Polysomnography (Sleep Study) Longevity-Medicine Target Ranges

The direct answer

Polysomnography measures airflow, respiratory effort, oxygen saturation, brain waves, and limb movement across a full night to diagnose obstructive sleep apnea (OSA) and score sleep architecture. The American Academy of Sleep Medicine (AASM) classifies an apnea-hypopnea index (AHI) under 5 events per hour as normal, 5-14.9 as mild OSA, 15-29.9 as moderate, and 30 or higher as severe, and it recommends treatment at AHI 15+ regardless of symptoms, or at 5-14.9 with daytime sleepiness, hypertension, or oxygen desaturation. Longevity-focused clinics frequently apply tighter internal targets, such as AHI under 2 or sleep efficiency above 88%, on the reasoning that cardiovascular and cognitive risk appear to rise gradually even within the "normal" range in some cohort studies. That reasoning is plausible but has not been tested as a treatment target in a trial; it is extrapolation from association, not a validated clinical threshold.

Why "normal" and "optimal" are not the same claim

A report that reads "AHI 4.8, normal study" tells you the patient does not meet the diagnostic criteria for OSA. It does not tell you whether six to nine hours of mild intermittent airflow limitation each night, repeated for a decade, carries measurable long-term risk. Multiple observational cohorts (Wisconsin Sleep Cohort, Sleep Heart Health Study, and others cited in the sleep medicine literature) have reported that cardiovascular and cognitive risk markers trend upward well before AHI crosses the diagnostic threshold for mild OSA. That is a genuine and reasonable basis for clinical concern.

Improving specific sleep metrics such as AHI from 4 to 1 or sleep efficiency from 85% to 90% has not been independently shown to alter hard outcomes including mortality, stroke, or dementia incidence. The published literature reviewed for this article contains no randomized trials comparing longevity outcomes between aggressive sleep targets and standard clinical thresholds. Values such as "AHI under 2" or "N3 above 18%" reflect targets adopted by certain longevity clinics as practical benchmarks and should not be interpreted as American Academy of Sleep Medicine recommendations. Quantitative findings from specific cohort studies cited in preliminary drafts could not be confirmed by direct review of the original papers and have been removed or restated in general terms pending verification.

OSA severity classification (established, AASM)

The AASM recommends treatment at AHI 15 or higher regardless of symptoms, and at 5-14.9 when accompanied by excessive daytime sleepiness, hypertension, mood disorders, or documented oxygen desaturation. This classification is a current clinical guideline position, not a longevity-specific construct, and general background on it is available at aasm.org.

Sleep architecture: general reference ranges

Standard sleep-medicine references describe roughly these ranges for a healthy adult under 60 across a full night, expressed as a percentage of total sleep time:

• N1 (lightest stage): typically under 5%; higher values suggest sleep instability
• N2: roughly 45-55%
• N3 (slow-wave sleep): roughly 15-25%, declining somewhat with age
• REM sleep: roughly 20-25%

Slow-wave sleep is the stage most closely linked to nocturnal growth hormone secretion, and REM sleep is linked to memory consolidation and to the nocturnal LH pulses that support morning testosterone. Both of these mechanistic links are well described in sleep physiology literature. The specific numeric associations sometimes quoted for these stages (for example, precise growth-hormone or dementia-risk percentages tied to a stage-percentage cutoff) require verification against a checked primary source before they should be used to counsel an individual patient, and none are reproduced here as fixed thresholds.

Alcohol, benzodiazepines, and several nonbenzodiazepine hypnotics (zolpidem, eszopiclone) are known to suppress N3 even at therapeutic doses. A patient reporting poor sleep quality despite an adequate total sleep time and a "normal" AHI is a reasonable candidate for architecture review, particularly if nightly sedative-hypnotic use is present.

Oxygen saturation during sleep

Oxygen desaturation index (ODI) counts drops in SpO2 of 3% or 4% below baseline per hour and tracks closely with AHI. Absolute SpO2 nadir below 90% is a recognized marker of clinical concern, and the AASM guidance is that sustained nocturnal SpO2 between 80% and 89% warrants treatment regardless of the AHI value. Time spent below 90% SpO2 (T90) has been proposed in the sleep literature as an AHI-independent marker of cardiovascular risk; the specific magnitude of that association reported in earlier material could not be verified here and should be confirmed against the primary study before being cited as a fixed number.

Sleep efficiency and the arousal index

Sleep efficiency is total sleep time divided by time in bed. A result at or above 85% is generally considered adequate. Efficiency below 80% overlaps with DSM-5 criteria for insomnia and is commonly associated with elevated evening cortisol and impaired glucose regulation in the broader sleep-medicine literature.

The arousal index counts EEG arousals per hour. A value above roughly 10 per hour indicates fragmented sleep even with a normal AHI. Frequent respiratory effort-related arousals (RERAs) without meeting the airflow-reduction threshold for a scored hypopnea describe upper-airway resistance syndrome (UARS), a pattern that can produce daytime fatigue and autonomic symptoms despite a technically normal AHI.

A decision framework for borderline and discordant PSG results

Most disagreements about a sleep study happen at the margins: an AHI that is "normal" but not reassuring, or one abnormal parameter buried inside an otherwise unremarkable report. This framework is meant to help a reader (or the ordering clinician) decide what a specific pattern actually calls for, and where the evidence support for that action is strong versus provisional.

