Why a Single Night of Testing Can Misclassify Your Sleep Apnea Severity
What night to night AHI variability means when your result sits near a threshold
One night of testing gets mild and moderate sleep apnoea right barely half the time, so here is exactly what to check and what to ask for next.
Night to night AHI variability is why one test night can misclassify you
A single night of testing classifies severe sleep apnoea correctly about 85% of the time, mild apnoea only 54% and moderate apnoea only 52% (Simonds, American Journal of Respiratory and Critical Care Medicine editorial, 2021). That gap has a name. It is night to night AHI variability. Your apnoea-hypopnoea index (AHI), the number of breathing stoppages and partial obstructions counted per hour of sleep, genuinely moves between nights in the same body, on the same equipment. A borderline result is one sample. It is not a fixed fact about you. If you have not yet read what your AHI score actually measures, start there.
So can a sleep study be wrong? Yes, and the size of the error is measurable. Across 24 studies and 3,250 participants, 49% of people changed severity class at least once across sequential sleep studies (Roeder, Thorax, 2020). The same problem sits underneath a normal study that contradicts how you feel. One night photographs a moving target.
- The error concentrates in the mild and moderate bands, where about half of single-night classifications are wrong.
- Severity reclassification between nights is common, not rare.
- If your number sits near 5, 15 or 30 events per hour, treat it as provisional.
How far your AHI actually moves from one night to the next
In about four people out of ten, the AHI shifts by more than 10 events per hour between consecutive nights. That is 41% of participants in the Thorax meta-analysis (Roeder, 2020). Same person. Same disease. Different number.
Night to night AHI variability, abbreviated NtNV in the research literature, has also been measured at scale. One group followed 67,278 people at home for an average of about 170 nights each, 11.6 million nights in total, and put the likelihood of misdiagnosis from a single night at roughly 20% to 50% (Lechat, American Journal of Respiratory and Critical Care Medicine, 2022).
The error is concentrated where you probably are
In 10,340 adults tested on three consecutive nights with a type III portable monitor, the home sleep apnea test (HSAT) that records airflow, effort and oxygen rather than brain activity, the pattern is stark (Punjabi, Chest, 2020).
| Your result on one night | How often that night matches the multi-night reference | What that means for you |
|---|---|---|
| Normal, no OSA | About 93% (Punjabi, Chest, 2020) | Reassuring, though persistent symptoms still deserve follow-up |
| Mild OSA, AHI 5 to 15 | 54% (Simonds, AJRCCM, 2021); about 20% misclassified against a three-night reference (Punjabi, Chest, 2020) | Close to a coin toss, and weak ground for a treatment decision |
| Moderate OSA | 52% (Simonds, AJRCCM, 2021); the same 20% misclassification (Punjabi, Chest, 2020) | Equally unstable, so one repeat night often changes it |
| Severe OSA, AHI 30 and above | 85% (Simonds, AJRCCM, 2021); 87% (Punjabi, Chest, 2020) | Solid enough to act on now. Start treatment, do not wait |
Being missed entirely is also measurable. Depending on the cut-off used, 12% of patients at 5 events per hour, 12% at 10 and 10% at 15 would have been missed on the first night alone (Roeder, Thorax, 2020). A false negative sleep study is close to a one-in-eight event at the lowest threshold. Different technology gives the same answer. In 51 patients tested at home for three nights with peripheral arterial tonometry (PAT, a wrist-worn method that reads blood-vessel tone), the AHI moved by more than 10 events per hour in 35% of patients, and severity was misclassified in 24% of cases with no first-night effect to blame (Tschopp, Journal of Clinical Sleep Medicine, 2021).
- A swing above 10 events per hour between nights affects roughly 41% of people.
- Severe results are stable enough to act on immediately. Mild and moderate results are not.
- The instability appears on home monitors, wrist tonometry and lab studies alike, so it is biology, not a faulty kit.
