Which number should I use?

Use the one you want to follow, taken from the same device and the same settings every night. ODI and T90 are steadier from night to night than lowest SpO2, which a single bad reading can set.

How many nights should I enter?

As many as you have, up to 14. With two or three nights the spread itself is a rough guess, so the nights-needed figure can swing a lot as you add more.

What margin should I pick?

The smallest change you would care about. If a shift of 2 events per hour in ODI would matter to you, use 2. A smaller margin needs more nights.

My nights are all over the place. What does that mean?

It means one night tells you little on its own. Check the recordings for gaps and loose-sensor stretches, keep your routine steady, and record more nights before you compare anything. If the numbers worry you, take the recordings to a doctor.

Is this the same as a sleep study?

No. This is arithmetic on numbers you type in. Home oximetry is not a sleep study, and this page says nothing about the cause of any number.

Night-to-night variation calculator

Overnight oximetry numbers move from one night to the next, even when nothing changes. Type in one number for each night you recorded and see how far they spread, how the average moves as nights are added, and roughly how many nights it takes for the average to settle. Nothing you enter is sent anywhere.

Enter at least 2 nights to see how they vary.

Planning more nights?

A baseline of ordinary nights, then one change at a time, is the simplest way to see how something affects your overnight oxygen. To put the recordings themselves side by side, use compare nights.

How many nights to record Compare nights

See the spread without typing

This calculator works from numbers you type in. An account keeps every night you record and shows the trend, so you can see how your nights vary over weeks and months.

Every recording saved and searchable
Compare any nights, any time
ODI, T90 and CVHR trends over time
Tag nights with what you changed
See which habits line up with your numbers
Describe a night in plain English

Start free See pricing

Your first 5 nights are free. No card needed.

How it's worked out

The average is the sum of the nights divided by the number of nights. The running average is the same sum after each night, so you can see it settle as nights are added.

SD = √( Σ(night − average)² ÷ (nights − 1) )

The standard deviation (SD) is the typical distance of a night from the average. It uses nights − 1, the usual choice for a small sample. The coefficient of variation is the SD as a percentage of the average, which lets you compare the spread of numbers on different scales.

nights ≈ (1.96 × SD ÷ margin)²

This is the textbook sample size for an average: how many nights it takes for the average to land within ± your margin of your longer-run average about 19 times in 20. Example: an ODI with an SD of 4 and a margin of 2 gives (1.96 × 4 ÷ 2)², about 15.4, so 16 nights. Halve the margin and the nights roughly quadruple.

Why your nights differ

How long you slept, and how much of it was on your back.
What happened that evening: a drink, a late meal, a blocked nose.
How the sensor sat on your finger. A loose fit or cold hands add gaps and false dips. See why O2Ring data has gaps.
Plain chance. Some nights just differ, with no cause you can point to.
Lowest SpO2 is a single moment in the night, so one bad reading can set it. Expect it to jump around more than ODI or T90.

What studies have found

Luu and colleagues looked at 132 three-night oximetry recordings from people with motor neuron disease. On a scale where 1 means identical nights, agreement was 0.94 for T90, 0.90 for ODI, 0.85 for baseline SpO2 and 0.77 for lowest SpO2. Even so, 11 of the 94 recordings that crossed the clinic's cut-off over three nights were under it on night one.

Lechat and colleagues gave 92 adults referred for suspected sleep apnea an under-mattress tracker for three months alongside a single-night sleep study. For 16 of 78 people, about 21%, the two landed on different sides of the same cut-off of 15 breathing events per hour. The 10 above it only on the tracker had bigger night-to-night swings than the 6 above it only on the sleep study.

Gardner and Burke had 26 people with heart failure wear an oximeter for six nights. The authors found statistically significant night-to-night differences in oxygen for all 26. One person spent 63% of night one below 92% SpO2 and 6% of night six.

None of these studies used the formula on this page. They show that one night is one sample. For a practical rule, read how many nights to record.

Common questions

Which number should I use?: Use the one you want to follow, taken from the same device and the same settings every night. ODI and T90 are steadier from night to night than lowest SpO2, which a single bad reading can set.
How many nights should I enter?: As many as you have, up to 14. With two or three nights the spread itself is a rough guess, so the nights-needed figure can swing a lot as you add more.
What margin should I pick?: The smallest change you would care about. If a shift of 2 events per hour in ODI would matter to you, use 2. A smaller margin needs more nights.
My nights are all over the place. What does that mean?: It means one night tells you little on its own. Check the recordings for gaps and loose-sensor stretches, keep your routine steady, and record more nights before you compare anything. If the numbers worry you, take the recordings to a doctor.
Is this the same as a sleep study?: No. This is arithmetic on numbers you type in. Home oximetry is not a sleep study, and this page says nothing about the cause of any number.

Sources

Not a medical device. Does not diagnose, treat or prevent any condition. Discuss results and changes with a clinician.