A patient sees a nominally normal AHI on their CPAP report — say, 2.1 — and still wakes up tired. The dealer says the therapy is working. The patient knows it isn’t. The explanation, more often than not, sits in a sub-apnea phenomenon that the AHI number does not contain: flow limitation. Partial inspiratory airway narrowing, insufficient to meet hypopnea criteria, but sufficient to fragment sleep and drive daytime symptoms. Modern APAPs detect it, react to it, and quietly adjust pressure because of it — but the headline AHI number doesn’t carry its imprint.
This article covers what flow limitation is physiologically, how it’s detected on the inspiratory flow waveform, how RERAs relate, how APAP algorithms respond to flow limitation (specifically by increasing pressure), and why flow-limitation-dominant patients remain symptomatic despite good AHI numbers.
What flow limitation is
Normal inspiratory airflow through a healthy upper airway follows a roughly sinusoidal time-course — flow rises from zero, peaks mid-inspiration, and falls back to zero. The flow contour is smooth and rounded. The airway is offering mechanical compliance that lets negative intrathoracic pressure during inspiration translate cleanly into flow.
When the pharyngeal airway partially narrows — because of soft-tissue crowding, reduced muscle tone in sleep, posterior tongue displacement, or mild edema — the airway becomes effectively rate-limited. Below a critical narrowing, increasing inspiratory effort no longer produces proportional flow increase; instead, the negative intrathoracic pressure pulls the partially collapsed airway walls further inward, and flow reaches a plateau. The inspiratory flow waveform visibly flattens — instead of a rounded peak, there is a plateau segment, sometimes with a small spike at peak inspiration.
This flow-contour flattening is the visible signature of flow limitation. It sits on a physiological continuum with hypopnea and obstructive apnea:
- No flow limitation — smooth rounded waveform, normal airway.
- Mild flow limitation — subtle flattening, possibly detectable only by trained eye or signal analysis. Usually doesn’t disrupt sleep.
- Moderate flow limitation — clear flattened plateau, rising respiratory effort to compensate, increased work of breathing. May terminate in arousal (a RERA) or resolve spontaneously as the patient shifts position or sleep deepens.
- Severe flow limitation — extended plateau, substantial flow reduction approaching hypopnea territory. Often culminates in a scored event (hypopnea or RERA).
- Hypopnea — flow reduction ≥ 30% with desat or arousal, lasting ≥ 10 seconds.
- Obstructive apnea — flow reduction ≥ 90% lasting ≥ 10 seconds.
The AASM scoring rules impose thresholds on this continuum. Events above threshold make AHI; events below threshold don’t. But the physiology doesn’t care about thresholds — the airway is doing the same thing, just less dramatically. A patient with a low AHI but high flow-limitation burden has real upper-airway resistance; the AHI simply misses it.
RERA — the flow-limitation event that does count
A RERA (respiratory effort-related arousal) is a flow-limitation sequence that terminates in an EEG-defined cortical arousal and doesn’t meet apnea or hypopnea criteria. AASM requires:
- ≥ 10 seconds duration,
- Increasing respiratory effort or progressive flow limitation (flattening),
- Terminated by an arousal,
- Not meeting apnea/hypopnea thresholds.
RERAs are included in RDI (respiratory disturbance index) but not AHI. A patient with AHI 3 and RDI 18 has a statistically normal AHI and a clinically abnormal respiratory picture. The 15 events per hour of RERA between those two numbers are fragmenting sleep.
This is the definitional basis of UARS (Upper Airway Resistance Syndrome) — a sleep-disordered breathing phenotype with:
- AHI < 5 (technically below OSA diagnostic threshold),
- RDI > 5 (often > 10),
- Symptomatic daytime sleepiness, morning headaches, unrefreshing sleep,
- Frequently a predominance of flow limitation and RERAs over classic apneas.
UARS is real, treatable, and systematically under-diagnosed when only AHI is reported. Women and thin younger men are over-represented in the UARS phenotype, often pushing back on OSA workup because “I’m not the typical apnea patient.” (AASM Practice Guidelines)
How APAP algorithms respond to flow limitation
An auto-titrating CPAP (APAP) ideally delivers the minimum pressure that keeps the airway adequately splinted — enough to suppress obstructive events, not so much that pressure intolerance or aerophagia develop. The algorithm must sense inadequate splinting and respond.