Where the usual advice about sleep study accuracy breaks down
Two flaws run through almost every guide here, and both work against a borderline reader.
Flaw one is treating a wrong result as an accident. The standard list blames a slipped sensor, a misreading algorithm, a glass of wine, a blocked nose. Those things happen. But the studies above followed people across consecutive nights on valid recordings, and the swing was still there. Variability is a property of your breathing, not a defect in the night.
Flaw two is discussing repeat testing only after a negative result. Guides list surgery, weight change, PAP titration and persistent symptoms after a normal study as reasons for a second test. Few address a positive but borderline result, which is exactly where single-night accuracy is worst.
The scoring rule can change your class without the night changing at all
An AHI is not one unambiguous number. A hypopnea, meaning a partial narrowing of the airway rather than a full stop, can be scored with a 4% oxygen desaturation rule or a 3% desaturation plus arousal rule, and that choice moves people across bands. In 147 participants who had two polysomnograms within 10 days, between-night classification disagreement was 29.9% with the 4% rule and 21.2% with the 3% plus arousal rule (Alavi, Chest, 2026). Hypoxic burden, which weighs the depth and duration of oxygen dips instead of counting events, was far more stable at 11.8% disagreement in the same study.
Vocabulary matters here too. A lab polysomnography (PSG) report gives an AHI or a respiratory disturbance index (RDI). A home study usually gives a respiratory event index (REI), counted per hour of recording rather than per hour of sleep, plus an oxygen desaturation index (ODI 3% or ODI 4%). Check which index and which rule produced your number.
- Two labs can score the same recording and hand you two different severities.
- Ask which index, AHI, REI or RDI, and which hypopnea rule your report used.
What drives night to night AHI variability in your own body
Two things predict it measurably: time with the head supine, and time in deep non-REM sleep. Those were the independent predictors of AHI variability in 160 adults who each completed three home sleep tests over a mean of 8.78 days with mandibular jaw movement monitoring (Martinot, Sleep Medicine, 2023). In that group, deciding from a single night would have produced 13.5% overtreatment and 6.0% undertreatment.
1How much of the night you spent on your back
Supine sleep time is the biggest lever most people have. With positional obstructive sleep apnoea, a night mostly on your side gives a mild number and a night flat on your back a moderate one. Most reports state supine time.
2How your sleep architecture fell that night
Deep non-REM sleep time independently predicted variability in the 2023 Sleep Medicine data. REM-related OSA adds a second layer. Events cluster in REM sleep, so a night with little REM may understate a real problem.
3Which night of the week you were tested
Across 70,052 adults monitored at home, the odds of a moderate-to-severe result, meaning an AHI of 15 or more, were 18% higher on Saturdays than on Wednesdays, and mean AHI ran 6% higher at weekends, an increase of 0.76 events per hour (Pinilla, American Journal of Respiratory and Critical Care Medicine, 2025). People who slept in by 45 minutes or more at weekends showed a further 47% increase in weekend odds. Social jetlag is a testing variable, so a routine Tuesday recording may undersample your real week.
4How much you actually slept
The first-night effect, meaning lighter and more fragmented sleep in an unfamiliar setting, is well described in sleep medicine. Check the total sleep time and sleep efficiency printed on your report. A short recording leaves the software less real sleep to score.
5What was different about that evening
Alcohol, sedatives and nasal congestion on the test night belong in a sleep diary, so your clinician can read the recording in context. Keep the framing honest. Treat them as context, not as proven drivers of internight AHI swings.
- Supine head time and deep non-REM sleep time are the verified predictors of your personal swing.
- A one-page diary of position, alcohol, congestion and bedtime turns a vague complaint into a specific request.
