Polysomnography (Sleep Study) Medication-Driven Changes: What Your Drugs Are Doing to Your Sleep Data

What Polysomnography Actually Measures

Polysomnography is the gold-standard overnight recording of sleep physiology. A standard in-lab study captures electroencephalography (EEG), electrooculography (EOG), chin and limb electromyography (EMG), nasal-oral airflow, respiratory effort belts, pulse oximetry, electrocardiography, and body position simultaneously. The American Academy of Sleep Medicine (AASM) 2023 scoring manual defines the thresholds used to stage each 30-second epoch and score respiratory events. [1]

Core PSG Channels and What They Detect

EEG distinguishes the four sleep stages: N1 (light sleep), N2 (sleep spindles and K-complexes), N3 (slow-wave or deep sleep), and REM. Airflow and effort channels identify obstructive apneas, central apneas, and hypopneas. Pulse oximetry produces the oxygen desaturation index (ODI). Chin EMG is required to score REM atonia, the loss of which defines REM sleep behavior disorder (RBD).

Normal PSG Reference Ranges in Adults

The AASM defines normal values as follows [1]:

| Metric | Normal Range | Optimal Target | |---|---|---| | AHI (no treatment) | <5 events/hr | <2 events/hr on CPAP | | REM sleep | 20 to 25% TST | 22 to 25% TST | | N3 (slow-wave) | 15 to 20% TST | 18 to 20% TST (<60 yr) | | Sleep efficiency | >85% | >90% | | Sleep onset latency | <20 min | <10 min | | SpO2 nadir | >88% | >90% | | ODI | <5 events/hr | <2 events/hr |

These numbers shift meaningfully with age. Adults over 60 normally produce less N3, often 5 to 10% of total sleep time, without any pharmacological explanation. [2]

How Scoring Rules Affect Results

A hypopnea under the AASM 2012 "alternative" criteria requires only a 30% flow reduction plus 3% desaturation or arousal. The 2017 recommended criteria require a 4% desaturation. A single lab switching criteria can change a patient's AHI from 4 to 14 events per hour on the same raw data. [3] Clinicians ordering PSG for medication-management decisions should confirm which scoring rules the reading lab applies.

Why Medications Change PSG Metrics

Drugs alter PSG through four major mechanisms: direct neurotransmitter modulation (GABA, serotonin, norepinephrine, orexin), upper-airway muscle tone changes, respiratory drive suppression, and circadian phase shifting. A 2019 review in Sleep Medicine Reviews catalogued more than 40 drug classes with documented PSG effects. [4] The clinical consequence depends on which channel is affected, the magnitude of the change, and whether the change is reversible on washout.

GABA-Ergic Drugs: Benzodiazepines and Z-Drugs

Benzodiazepines (temazepam, clonazepam, diazepam) suppress N3 and increase N2 spindle density. In a crossover study comparing temazepam 15 mg to placebo, N3 fell from 18.4% to 9.7% of TST while sleep efficiency rose to 91%. [5] That suppression persists even after subjective tolerance develops.

Z-drugs (zolpidem, eszopiclone, zaleplon) have a similar but smaller N3 suppression effect. Eszopiclone 3 mg in a 6-week polysomnographic trial reduced N3 by approximately 4 percentage points relative to placebo. [6] Critically, neither class significantly worsens AHI in patients without pre-existing OSA, but both can raise AHI by 3 to 6 events per hour in patients with mild-to-moderate OSA by relaxing pharyngeal dilator muscles. [7]

Opioids: A Distinct Pattern

Opioids suppress both N3 and REM, increase central apneas, and produce a characteristic saw-tooth airflow pattern called cluster or ataxic breathing. A study of chronic opioid users (N=392) found central apnea prevalence of 24% versus 5% in matched controls. [8] AHI on opioids often underestimates the true respiratory burden because central events, which are scored differently from obstructive events, may dominate. Methadone produces the most pronounced central apnea burden among commonly prescribed opioids. [9]

Stimulants, Antidepressants, and Alerting Agents

SSRIs and SNRIs: REM Suppression Is the Rule

Every clinically used SSRI and SNRI substantially suppresses REM sleep. Fluoxetine 20 mg/day reduces REM from roughly 22% to 12% of TST within the first week and the effect persists throughout treatment. [10] This has two clinical consequences. First, patients on SSRIs who undergo PSG for suspected REM sleep behavior disorder may have falsely low REM percentages, obscuring the diagnosis. Second, SSRIs modestly improve OSA severity in some patients by increasing upper-airway tone via serotonergic motor neuron activation. A 2022 meta-analysis (8 RCTs, N=298) found paroxetine reduced AHI by a mean 5.1 events per hour versus placebo. [11]

Tricyclic antidepressants (TCAs) suppress REM even more aggressively than SSRIs. Amitriptyline 50 mg can reduce REM to under 5% of TST. [10] TCAs also carry significant anticholinergic burden, which may worsen periodic limb movements of sleep (PLMS).

