Oxygen Desaturation Index Calculator (ODI)
Oxygen Desaturation Index Calculator (ODI)
Compute the ODI at both the 3% and the 4% desaturation rule from one night’s counts, and see the gap — the stricter rule can move the same recording across a threshold without anything about the patient changing.
ODI at 3% and at 4%
Both rules, one night142 desaturations of 3% or more, of which 88 reached 4%, over 360 minutes
Formula
- 3% rule
- the AASM’s recommended hypopnoea criterion 1A requires a drop in peak signal excursion of 30% or more lasting 10 seconds or more, with a 3% oxygen desaturation from pre-event baseline and/or an associated arousal. The ODI counted on the same 3% threshold is the index that goes with it
- 4% rule
- the AASM’s alternative criterion 1B requires the same 30% drop for 10 seconds or more with a 4% oxygen desaturation from pre-event baseline, and the AASM’s own note on it is explicit: the criterion involving arousals is included in 1A and excluded from 1B. Every arousal-only event therefore disappears
- why both are in circulation
- the AASM suspended indefinitely the requirement to score at 3% in August 2013, stating that Medicare and other payers continue to require a desaturation of 4% or more, and that centres must document which criterion they used. Version 3 of the manual went further and made the 4% rule optional rather than acceptable
- the denominator
- sleep time on an attended polysomnogram, recording time on a home or oximetry-only study. The ODI inherits the whole AHI-versus-REI problem: identical counts over a longer denominator give a smaller index
- baseline
- a desaturation is measured from the PRE-EVENT baseline, not from 100% and not from the night’s mean. Medicare’s oxygen policy, which uses a different construct for a different purpose, defines baseline saturation as the mean saturation level during the duration of the test — a reminder that the word baseline is not portable between documents
Worked example
142 desaturations of 3% or more, of which 88 reached 4%, over 360 minutes
Hours = 360 ÷ 60 = 6.0
ODI at 3% = 142 ÷ 6.0 = 23.7 events/h
ODI at 4% = 88 ÷ 6.0 = 14.7 events/h
The 15-per-hour mark falls between them. One recording, one patient, one oximetry trace: the recommended rule puts the index above 15 and the alternative rule puts it below. Nothing measured changed — 54 of the 142 falls simply did not reach 4 percentage points
The 4% index is 88 ÷ 142 = 62.0% of the 3% index, and that ratio is just the ratio of the counts: the denominator cancels, so it is the same whatever the sleep time
Read against the adult AHI strata out of habit, as laboratories and readers often do, the two would be called moderate and mild respectively. The ODI has no published severity strata of its own, which is why this page prints no band
Change the sleep time to 240 minutes and the 3% index becomes 35.5; to 720 minutes and it becomes 11.8. Same events, three answers, and only the denominator moved
Enter a 4% count larger than the 3% count and the page refuses: a fall of 4 percentage points is also a fall of 3, so the second count is a subset of the first and cannot exceed it
The two rules, side by side
| Rule 1A (recommended) | Rule 1B (alternative) | |
|---|---|---|
| Flow reduction | 30% or more of pre-event baseline | 30% or more of pre-event baseline |
| Duration | 10 seconds or more | 10 seconds or more |
| Desaturation | 3% or more from pre-event baseline | 4% or more from pre-event baseline |
| Arousal counts instead | yes | no |
| Status in manual version 3 | recommended | optional |
| Required by Medicare | no | yes |
Why the AASM demoted this index in version 3
| Reporting rule | Version 2.x | Version 3 |
|---|---|---|
| Oxygen desaturation index | recommended | optional |
| Time below a specified saturation threshold | — | recommended |
| Central apnoea index | — | optional |
| Hypopnoeas scored at a 4% desaturation | acceptable | optional |
One oximetry trace, two indices, and a threshold between them
The oxygen desaturation index looks like the simplest number on a sleep report: count the falls in saturation, divide by the hours. The difficulty is that there is no single definition of a fall. The AASM’s recommended hypopnoea criterion counts a 3% desaturation, or an arousal with no desaturation at all; its alternative criterion requires 4% and credits no arousal. Both rules are in use, both are current, and the gap between them is not small. On the worked night above, 54 of 142 falls reached 3 percentage points and not 4 — enough to move the index from 23.7 to 14.7 and across the 15-per-hour mark.
