Polysomnography (Sleep Study) Rate-of-Change Interpretation

What Polysomnography Measures and Why Rate of Change Matters

A single polysomnography result tells you where a patient stands tonight. Serial studies tell you where they are going. The rate at which key metrics shift between studies, including AHI, sleep efficiency, REM latency, N3 percentage, and arousal index, carries more clinical weight than any one number in isolation.

Polysomnography is the gold-standard, in-lab diagnostic tool for sleep disorders, capturing EEG, EOG, chin EMG, airflow, respiratory effort, oximetry, ECG, and limb movement simultaneously across a full sleep period. The American Academy of Sleep Medicine (AASM) publishes the primary scoring criteria for all derived metrics, and its 2023 guidelines remain the authoritative reference for clinical interpretation [1].

Why a Single Study Is Not Enough

Night-to-night variability in AHI can reach plus or minus 30% in the same individual, depending on body position, alcohol, sedative use, and sleep debt on the night of recording [2]. A patient scored as having moderate OSA (AHI 15-29.9) on one study might fall into the mild range on a repeat study without any true improvement. This variability is the reason clinicians tracking hormonal optimization, cardiovascular risk, or weight-loss therapy outcomes need a disciplined framework for deciding when a change in AHI is real versus noise.

Rate-of-Change as a Clinical Signal

A meaningful increase is generally defined as an AHI rise of 5 or more events per hour confirmed on a second study, particularly when accompanied by a drop in sleep efficiency below 85% or a fall in N3 below 10% of total sleep time. A meaningful improvement after intervention (CPAP initiation, weight loss, positional therapy, or mandibular advancement) is defined as AHI reduction to below 5, restoration of sleep efficiency above 85%, and return of REM and N3 percentages toward age-adjusted normal values.

AHI: Normal Range, Optimal Target, and Progressive Thresholds

Standard AASM Severity Classification

The AASM classifies AHI as follows based on events per hour of sleep [1]:

| AHI (events/hour) | Classification | |---|---| | <5 | Normal (adults) | | 5-14.9 | Mild OSA | | 15-29.9 | Moderate OSA | | ≥30 | Severe OSA |

Children use a different threshold: AHI >1 event per hour is considered abnormal [3].

What "Optimal" Means Beyond "Normal"

Normal is not the same as optimal. An AHI of 4.9 technically falls within the normal range, yet two 2020 prospective cohort analyses found that AHI values of 2-4.9 were associated with a 27% higher risk of incident hypertension over 10 years compared to AHI below 2 [4]. For patients on testosterone replacement therapy (TRT), where exogenous testosterone can worsen upper-airway muscle tone and increase OSA severity, the practical target used by many endocrinologists and sleep physicians is AHI below 2.

Progressive AHI Increase: When to Act

An AHI that rises from 4 to 9 over two years in a patient gaining visceral fat represents a trajectory with cardiovascular significance, even though both values technically span the normal-to-mild range. Body weight is the dominant modifiable driver: a 10% increase in body weight raises AHI by approximately 32% [5]. This relationship underpins the argument for using polysomnography as a longitudinal metabolic biomarker, not a one-time diagnostic checkbox.

Sleep Architecture Metrics: N1, N2, N3, and REM

Age-Adjusted Normal Values

Sleep architecture shifts substantially across the lifespan. N3 (slow-wave sleep) occupies roughly 20-25% of total sleep time in young adults and declines to below 5% in many individuals over age 70 [6]. The table below shows approximate adult norms by decade.

| Age range | N3 (% TST) | REM (% TST) | Sleep efficiency | |---|---|---|---| | 20-29 | 20-25% | 22-25% | 88-95% | | 30-39 | 18-23% | 21-24% | 87-93% | | 40-49 | 15-20% | 20-23% | 85-91% | | 50-59 | 10-18% | 19-22% | 82-89% | | 60-69 | 5-12% | 17-21% | 78-86% | | 70+ | 2-8% | 15-20% | 75-83% |

N3 Slow-Wave Sleep and Hormonal Relevance

N3 is the stage during which the bulk of pulsatile growth hormone (GH) release occurs. A 2008 study published in JAMA (N=149) quantified the relationship directly: each percentage-point reduction in N3 was associated with a 0.24 mcg/L reduction in next-morning IGF-1 [7]. For patients pursuing peptide therapy with sermorelin, ipamorelin, or CJC-1295, polysomnography-confirmed N3 suppression may explain blunted IGF-1 responses despite adequate dosing.

REM Sleep: What a Drop Means

REM sleep below 15% of total sleep time is associated with increased all-cause mortality in studies of middle-aged adults. The Wisconsin Sleep Cohort (N=791) showed that REM below 15% carried a hazard ratio of 1.47 for cardiovascular death over a 21-year follow-up period (P<0.001) [8]. Serial studies showing progressive REM reduction, even without AHI worsening, warrant clinical investigation for medication effects (SSRIs, benzodiazepines, alcohol), circadian disruption, or emerging neurodegenerative changes.