What your national pathway actually allows
Night to night AHI variability is the same everywhere. The pathway that answers it is not. A German patient asking for a second night is asking for routine practice. A UK patient is asking for an exception.
| Country | Standard first test | Where a second night sits | Detail worth quoting |
|---|---|---|---|
| France | Polygraphie ventilatoire nocturne at home, or polysomnographie in a unite du sommeil, recorded over one night, or a night and a day | A repeat is usually a hospital appointment, not another posted device | Assurance Maladie requires the polygraphie to record at least six hours, so a short or fragmented night can invalidate it outright |
| Germany | Testing in a Schlaflabor, described in public patient information as taking place for one or more nights | Already inside routine practice, so the request is normal rather than exceptional | IQWiG patient information, updated 8 July 2026, uses the phrase eine oder mehrere Naechte |
| United Kingdom | GP referral to a specialist sleep clinic, then a device worn for one overnight recording, usually at home | No routine second night in the standard NHS pathway | Severity is set from that single AHI score, so a borderline result needs an explicit argument |
Match the words to your own paperwork. A kardiorespiratorische Polygraphie is the German name for respiratory polygraphy, and German reports call this problem Nacht-zu-Nacht-Variabilitaet of the Apnoe-Hypopnoe-Index.
The French bands are also drawn differently from the ones most English-language pages quote. Assurance Maladie states the indice d'apnées-hypopnées (IAH) bands as 5 to 15 légère, 16 to 30 modérée and above 30 severe. An IAH of exactly 15 or 16 sits on a different line in France than a reader elsewhere would assume from a 15 to 30 band.
There is now a European-led consensus you can name out loud. A two-round modified Delphi study of 13 international sleep medicine experts, run from the University of Zurich and published in Sleep Medicine in 2026, agreed on 12 of 17 statements. It supports multi-night testing for a borderline AHI or when the clinical picture contradicts the result, endorses three nights no more than a month apart, and puts treatment decisions on the mean AHI across valid nights.
- France sets a minimum six-hour recording, a concrete and citable reason to ask for a repeat.
- The NHS pathway assumes a single night, so name the 2026 Zurich Delphi consensus when you ask.
A three-night protocol you can put in front of your clinician
Here is the sequence that works.
- Read the report, not the label. Find total sleep time, recording duration, supine time, REM percentage, the index used, and the hypopnea rule, either 4% desaturation or 3% plus arousal.
- Check the disqualifiers. A French polygraphie ventilatoire recorded for less than the six-hour minimum, a heavily fragmented night, or a night spent almost entirely on one side are concrete grounds to ask for a repeat.
- Write four nights of diary. Bedtime, alcohol, sedatives, nasal congestion, sleeping position, weekday or weekend. Add your Epworth Sleepiness Scale answers or a STOP-BANG score, because symptoms argue alongside the number. If you own a wearable or an under-mattress sleep sensor, bring its trend as context, never as proof.
- Ask for three nights, not one repeat. Request three recording nights no more than a month apart, with the decision resting on the mean AHI across valid nights, citing the 2026 Delphi consensus from the University of Zurich.
- Agree the decision rule in advance. Ask what happens if the mean lands at 15 rather than 16, the exact line between légère and modérée in the French bands. That is the whole point when you sit near the CPAP eligibility threshold in your country.
- Set a review date. If the answer is no, ask for a documented review in three to six months. Our guide on timing a follow-up sleep study covers when re-testing genuinely adds information.
How many nights you actually need
Three nights is the pragmatic answer to night to night AHI variability, and an honest compromise. Diagnostic performance rises from a one-night F1-score of 0.77 to 0.94 at 14 nights, then stays flat (Lechat, American Journal of Respiratory and Critical Care Medicine, 2022). One extra night is a real improvement rather than a cure. Three nights within a month captures most of the gain at a cost a clinic can accommodate.
- Ask for three nights and a decision on the mean AHI, not a single repeat.
- Bring the report details and a diary. Specifics get answered, vague doubt does not.
