OSA Severity Grading Interpreter
OSA Severity Grading Interpreter
Read an index against the published severity strata — and see why the strata are not the diagnostic criteria, why the adult and paediatric stratifications are different, and what the strata are criticised for.
Severity strata and diagnostic criteria
Strata are not criteriaAn index of 16.8 events per hour in an adult, with a documented symptom
The strata, and the criteria, are two different things
| Question | What the source says | Source |
|---|---|---|
| How severe is this index? | under 5 per hour normal for adults; 5 to 14.9 mild; 15 to 29.9 moderate; 30 or more severe | AASM, 2025 specification of quality measure 277 |
| Is the diagnostic criterion met? | an obstructive RDI of 5 or more per hour with typical symptoms or a listed comorbidity, OR 15 or more per hour regardless of symptoms | ICSD-3, as quoted in the AASM’s 2017 guideline |
| Does a payer cover a device? | 15 or more per hour with at least 30 events, or 5 to 14 with at least 10 events plus a symptom or comorbidity | Medicare PAP local coverage determination L33718 |
| How severe is a child’s index? | above 1 per hour as the syndrome cut-off by polygraphy; 2 or more as one definition’s lower bound; above 5 described as moderate-to-severe. No threshold at 10 | ERS task force statement, 2- to 18-year-olds |
What the severity strata are criticised for
| Criticism | Evidence | Cohort |
|---|---|---|
| The index does not predict cardiovascular outcome | the hypoxic burden — the area under the respiratory event-related desaturation curve — carried hazard ratios of 1.81 and 2.73 for the upper two quintiles, and 1.96 in the second cohort, where the AHI poorly predicts the adverse outcomes of sleep apnoea | 2743 men in MrOS; 5111 adults in the Sleep Heart Health Study (Azarbarzin 2019) |
| The index treats unequal events as equal | it disregards the depth and the duration of the ventilatory disturbance, so a long event with a shallow desaturation and a short one with a deep desaturation count the same | argued in Mansour and Won’s review of the index’s limits |
| One night is one sample | about 29% of people whose 28-night average was moderate to severe were misclassified from a single night, and about 15% of those with none or mild were classified as moderate to severe | 32,775 people in the Lechat authors’ reply |
| Time below 90% outperformed the desaturation index | T90 predicted cardiovascular mortality (hazard ratio 1.21, then 1.16 adjusted) where the oxygen desaturation index did not reach significance (1.13, p = 0.06) | 2840 men in MrOS (Baumert 2020) |
The strata, the criteria, and what the index does not predict
Three sentences do most of the damage in this subject, and they are all variants of one confusion: that the severity band and the diagnostic criterion are the same statement. They are not. The AASM’s own specification of its sleep apnoea severity quality measure sets the adult bands at under 5 per hour, 5 to 14.9, 15 to 29.9 and 30 or more. The diagnostic definition the AASM’s 2017 guideline quotes from ICSD-3 is a different shape altogether: an obstructive respiratory disturbance index of 5 or more per hour with typical symptoms or a listed comorbidity, or 15 or more per hour regardless. So an index of 7 in a person with no symptoms and no comorbidity lands in the mild band and does not meet the definition, and saying both of those at once is the honest answer.
The second confusion is between populations. The adult strata and the European Respiratory Society task force statement’s paediatric marks are an order of magnitude apart — 5, 15 and 30 against 1, 2 and 5 — because a child’s airway tolerates far less. Reading a child’s index against the adult bands makes a markedly abnormal study look normal. Note also what the paediatric statement does not contain: no threshold at 10 appears in it, so the widely quoted paediatric triad of 1, 5 and 10 is not this source’s, and none is printed here.
The third is treating the band as a measure of how ill someone is. The literature against that is substantial. Azarbarzin and colleagues, working in 2743 men of the MrOS cohort and 5111 adults of the Sleep Heart Health Study, found that the index poorly predicts the adverse outcomes of sleep apnoea, while a measure of the area under each event’s desaturation curve carried hazard ratios for cardiovascular death of 1.81, 2.73 and 1.96. Baumert and colleagues, in 2840 of the same men, found time below 90% saturation predicted cardiovascular mortality where the desaturation index did not. The newer metrics exist because of this. Most of them need the saturation trace rather than a number on a report, so they cannot be computed here, and they are named instead.
A severity band is not a diagnosis. The diagnostic criteria and the severity strata are two different published things, and conflating them is the commonest error on this page’s subject. Every number on a sleep study is scoring-rule-dependent and night-to-night variable: the same recording scored under two rules, or the same patient studied on two nights, gives different figures. A coverage or regulatory threshold is jurisdiction-specific and is quoted here as what that named authority publishes, not as a clinical standard and not as advice. The AASM’s recommended rule 1A credits a 3% desaturation or an arousal; its alternative rule 1B requires 4% and gives no arousal credit, and Medicare’s own coverage documents require 4%. The same night scored both ways gives two different numbers. This page does the arithmetic and names the body that publishes the threshold. It renders no diagnosis and no clinical decision.