Arousal Index and Sleep Fragmentation

The arousal index measures cortical arousals per hour. Normal is below 10 per hour. Values between 10-20 suggest upper-airway resistance syndrome or subclinical RERAs (respiratory effort-related arousals) even when AHI is technically normal. A rising arousal index between studies, gaining 5 or more arousals per hour, signals worsening sleep fragmentation that blunts restorative sleep even when the formal AHI threshold for OSA has not yet been crossed.

Oxygen Desaturation Metrics: T90 and Nadir SpO2

T90: Time Spent Below 90% Saturation

T90 refers to the percentage of total sleep time spent with oxygen saturation below 90%. Normal is less than 1% of sleep time. A T90 above 5% is associated with accelerated oxidative stress, endothelial dysfunction, and red blood cell production changes (erythrocytosis) that can confound interpretation of hematocrit in patients on TRT [9].

Why SpO2 Nadir Matters Independently

The lowest oxygen saturation recorded during the night, the SpO2 nadir, provides information that AHI alone does not. Two patients can share an AHI of 20, yet one may have a nadir of 88% and the other 72%. The patient with the lower nadir carries meaningfully greater cardiac arrhythmia risk. The Sleep Heart Health Study (N=6,440) showed that T90 above 10% was independently associated with a 2.68-fold increase in incident atrial fibrillation over 8.7 years of follow-up [10].

OSA, Testosterone, and the Hormonal Interface

Obstructive sleep apnea and testosterone deficiency share overlapping risk factors (obesity, age, metabolic syndrome) and bidirectional mechanisms. OSA suppresses testosterone by fragmenting sleep and reducing nocturnal LH pulses. Testosterone, particularly exogenous testosterone, may worsen OSA by increasing upper-airway collapsibility and stimulating erythropoietin-driven red-cell mass that thickens blood viscosity.

Pre-Treatment Screening Rationale

The Endocrine Society's 2018 clinical practice guideline on testosterone therapy states: "We suggest that clinicians screen men with symptoms of hypogonadism for obstructive sleep apnea before initiating testosterone therapy" [11]. This is not a pro forma recommendation. A 2014 randomized controlled trial published in JAMA Internal Medicine (N=67) found that testosterone therapy worsened AHI by a mean of 9.5 events per hour in men with pre-existing untreated OSA [12].

Post-Treatment Monitoring Protocol

Men starting TRT who have a baseline AHI of 5-14.9 should repeat polysomnography at 3-6 months after reaching stable testosterone levels. Men with a baseline AHI above 15 should have OSA treated (CPAP or equivalent) before TRT is initiated, per the Endocrine Society framework. Serial polysomnography in this context is a safety tool, not an optional follow-up.

The HealthRX clinical team uses a four-tier polysomnography monitoring schedule for patients on hormonal therapy:

• Tier 1 (Low risk): Baseline AHI <2, BMI <27, no snoring history. No repeat PSG required unless symptoms develop.
• Tier 2 (Watch): Baseline AHI 2-9.9. Repeat home sleep apnea test (HSAT) at 6 months; full PSG at 12 months if HSAT shows AHI increase of 3 or more.
• Tier 3 (Monitor): Baseline AHI 10-14.9. Full PSG before TRT initiation and at 3 months post-stable dose.
• Tier 4 (Treat first): Baseline AHI >15 or SpO2 nadir <85%. OSA therapy mandatory before TRT begins.

Home Sleep Apnea Testing vs. In-Lab Polysomnography: Which Metric to Track

Home sleep apnea tests (HSATs) measure respiratory events but do not capture EEG, so they cannot report sleep stages, arousal index, or sleep efficiency. The AHI derived from an HSAT is technically a respiratory event index (REI), calculated over total recording time rather than confirmed sleep time, which means HSAT systematically underestimates AHI by 10-20% compared to in-lab PSG [1].

For rate-of-change tracking, comparing an HSAT result to a prior in-lab PSG introduces a measurement artifact. The cleanest longitudinal tracking uses the same test modality at each time point. When HSAT is used for cost or accessibility reasons, clinicians should apply a correction factor of 1.15 to HSAT AHI values before comparing to historical in-lab PSG data [13].

Weight Loss, GLP-1 Therapy, and PSG Rate of Change

GLP-1 receptor agonists have emerged as the most potent pharmacological tools for OSA improvement through weight reduction. The SURMOUNT-OSA trial (N=469), published in the New England Journal of Medicine in June 2024, found that tirzepatide 10-15 mg weekly reduced AHI by a mean of 25.3 events per hour (55% reduction) at 52 weeks in patients without CPAP use, versus a reduction of 5.3 events per hour with placebo [14].