The reason both survive is administrative rather than physiological. The AASM suspended the requirement to score at 3% in August 2013, saying in the same announcement that Medicare and a number of other payers continue to require a 4% desaturation, and that accredited centres must document which criterion they used. Version 3 of the manual, from February 2023, went further and reclassified the 4% rule from acceptable to optional. So the field has a recommended rule, an optional rule, and a payer that requires the optional one. 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.
The denominator does the same damage here as it does to the AHI. An ODI from an attended study divides by scored sleep; an ODI from a home device or a bare overnight oximetry divides by whatever it recorded, including the hours the patient lay awake. An index is only as meaningful as its denominator: the same event count over a shorter sleep time is a larger index, and nothing about the patient changed. An oximetry-only study has no other denominator available, so its index is systematically the smaller of the two.
The last thing worth knowing is that the AASM is quietly moving away from this index. Version 3 demoted reporting it from recommended to optional, and in the same revision made reporting the time below a specified saturation threshold recommended. Baumert and colleagues, in 2840 men of the MrOS cohort, found time below 90% saturation predicted cardiovascular mortality with a hazard ratio of 1.21, and 1.16 after adjustment, while this index reached 1.13 at p = 0.06 and was not significant. Read this against the reference range your own laboratory printed beside the metric. Normal values here are age- and sex-specific distributions published as tables, and this page does not approximate one. 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. This page does the arithmetic and says where every figure it quotes comes from. It renders no diagnosis and no clinical decision.
Frequently asked questions
What is the difference between a 3% and a 4% ODI?
The threshold a fall in saturation must reach to be counted. Every 4% fall is also a 3% fall, so the 4% index is always the smaller of the two, and the ratio between them is just the ratio of the counts. On the worked night above they are 23.7 and 14.7 events per hour — the 15-per-hour mark sits between them. The AASM recommends the 3% criterion; Medicare requires 4%.
Which ODI should I use?
Whichever the report states, and if it states neither, the index cannot be compared with anyone else’s number. The AASM’s recommended hypopnoea criterion uses 3%; its alternative uses 4% and is what Medicare’s coverage documents require; accredited centres are expected to document which they applied. This page prints both so a reader holding one can see the other.
Does the ODI have mild, moderate and severe bands?
No body publishes severity strata for it, which is why this page shows no coloured band. The adult 5, 15 and 30 thresholds belong to the apnoea-hypopnoea index, and laboratories that apply them to the ODI are borrowing them. Reference ranges for oximetry indices are also age-dependent and laboratory-specific, so use the range printed beside your own result.
Why did the AASM make reporting the ODI optional?
Its public Summary of Updates for version 3 of the scoring manual records that reporting the oxygen desaturation index was changed from a recommended to an optional rule, while reporting time below a specified saturation threshold became recommended. The AASM does not give a reason there, but the outcome literature points the same way: in 2840 MrOS men, time below 90% predicted cardiovascular mortality and the ODI did not reach significance.
Is a desaturation measured from 100% or from the baseline?
From the pre-event baseline, in both AASM rules — not from 100% and not from the night’s mean. That matters in a patient whose baseline is already low, where a fall to 84% from a baseline of 88% is a 4% desaturation while a fall to 94% from 98% is also one. Medicare’s oxygen policy uses baseline to mean the mean saturation during the test, which is a different construct for a different purpose.
Related calculators
References
- American Academy of Sleep Medicine. AASM clarifies hypopnea scoring criteria. Darien, IL: AASM.
- American Academy of Sleep Medicine. AASM suspends required use of 3 percent hypopnea desaturation scoring criterion. 27 August 2013.
- American Academy of Sleep Medicine. Summary of Updates in Version 3 of The AASM Manual for the Scoring of Sleep and Associated Events. Darien, IL: AASM; February 2023.
- 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.
- Centers for Medicare & Medicaid Services. Local Coverage Determination L33718: Positive Airway Pressure (PAP) Devices for the Treatment of Obstructive Sleep Apnea.
- Centers for Medicare & Medicaid Services. Local Coverage Determination L33797: Oxygen and Oxygen Equipment, including the revision history at R9 (1 January 2023).
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/