The most sensitive pre-event signal the APAP has is flow limitation. Apneas and hypopneas, by definition, have already happened by the time the algorithm sees them. Flow-limitation detection lets the algorithm respond before events manifest, raising pressure when the flow contour starts flattening and backing off when the contour normalises.
Different manufacturers weigh flow limitation differently:
ResMed AutoSet algorithm reacts quickly to flow-limitation onset. Flow-shape analysis on the inspiratory waveform detects flattening with reasonable sensitivity, and the algorithm increases pressure in small increments (typically 0.5–1 cmH₂O per minute of sustained flow limitation). The AutoSet family’s reputation for tight residual AHI control derives substantially from this responsive flow-limitation handling.
Philips DreamStation detects flow limitation and responds, but with less aggressive pressure increments in published comparative evaluations. Residual flow-limitation index on equivalent patients tends to run slightly higher on DreamStation than on AirSense.
BMC APAP family has flow-limitation detection in firmware, but the responsiveness and threshold tuning vary across generations and are less transparent in public documentation.
The cross-device comparison is not about “better” or “worse” in isolation — a more sensitive algorithm produces lower residual flow limitation but higher 95th-percentile pressures, with the aerophagia / pressure-tolerance trade-off that implies. Patient-specific preference and tolerance matter.
Why a low AHI doesn’t always mean therapy is working
The scenario: a patient is on well-titrated CPAP, home AHI averaging 2, leak within specification, usage > 5 hours per night. The patient still reports unrefreshing sleep and daytime sleepiness. The AHI says therapy is adequate; the patient says otherwise.
Several failure modes to investigate, in order:
1. Residual flow limitation. Download the device report and specifically look at the flow-limitation index. If it’s elevated (typically > 0.3 on ResMed, or the equivalent metric on other brands), residual partial obstruction is still fragmenting sleep without meeting AHI criteria. Solution: raise the APAP upper pressure limit (or raise fixed CPAP pressure) to allow the algorithm to respond more aggressively.
2. Arousals from other causes. PLMS (periodic limb movements), nocturnal bladder, partner’s snoring, GERD, anxiety, non-respiratory insomnia. The CPAP addresses respiratory arousals; other causes don’t respond to pressure. Investigation may require a repeat PSG on therapy.
3. Mask or leak intrusion. Mouth leaks, dry mouth, mask-related awakenings. Not captured well in AHI; captured in leak numbers.
4. UARS phenotype undetected at diagnosis. If the patient was diagnosed on a Type III HSAT (no EEG, no arousal scoring), RERAs and UARS were missed. CPAP on an incomplete diagnosis may address the wrong problem.
5. Persistent hypoxic burden despite low AHI. The ODI (oxygen desaturation index) may reveal desaturations that the AHI misses — short-duration events that fall below hypopnea criteria but accumulate hypoxic burden over the night. (Azarbarzin A et al, Eur Heart J 2019)
The flow-limitation index — what value is normal
There is no universally accepted numeric threshold for flow-limitation burden. Rough heuristics from device data:
- < 0.1 — minimal residual flow limitation; therapy clean.
- 0.1–0.3 — modest residual; acceptable in an asymptomatic patient.
- > 0.3 — substantial residual; should prompt investigation in a symptomatic patient.
- > 0.5 — high residual; APAP upper limit is constraining the algorithm’s response.
These are not hard diagnostic cut-offs — the index is proprietary and varies across brands. Trend matters more than absolute value: a flow-limitation index that has drifted upward over months signals something changing (weight gain, mask deterioration, nasal patency changes, leak-driven false flow-limitation artefacts).
The clinical question: raise pressure, or raise mode?
A patient with AHI < 5 but high flow-limitation index and persistent symptoms has several options:
1. Raise the APAP upper pressure limit. If the current upper limit is 14 cmH₂O and the 95th percentile is consistently hitting 14, the algorithm is constrained. Raising to 16 cmH₂O lets the algorithm react to flow limitation. Watch for aerophagia and leak as pressure climbs.