Stimulants and Wake-Promoting Agents

Amphetamine-class stimulants (lisdexamfetamine, mixed amphetamine salts) markedly delay sleep onset and reduce TST when taken after noon. Modafinil and armodafinil have less total-sleep-time suppression but reliably increase sleep onset latency by 15 to 25 minutes at therapeutic doses. [12] A patient with untreated ADHD taking lisdexamfetamine 50 mg at 7 AM may still show a 35-minute sleep onset latency on PSG at 10 PM, which could be misread as primary insomnia.

Bupropion inhibits dopamine and norepinephrine reuptake and is the antidepressant least likely to suppress REM, making it a preferred choice when sleep architecture preservation matters. [13]

Testosterone, GLP-1 Agonists, and Hormonal Therapies

Testosterone and OSA: A Bidirectional Problem

Testosterone therapy worsens OSA in a dose-dependent fashion. A randomized placebo-controlled trial (N=67) found that testosterone gel 100 mg/day raised AHI from 7.4 to 12.3 events per hour over 18 weeks. [14] The proposed mechanism is testosterone-driven increase in upper-airway length and reduction in hypercapnic ventilatory response.

This has direct clinical relevance for TRT management. Men starting testosterone with a baseline AHI of 10 to 15 events per hour may cross the moderate OSA threshold (AHI >15) within months. The Endocrine Society's 2018 testosterone clinical practice guideline recommends screening for OSA before initiating therapy and monitoring for worsening symptoms during treatment. [15]

Men with severe untreated OSA also produce less testosterone. A cross-sectional study of 1,312 men found free testosterone inversely correlated with AHI severity (r = -0.31, P<0.001). [16] Treating OSA with CPAP raises morning testosterone by a mean of 72 ng/dL over 12 weeks. [16] The implication: a low testosterone reading drawn before OSA treatment may normalize without exogenous testosterone once CPAP is established.

GLP-1 Receptor Agonists and Sleep Apnea

Semaglutide has emerged as an active OSA treatment. The SURMOUNT-OSA trial (N=469) demonstrated that semaglutide 2.4 mg subcutaneous weekly reduced AHI by 29.4 events per hour (63% reduction) in adults with obesity-related OSA over 52 weeks, compared to 5.5 events per hour with placebo. [17] The FDA approved semaglutide (Wegovy) specifically for moderate-to-severe OSA in adults with obesity in June 2024. [18]

Weight loss of 10 to 15% of body mass, which semaglutide reliably produces, reduces pharyngeal fat deposition and raises the arousal threshold. Patients on semaglutide who undergo serial PSG may see AHI drop by 50% or more even before any CPAP pressure adjustment, requiring re-titration of CPAP to avoid over-treatment with pressure. [17]

Menopausal Hormone Therapy

Postmenopausal women on combined estrogen-progestogen therapy have lower AHI than untreated women. A study of 2,852 postmenopausal women in the Sleep Heart Health Study found hormone therapy users had 0.5 odds of moderate-to-severe OSA (AHI >15) compared to non-users (OR 0.5, 95% CI 0.3 to 0.8). [19] Progesterone specifically acts as a respiratory stimulant, increasing hypercapnic ventilatory response, which reduces the apnea threshold. Women who discontinue hormone therapy after years of use may develop new-onset OSA that only becomes apparent on PSG.

Alcohol and Cannabis: Acute vs. Chronic Effects

Alcohol's PSG Signature

Alcohol consumed within 2 hours of sleep onset suppresses REM in the first half of the night and produces REM rebound in the second half. A controlled crossover study (N=24) found 0.6 g/kg ethanol reduced first-half REM from 14% to 6% of TST and increased AHI by 8.4 events per hour in habitual snorers. [20] A PSG performed on a patient who consumed alcohol that evening will show artificially worsened OSA metrics, falsely elevated PLMS indices in the second half, and distorted slow-wave sleep distribution.