Frequently asked questions
What AHI counts as mild, moderate and severe?
The AASM’s own 2025 specification of its sleep apnoea severity quality measure states it as: an AHI or REI under 5 per hour is normal for adults, 5 to 14.9 is mild, 15 to 29.9 is moderate and 30 or more is severe. A 2022 Sleep Science review states the same bands as mild 5 to under 15, moderate 15 to 30 and severe above 30, which puts exactly 30.0 in a different band; this page follows the AASM’s wording. The AASM’s 2017 specification of that same measure stated the strata over the RDI rather than the AHI.
Is an AHI of 5 a diagnosis of obstructive sleep apnoea?
No. The definition the AASM’s 2017 guideline quotes from ICSD-3 asks for an obstructive respiratory disturbance index of 5 or more per hour together with typical symptoms — unrefreshing sleep, sleepiness, fatigue or insomnia, waking with a gasping or choking sensation, loud snoring, witnessed apnoeas — or together with obesity, hypertension, stroke or congestive heart failure. Without one of those, an index of 5 sits in the mild band and does not meet that definition. At 15 or more the symptom requirement drops away.
Are the paediatric thresholds the same as the adult ones?
No, and the difference is about fivefold. The European Respiratory Society task force statement on 2- to 18-year-olds uses an AHI above 1 episode per hour as its polygraphy cut-off, an obstructive AHI of 2 or more as one definition’s lower bound, and describes above 5 as moderate-to-severe. It publishes no threshold at 10. An index that is the adult mild band is well into the abnormal range for a child.
Why is the AHI criticised as a severity measure?
Because it counts events without weighting them, and the count has turned out to predict outcome poorly. Azarbarzin and colleagues, in 2743 MrOS men and 5111 Sleep Heart Health Study adults, reported that the AHI poorly predicts the adverse outcomes of sleep apnoea while the area under each event’s desaturation curve carried hazard ratios of 1.81, 2.73 and 1.96 for cardiovascular death. Baumert and colleagues found time below 90% saturation significant where the desaturation index was not. The index also ignores event depth and duration, and varies between nights.
Does it matter which index the strata are read over?
Yes, and the sources are not consistent about it. The AASM’s 2025 measure specification states the strata over the AHI or the REI; its 2017 specification stated them over the RDI. Those are three different numbers on one night: the RDI adds respiratory effort-related arousals, and the REI divides by recording rather than sleep time. A report that gives a number without naming the index leaves the band undecidable.
Related calculators
References
- American Academy of Sleep Medicine. Sleep Apnea Severity (MIPS clinical quality measure 277, 2025 specification). Darien, IL: AASM; 2025.
- American Academy of Sleep Medicine. Sleep Apnea Severity (quality measure 277, 2017 specification). Darien, IL: AASM; 2017.
- Kapur VK, Auckley DH, Chowdhuri S, et al. Clinical practice guideline for diagnostic testing for adult obstructive sleep apnea: an American Academy of Sleep Medicine clinical practice guideline. J Clin Sleep Med. 2017;13(3):479–504.
- Kaditis AG, Alonso Alvarez ML, Boudewyns A, et al. Obstructive sleep disordered breathing in 2- to 18-year-old children: diagnosis and management. Eur Respir J. 2016;47(1):69–94.
- Azarbarzin A, Sands SA, Stone KL, et al. The hypoxic burden of sleep apnoea predicts cardiovascular disease-related mortality in the Osteoporotic Fractures in Men Study and the Sleep Heart Health Study. Eur Heart J. 2019;40(14):1149–57.
- Baumert M, Immanuel SA, Stone KL, et al. Composition of nocturnal hypoxaemic burden and its prognostic value for cardiovascular mortality in older community-dwelling men. Eur Heart J. 2020;41(4):533–41.
- Soori R, Baikunje N, D’sa I, et al. Pitfalls of AHI system of severity grading in obstructive sleep apnoea. Sleep Sci. 2022;15(S1):285–8.
- Mansour M, Won C. Apnea-hypopnea index: limitations and future directions. Pulmonary Health Hub commentary (read as a secondary review; its primary figures were re-read in the papers it cites).
- Lechat B, Catcheside PG, Reynolds AC, et al. Reply to Martinez-Garcia et al. and to Abreu and Punjabi. Am J Respir Crit Care Med. 2022;206(1):126–9, on Lechat B, et al. 2022;205(5):563–9.
- Centers for Medicare & Medicaid Services. Local Coverage Determination L33718: Positive Airway Pressure (PAP) Devices for the Treatment of Obstructive Sleep Apnea.
Not medical advice. For healthcare professionals and education. Reference intervals vary by laboratory and assay — always use your own laboratory's. Never base a dose or a treatment decision on this page alone. Full disclaimer at calcengines.com/disclaimer/