How to Read Your CPAP Data
Your machine reports four numbers that tell you whether treatment is working. Read in the wrong order they mislead — a residual AHI is meaningless until you have checked the leak figure.
A CPAP machine is the only treatment on this site that reports on itself every night. That is a genuine advantage and most people never use it, because the numbers arrive without explanation and in the wrong order of importance.
Here is what each one means, and the order to read them in.
One thing to be clear about first: these are the machine's own estimates, not a scored sleep study. A device sees airflow and pressure. It has no EEG, so it cannot see arousals or stage your sleep, and it infers events from flow patterns rather than from the full signal set a technician scores. Treat the numbers as a useful ongoing signal, not as a diagnostic-grade measurement. I am not aware of evidence establishing that device-reported indices are equivalent to a scored polysomnogram, and the mechanics above are the reason to expect they are not.
If your machine is reporting central events, that has its own explanation and its own likely course — central sleep apnea, and the kind that appears on CPAP.
Read them in this order
1. Leak — first, always
Why first: every other number depends on it. A machine measures the air it delivers. Air escaping around a mask edge or through an open mouth is air the device cannot account for, and large leak makes its event detection unreliable. A residual AHI recorded during heavy leak is not trustworthy.
Most reports give a median and a 95th percentile leak. The 95th percentile is the informative one — it captures the bad part of the night rather than averaging it away. Manufacturers publish their own acceptable ranges and they differ between machines and mask types, so compare against your device's guidance rather than a number from an article.
What large leak usually means: the mask is the wrong size or shape, the straps are over-tightened (which deforms the cushion and causes leak rather than preventing it), the cushion is worn out, or your mouth is falling open on a nasal mask. That last one is common and fixable. Mouth leak on CPAP.
2. Usage
Hours per night, and nights used. This is the number your clinician and, in the US, your insurer will look at — many insurers apply a minimum usage rule, commonly expressed as four hours a night on most nights, though the specifics vary by payer and are worth confirming rather than assuming.
Treat that threshold as an administrative floor, not a clinical target. There is no physiological reason benefit stops at four hours; more hours means more of the night treated.
For commercial drivers this number is a professional document, because it is what demonstrates treatment. Sleep apnea and your DOT physical.
3. Residual AHI
Events per hour the machine believes are still happening on treatment. A well-titrated setup usually brings this well below 5, and most clinicians want it low single digits or under.
Read it only after the leak figure. A residual AHI of 2 with large leak tells you less than a residual AHI of 6 with a good seal.
If yours is persistently high with acceptable leak, the possibilities are worth knowing:
- Pressure too low for your needs, or an auto range set too narrow
- Positional events — obstruction that appears only on your back, which the machine cannot distinguish by position
- Central events rather than obstructive ones, which pressure does not fix and can occasionally worsen
- Mouth breathing on a nasal mask, which both leaks and lets the airway destabilise
4. Pressure
On a fixed-pressure machine this is simply your setting. On an auto-adjusting machine you get a range and usually a 95th percentile pressure — the level at or below which you spent 95% of the night.
Two patterns are worth noticing. If your 95th percentile sits at the top of your allowed range, the machine may be hitting its ceiling and the range may be set too low. If it sits far below your maximum, the ceiling is not the constraint.
5. Central apnea index, if reported
Some machines separate central events — where effort stops rather than the airway closing. A rise in central events after starting treatment is a recognised phenomenon, and it is not fixed by more pressure. It relates to ventilatory control instability rather than obstruction. Why treatments work for some people and not others.
What good data looks like
Leak within your device's stated range, usage covering most of your sleep, residual AHI low single digits, and pressure not pinned at the ceiling. If all four hold and you still feel unrefreshed, the useful conclusion is that your apnoea is being treated and something else is causing the symptom — insufficient sleep time, another sleep disorder, or one of the ordinary medical causes of fatigue. That is worth knowing rather than escalating pressure. When snoring itself causes fatigue.
When to act
Contact your provider if: residual AHI is persistently above about 5 with acceptable leak; leak is consistently high despite mask adjustment; central events are rising; or you feel worse rather than better.
Do not change your own pressure settings. They are prescribed, and in most places clinical menus are locked for good reason. Bring the data to the person who prescribed it.
Do not read one bad night as a trend. Night-to-night variability in respiratory events is substantial — around half of people change severity class between sequential diagnostic nights (Roeder et al., Thorax, 2020). That variability does not disappear once you are on treatment. Look at weekly and monthly patterns, not last night.
What the data cannot tell you
Whether you slept well. No EEG means no sleep staging and no arousal index. You can have excellent machine numbers and badly fragmented sleep.
Whether your cardiovascular risk is improving. Residual AHI is a process measure. Recent analysis suggests CPAP's cardiovascular benefit is concentrated in higher-risk patients rather than distributed evenly, so a good number is not itself an outcome. Does CPAP reduce cardiovascular risk.
Whether your pressure is optimal, as opposed to adequate. That is a titration question.
And the thing that is not a data problem
If the equipment is the obstacle — mask discomfort, dryness, claustrophobia, aerophagia — those are solvable and are the main reason people abandon treatment. Reading the data well does not help if the mask is unwearable. Getting used to CPAP · The alternatives, if it genuinely will not work.