Once the number is settled, your options change
If your mean AHI across valid nights lands at 30 or above, pursue CPAP. Severe results were correctly classified 85% to 87% of the time on a single night (Simonds, AJRCCM, 2021; Punjabi, Chest, 2020), so that band can be trusted and delay costs you most. No home device substitutes for CPAP in severe obstructive sleep apnoea.
In the mild-to-moderate range, an IAH of 5 to 30 under the French bands and averaged across valid nights, the conversation genuinely opens. Positional therapy has a clear rationale when your reports show high supine sleep time. A mandibular advancement device, the orthèse d'avancée mandibulaire fitted by a dentist, is commonly offered for mild-to-moderate disease. Alcohol timing, weight and nasal patency sit in the same picture. A nasal stent, meaning an intranasal airway device such as Back2Sleep, a CE-certified Class I soft silicone tube that keeps the nasal airway open during sleep and is sold without prescription in a four-size starter kit, belongs in that conversation too. No machine, no hose, no electricity.
- An averaged AHI of 30 or above means CPAP, and you can act on it without further testing.
- Mild-to-moderate results support positional therapy, a mandibular advancement device or a nasal stent, chosen against your own supine and REM data.
What Back2Sleep Users Say
Frequently Asked Questions
How many nights do you need for a sleep study to be accurate?
Three is the practical answer. The 2026 University of Zurich Delphi consensus endorses a three-night protocol, no more than one month apart, with treatment decided on the mean AHI. Accuracy climbs from a one-night F1-score of 0.77 to 0.94 at 14 nights, then plateaus (Lechat, AJRCCM, 2022).
Can a sleep study be wrong?
Yes, and it is measurable. About 49% of people change severity class at least once across repeat studies, and 41% shift by more than 10 events per hour between nights (Roeder, Thorax, 2020). Errors cluster in the mild and moderate bands rather than the severe one.
Can I ask for a second sleep study if my result was borderline?
Yes. A borderline AHI is one of the situations the 2026 Zurich Delphi consensus names for multi-night testing. Bring specifics: total sleep time, supine time, and the hypopnea scoring rule used. In France, a polygraphie ventilatoire shorter than six hours can be challenged on Assurance Maladie criteria alone.
Does it count if I barely slept during my sleep study?
It depends on how much scored sleep the recording captured. Very short or heavily fragmented nights give the software little to work with, and France requires a polygraphie ventilatoire to record at least six hours. Ask for the total sleep time and sleep efficiency figures before accepting the result.
What is night-to-night variability in sleep-disordered breathing?
Night-to-night variability, or NtNV, is the genuine change in your apnoea-hypopnoea index from one night to the next. It is not equipment failure. Time with the head supine and time in deep non-REM sleep were the independent predictors of that swing in a 2023 Sleep Medicine study.
Is a home sleep apnea test as accurate as a lab study?
Both share the same weakness: one night. A home test over three consecutive nights classified normal results correctly about 93% of the time and severe results 87%, but misclassified roughly 20% of mild and moderate cases against the three-night average (Punjabi, Chest, 2020). Nights matter more than setting.
Can a GP test for sleep apnoea or do I need a sleep clinic referral?
In the UK, a GP refers you to a specialist sleep clinic, which then arranges one overnight recording, usually at home. Germany routinely records one or more nights in a Schlaflabor. France uses a polygraphie ventilatoire at home or polysomnographie in a hospital sleep unit.
Does the day of the week affect sleep apnea test results?
Yes. Across 70,052 adults, the odds of a moderate-to-severe result were 18% higher on Saturdays than Wednesdays, and mean AHI ran 6% higher at weekends (Pinilla, AJRCCM, 2025). A midweek test can undersample your worst nights, so tell your clinician which night you recorded.
Ready for quieter nights? Discover the Back2Sleep starter kit and find the right fit for you.
Not sure if you are at risk? Take our sleep risk screening to find out in just a few minutes.
Want to learn how it works? Explore the Back2Sleep nasal stent designed for comfortable, effective relief.