How to Read Your Sleep Study
The AHI gets all the attention and is the least stable number on the page. Here is what each figure means, which ones predict outcomes better, and what to ask for when your report is missing them.
Most people leave a sleep clinic with a number and a severity label, and nothing else that makes sense. The report itself is dense, written for clinicians, and contains several figures that matter more than the one everyone quotes.
This is what each number means, and which ones deserve your attention.
Nothing here replaces the clinician who ordered the test. The point is to let you read your own report and ask better questions about it.
The headline number: AHI
The apnoea-hypopnoea index is the average number of apnoeas (breathing stops) and hypopnoeas (breathing roughly halves) per hour of sleep. It drives the severity label:
| AHI | Label |
|---|---|
| Under 5 | Normal |
| 5 to under 15 | Mild |
| 15 to 30 | Moderate |
| Over 30 | Severe |
Those bands come from clinical guidance (AASM clinical practice guideline, Journal of Clinical Sleep Medicine, 2017) and they are conventions rather than biological cliffs. An AHI of 14.8 and an AHI of 15.2 are the same physiology with different labels.
Why the AHI is the least stable number on your report
This is the part almost nobody is told, and it changes how much weight a single figure deserves.
A systematic review of 24 studies covering 3,250 participants looked at how much respiratory events vary between one night and the next (Roeder et al., Thorax, 2020).
On average, across groups, there was essentially no difference — the mean AHI difference between first and second night was −1.70 per hour (95% CI, −3.61 to 0.02) [Roeder 2020].
Individually, it was a different story:
- 41% of participants (95% CI, 27–57%) showed changes of more than 10 events per hour from one night to the next [Roeder 2020]
- 49% (95% CI, 32–65%) changed severity class at least once across sequential studies [Roeder 2020]
- Depending on the diagnostic threshold used, around 10–12% would have been missed entirely by a single night of testing [Roeder 2020]
Roughly half of people would be put in a different severity band by a second night. That is not a reason to distrust your report — it is a reason to treat the number as an estimate with a range around it rather than a fixed property of your body, and a strong reason not to draw conclusions from a small difference.
If your result came back just under a threshold and your symptoms say otherwise, that discrepancy is worth raising rather than accepting.
The numbers that often matter more
Oxygen desaturation index, and how low you went
The ODI counts how often your oxygen saturation dropped by a set amount (usually 3% or 4%) per hour. Oxygen nadir or minimum SpO2 is the lowest point reached, and time below 90% is how long you spent there.
Two people with identical AHIs can have very different oxygen profiles. Brief events that barely move saturation are not equivalent to events that drop you into the low 80s — and the AHI cannot tell them apart, because it counts events without weighing them.
Hypoxic burden
The metric that captures what the AHI misses, and the one with the better outcome evidence behind it.
Hypoxic burden measures the area under the desaturation curve for each respiratory event — combining how often you desaturate with how deep and how long each drop is.
In two large cohorts — 2,743 men in MrOS and 5,111 adults in the Sleep Heart Health Study — hypoxic burden strongly predicted cardiovascular mortality while the AHI did not (Azarbarzin et al., European Heart Journal, 2019). Participants in the highest quintiles carried hazard ratios of 1.81 (95% CI, 1.25–2.62) and 2.73 (95% CI, 1.71–4.36) in MrOS, and 1.96 (95% CI, 1.11–3.43) in SHHS [Azarbarzin 2019].
Most reports do not include it yet. What hypoxic burden is and why it matters.
Arousal index
How many times per hour your brain briefly woke — measured on EEG, mostly without you remembering any of it.
This is where the daytime symptoms usually live. Someone with a modest AHI and a high arousal index can be genuinely exhausted, and a report showing that combination explains a lot. It also matters for the pattern where breathing effort fragments sleep without qualifying as apnoea. When snoring itself causes fatigue.
Supine versus lateral AHI
Ask for this one specifically if it is not there. It is the single most actionable line on a sleep study.
If your AHI is 40 on your back and 8 on your side, you have positional apnoea and position is treating a large share of your problem. If it is 30 either way, position is close to irrelevant for you and positional advice is wasted effort. Same report, opposite conclusions. Positional therapy, done properly · Positional trainers.
REM AHI
Events cluster in REM sleep, when muscle tone is lowest. A study with little REM may understate your severity — which is one mechanism behind the night-to-night variability above, since REM time differences were significantly associated with AHI differences [Roeder 2020].
Total sleep time and sleep efficiency
The denominator. An AHI is events divided by hours of sleep, so a study where you slept two broken hours in an unfamiliar bed is a less reliable estimate than one where you slept seven. Home tests often estimate sleep time rather than measuring it, which shifts the number.
RDI and RERAs
The respiratory disturbance index includes respiratory effort-related arousals — events where breathing effort rose enough to wake you without meeting apnoea or hypopnoea criteria. A report with an RDI meaningfully higher than its AHI is describing exactly that pattern, and it explains symptoms an AHI alone would not.
If your report shows only an AHI and your symptoms do not match it, asking whether RERAs were scored is a specific and reasonable question.
Home test versus lab study
A home test measures fewer channels and usually has no EEG, so it cannot see arousals or stage your sleep, and it typically estimates sleep time from recording time. That tends to dilute the index — events divided by a larger denominator.
Practical consequence: a home test is good at confirming apnoea and less good at ruling it out. A negative home study in someone with strong symptoms is not the end of the question. What home testing involves.
Questions worth asking about your report
- What was my supine versus non-supine AHI?
- Were RERAs scored, and what is my RDI?
- What was my oxygen nadir and my time below 90%?
- How much did I actually sleep, and how much REM did I get?
- Given the night-to-night variability, how confident are you in this severity class?
- Is my hypoxic burden available, or calculable from the raw data?
Take the report to the appointment rather than the label. What to tell the sleep doctor.
What a sleep study does not tell you
Where your airway collapses. It tells you that it does and how badly, not whether the problem is your nose, your palate or your tongue base. That distinction decides which remedies can physically work. Where your airway narrows.
How loudly you snore. Snoring volume is poorly captured and often not reported, and it correlates weakly with severity. A quiet study is not a reassuring study.
Whether you will tolerate a treatment, which is the variable that decides real-world benefit more than any number on the page. CPAP versus an oral appliance.
And the reverse of the reassurance trap: a normal study does not mean nothing is wrong. It means this night, measured these ways, did not cross these thresholds. Snoring versus sleep apnea.