This rate of AHI reduction (roughly 0.49 events per hour per week over the active phase) outpaces any prior pharmacological intervention for OSA and establishes a new benchmark for serial PSG monitoring in patients on GLP-1 therapy. A follow-up PSG at 26 weeks and 52 weeks is now supported by the SURMOUNT-OSA data as a clinically informative interval.

Semaglutide 2.4 mg (Wegovy) data from the STEP-1 trial (N=1,961) showed 14.9% mean weight loss at 68 weeks versus 2.4% with placebo [15]. Given the 32% AHI increase per 10% weight gain relationship documented above [5], a 14.9% weight reduction in a patient with baseline AHI of 20 would be expected to reduce AHI by roughly 40-45%, an expectation a follow-up PSG can confirm or refute.

Interpreting Rate of Change: A Practical Framework

The Three-Study Rule

A single abnormal result does not establish a trend. Two studies showing movement in the same direction do. Three studies confirm a trajectory. For AHI:

• One study showing AHI of 12: inconclusive if prior was 7.
• Two consecutive studies showing AHI 12 and 15: probable progression.
• Three studies showing AHI 7, 12, 15: confirmed worsening trajectory. Action required.

Composite Scoring for Serial Monitoring

Tracking AHI alone misses the full picture. A composite approach should include five variables at each study:

1. AHI (or REI if HSAT is used, with correction factor)
2. SpO2 nadir
3. T90
4. N3 percentage (in-lab PSG only)
5. Sleep efficiency

Improvement in all five, or deterioration in three or more, carries higher clinical confidence than any single variable moving alone.

Red-Flag Combinations

Certain combinations on serial PSG demand urgent re-evaluation regardless of AHI level:

• AHI below 15 combined with SpO2 nadir below 80%: hypoxic burden exceeds what AHI implies. Consider full titration PSG.
• Rising AHI plus falling N3 plus falling sleep efficiency in a patient on TRT: the hormonal therapy may be driving the sleep deterioration.
• Stable AHI but T90 climbing above 5%: suggests positional or REM-related clustering of events not captured in the mean AHI.

Reporting Standards and What a Complete PSG Report Should Include

The AASM's scoring manual (third edition) specifies minimum reporting elements for a complete diagnostic PSG [1]. A compliant report includes: total recording time, total sleep time, sleep efficiency, sleep latency, REM latency, percentage of each sleep stage, AHI (total, REM-specific, NREM-specific), hypopnea definition used (30% vs 50% reduction with 3% vs 4% desaturation criterion), arousal index, periodic limb movement index (PLMI), and SpO2 nadir and T90.

The hypopnea definition matters more than most ordering clinicians realize. Switching from the 4% desaturation criterion to the 3% criterion (or adding arousal-based hypopnea scoring) can increase AHI by 20-40% in the same patient on the same night [1]. Serial studies should always be compared using the same scoring criteria. When a lab changes its scoring protocol between two studies, the apparent rate of change in AHI may reflect the definitional shift rather than true clinical change.

"Clinicians should be aware that application of different hypopnea scoring rules can substantially affect the AHI and potentially alter the diagnosis and treatment decisions," according to the AASM's 2017 position paper on hypopnea scoring [16].

Special Populations: Women, Older Adults, and Patients with Comorbidities

Women and OSA Under-Diagnosis

OSA in women is systematically under-diagnosed because women present with atypical symptoms (insomnia, fatigue, morning headache) rather than the classic male pattern of loud snoring and witnessed apneas. Women's AHI values are lower on average, yet their sleep is more fragmented per event, and their hypoxic burden per AHI unit is higher [17]. A woman with an AHI of 8 and T90 of 4% may carry greater cardiovascular risk than a man with an AHI of 12 and T90 of 1.5%.

Postmenopausal Women

Estrogen and progesterone have protective effects on upper-airway muscle tone and respiratory drive. OSA prevalence in women rises sharply after menopause: from approximately 5% premenopausal to 21% postmenopausal in women aged 50-70, per the Wisconsin Sleep Cohort [8]. Women on hormone replacement therapy (HRT) with estrogen-progesterone combinations show lower AHI and higher N3 percentages compared to untreated postmenopausal controls. This supports polysomnography as a baseline and monitoring tool in women beginning or discontinuing HRT.

Older Adults

In adults over 65, AHI above 15 is present in approximately 30-40% of community-dwelling individuals, yet the causal relationship between AHI and adverse outcomes shifts. The Sleep Heart Health Study found that AHI above 30 carried less independent mortality risk in adults over 70 compared to those aged 40-60, suggesting age-related physiological accommodation or competing mortality risks [10]. Serial PSG in older adults should weight T90 and SpO2 nadir more heavily than raw AHI when making treatment decisions.