The common failure mode this framework is built to catch: treating a single normal number (usually AHI) as if it clears the whole study, when oxygen data, architecture, or arousal burden tell a different story sitting one line below it.

Testosterone therapy and sleep apnea: a bidirectional relationship

Exogenous testosterone can reduce upper-airway muscle tone and, in some patients, worsen airway collapsibility, which is why professional guidance recommends screening for OSA before starting therapy in men with sleep-apnea symptoms or risk factors such as obesity, and monitoring for worsening OSA as a recognized adverse effect once therapy begins. Separately, disrupted or fragmented sleep from OSA can suppress the nocturnal LH pulses that support testosterone production, so treating significant OSA is a reasonable step to take before attributing low testosterone to age or other causes.

The specific effect sizes sometimes quoted for AHI change on TRT, or testosterone change after CPAP treatment, could not be verified against a checked primary source in this review and are not reproduced as fixed numbers. A clinician making an individual treatment decision should confirm the current literature and the patient's own baseline study rather than relying on a population average.

GLP-1 receptor agonists and sleep apnea

Weight loss from GLP-1 receptor agonists such as semaglutide and tirzepatide is a recognized mechanism for OSA improvement, supported by trial evidence including a large tirzepatide-versus-placebo trial in adults with obesity and OSA published in the New England Journal of Medicine in 2024. The precise magnitude of AHI reduction reported in that trial should be confirmed against the published paper before being quoted to a patient; this article intentionally does not restate specific numbers here pending that verification. A reasonable clinical practice is to reassess CPAP pressure requirements 6-12 months after substantial weight loss, since a pressure that was correct at a higher body weight can become excessive afterward and itself disrupt sleep.

Periodic limb movements

A PSG also scores periodic limb movements of sleep (PLMS). A PLMS index above roughly 15 per hour with associated arousals is classified as periodic limb movement disorder, which can fragment sleep in a way that produces daytime symptoms similar to mild OSA, despite a normal AHI. In a patient with non-restorative sleep and a normal AHI, reviewing the PLMS index is a reasonable next step before concluding the study is unremarkable. Reported prevalence figures for PLMD in older adults vary by study population and detection method; a specific percentage is not restated here without a verified source.

Choosing between an in-lab study and a home test

A home sleep apnea test (HSAT) records airflow, respiratory effort, and oximetry, typically three to four channels, but does not record EEG and so cannot stage sleep or score arousals. The AASM considers HSAT appropriate for adults with a high pre-test probability of moderate-to-severe OSA and no significant comorbidities. An in-lab PSG records considerably more channels and is the appropriate choice when sleep architecture, arousal scoring, or limb-movement data are clinically relevant, which is most of the time in a longevity-medicine evaluation. HSAT devices are also understood to underestimate AHI relative to in-lab studies because they typically use total recording time rather than confirmed sleep time as the denominator; the exact underestimation range should be confirmed against current device-specific literature rather than assumed to be a fixed figure.

What a complete report should include, and when to repeat testing

A report that states only "AHI 3.2, normal study" is not sufficient for a longevity-oriented evaluation. A more complete requisition or report should include AASM-compliant staging percentages, ODI at the 3% and 4% thresholds, T90, SpO2 nadir, arousal index broken out by type, and a PLMS index with and without associated arousals.

Reasonable, though not formally guideline-mandated, retesting triggers include: initiating or discontinuing testosterone therapy, initiating or discontinuing a GLP-1 receptor agonist alongside significant weight change, new or worsening bed-partner-reported snoring, or an unexplained drop in morning testosterone without a dose change. A borderline baseline study (AHI 2-5, for example) is a reasonable candidate for a repeat study a few months after starting testosterone therapy, given the guideline-recognized risk of worsening OSA on treatment.

When to seek care sooner than a scheduled repeat study

Witnessed pauses in breathing accompanied by chest pain, new or worsening high blood pressure, morning headaches with confusion, or oxygen saturation readings in the 80s reported by a home pulse oximeter warrant prompt medical evaluation rather than waiting for a routine follow-up sleep study. Severe daytime sleepiness that affects safe driving is also a reason to seek evaluation without delay.

Evidence boundary

Established: the AASM's AHI-based severity classification and treatment thresholds; the recommendation to screen for OSA before starting testosterone therapy in at-risk men and to monitor for worsening OSA during treatment; the general mechanism by which GLP-1-driven weight loss improves OSA; the basic physiology linking slow-wave sleep to growth hormone release and REM sleep to memory consolidation and gonadal hormone pulsing.

Plausible but unproven: that pushing PSG metrics beyond standard "normal" thresholds toward stricter longevity-practice targets (AHI under 2, sleep efficiency above 88%, N3 above 18%) produces measurable improvements in lifespan, cardiovascular events, or dementia risk beyond what standard-of-care OSA treatment already achieves.

Not established from the material available for this review: precise numeric effect sizes for testosterone change on CPAP, AHI change on TRT, AHI reduction from specific GLP-1 agents, or PLMD prevalence in older adults. These figures appeared in an earlier version of this article attached to citation identifiers that could not be verified, and they have been generalized or removed rather than restated as fact.