2. Switch to BiPAP. If raising CPAP pressure causes exhalation intolerance, switching to BiPAP (e.g., IPAP 16 / EPAP 10) splits the pressure into a higher inspiratory pressure (better airway splinting during inspiration, when flow limitation matters) and a lower expiratory pressure (easier exhalation).
3. Add oral appliance combination therapy. In selected patients with residual flow limitation, a mandibular advancement device alongside CPAP can reduce the mechanical load. Niche but occasionally valuable.
4. Investigate surgical / anatomical contributors. Nasal obstruction (septal deviation, turbinate hypertrophy) and retrognathic anatomy drive flow limitation and sometimes benefit from ENT consultation.
The choice among these depends on residual symptom burden, patient tolerance, and available services. A metropolitan Indian sleep-medicine practice has all four options; a tier-2 city practice may be limited to option 1 and referrals for 3 and 4.
Inspiratory flow-waveform morphology — what to look for in a trace
When a clinician opens the raw flow trace (via OSCAR on a home device, or the lab’s PSG software), several morphological features indicate flow limitation:
- Flattened peak (plateau). Instead of a rounded inspiratory peak, a flat segment where flow does not increase despite continuing inspiratory effort. This is the classic flow-limitation signature.
- Early-peak flattening. Flow rises rapidly at inspiration onset, then plateaus or decreases while inspiratory effort continues. Suggests early-inspiration airway narrowing.
- Late-inspiration drop. Flow rises normally initially but falls off toward end-inspiration while effort persists. Can indicate tissue collapse at the end of inspiration.
- Notched peaks or double peaks. Irregular flow contour with multiple peaks. Less specific but often coincident with flow limitation.
Trained polysomnographic technologists score flow limitation visually against these morphological cues. Automated flow-limitation scoring in APAP firmware uses similar features extracted via signal-processing algorithms — typically a combination of peak-to-plateau ratios, spectral analysis, and shape-metric calculations. (AASM Scoring Manual)
The UARS demographic — who gets missed
UARS and flow-limitation-dominant phenotypes are over-represented in:
- Women. Flow-limitation events and RERAs are a more common presentation pattern in female OSA cohorts than in male cohorts. Women also present at lower average AHI values for equivalent symptom burden.
- Thinner patients. BMI-independent anatomical narrowing (retrognathia, high-arched palate, elongated soft palate) drives flow limitation without the soft-tissue crowding that produces frank apneas.
- Younger adults. Classical OSA picture with loud apneas is more common in middle-aged to older men; younger adults often present with UARS-spectrum disease.
A patient in any of these groups with low AHI but persistent symptoms warrants careful flow-limitation evaluation. The default assumption that a normal AHI rules out sleep-disordered breathing is wrong in this demographic and leads to systematic under-diagnosis.
Clinical takeaway
Flow limitation is the invisible part of sleep-disordered breathing — physiologically real, clinically important, absent from the headline AHI number, but captured in the APAP’s flow-limitation index and visible on careful inspiratory-waveform inspection. Patients who remain symptomatic despite good-looking AHI should have the flow-limitation metric reviewed. Don’t accept “AHI is fine, therapy is working” as sufficient when the patient still reports unrefreshing sleep.
HHZ’s editorial view: every CPAP data-review consultation should include a specific look at the flow-limitation trend, not just the AHI. This is the single metric most often ignored and most often responsible for the gap between numerical adequacy and patient-reported outcomes.
Consult your sleep physician if you remain symptomatic on CPAP despite a good AHI — flow limitation, UARS, and other sub-apnea phenomena may require a specific investigation and adjustment.
References: Guilleminault C et al — UARS classical description [CITATION]; AASM Manual v3 — RERA scoring [CITATION]; ResMed AutoSet algorithm white paper [CITATION]; Azarbarzin A et al, Eur Heart J 2019 — hypoxic burden [CITATION]; Philips DreamStation 2 clinician guide [CITATION].