The standard pre-PSG instruction to abstain from alcohol for 24 hours exists precisely because of this acute distortion. If there is any clinical reason to suspect alcohol was consumed before a study, the study may need to be repeated. [1]

Cannabis and Cannabinoids

THC acutely suppresses REM sleep in a dose-dependent fashion. A 4-week study of daily cannabis users (N=27) documented REM suppression from 21% to 10% of TST during active use, with REM rebound to 29% after 3 nights of abstinence. [21] CBD has not shown consistent REM effects at doses below 300 mg/day. Dronabinol (synthetic THC) has been studied as an OSA treatment at doses of 2.5 to 10 mg, with the PACE trial (N=73) showing AHI reduction of 12.9 events per hour at 10 mg versus 3.4 events per hour with placebo. [22]

Clinicians ordering PSG in cannabis-using patients should document last use and dose, since THC abstinence of fewer than 72 hours may produce rebound REM that inflates the PSG REM percentage and obscures a baseline architecture diagnosis.

Interpreting PSG in the Testosterone Optimization Context

Men presenting to a hormone optimization clinic often have overlapping conditions: low testosterone, obesity, metabolic syndrome, and undiagnosed OSA. The PSG in this context is not just a sleep diagnostic, it is a safety screen before prescribing testosterone.

Pre-TRT Screening Protocol

The Endocrine Society guideline [15] and AASM position statement both support OSA screening before testosterone initiation. A reasonable pre-TRT PSG interpretation framework follows this logic:

• AHI <5: OSA excluded. TRT can proceed with standard monitoring.
• AHI 5 to 14 (mild OSA): Discuss CPAP initiation. Repeat PSG at 3 months post-TRT start.
• AHI 15 to 29 (moderate OSA): CPAP should be established and effective (AHI <5 on treatment) before TRT begins, per Endocrine Society guidance.
• AHI >30 (severe OSA): TRT is relatively contraindicated until OSA is controlled. Consider the semaglutide pathway if BMI >30 is present. [17]

On-TRT Monitoring

Once testosterone therapy is established, a follow-up PSG at 6 to 12 months is reasonable if baseline AHI was 10 to 15 events per hour, even with initial CPAP. A mean AHI increase of 4.9 events per hour has been documented over 18 weeks of testosterone gel therapy. [14] If a patient on CPAP reports re-emergent snoring or daytime sleepiness, a repeat PSG with the current CPAP pressure is indicated before attributing symptoms to inadequate testosterone dose.

Home Sleep Apnea Testing vs. In-Lab PSG: What Medications Change

Home sleep apnea tests (HSATs) record fewer channels than in-lab PSG, typically airflow, respiratory effort, and oximetry only. EEG staging is absent. This creates a measurement problem in medicated patients.

Why HSAT Underestimates AHI on Certain Drugs

Because HSAT does not record EEG arousals, it cannot score hypopneas by the arousal criterion. A patient on benzodiazepines, whose OSA produces more arousals than desaturations, will have a lower AHI on HSAT than on PSG. A 2016 study (N=120) comparing simultaneous PSG and HSAT channels found HSAT underestimated AHI by a mean of 6.2 events per hour in patients on GABAergic hypnotics. [23]

Opioid-related central apneas are also poorly captured by HSAT, since many HSATs do not reliably distinguish obstructive from central events without the effort-belts interpretation by an experienced scorer. For patients on chronic opioids or high-dose benzodiazepines, in-lab PSG is the appropriate test. [3]

Drug Washout Before PSG: Practical Guidance

Most labs recommend a 5-half-life washout for drugs with known PSG effects. Practical washout periods for common agents [4] [5] [6]:

• Zolpidem (t1/2 2.5 hr): 12-hour washout adequate for most patients
• Temazepam (t1/2 8 to 20 hr): 48-hour washout recommended
• Diazepam (t1/2 20 to 100 hr): 7-day minimum washout, active metabolites extend this
• Fluoxetine (t1/2 1 to 4 days, active metabolite norfluoxetine t1/2 4 to 16 days): Full washout takes 5 to 6 weeks and is rarely practical. Document on the PSG report.
• Methadone (t1/2 24 to 36 hr): Washout is not clinically safe. Perform PSG on maintenance dose and note it explicitly.
• Testosterone (topical): No washout needed. Note dose and duration on the order.
• Semaglutide: No washout needed for OSA-focused PSG. Document dose and weeks on therapy.

When washout is not safe or practical, the interpreting physician must annotate the PSG report with the specific drug and dose, so the AHI and architecture findings are contextualized appropriately. [1]

Key PSG Metrics to Track During Medication Titration

Serial PSG or HSAT is a monitoring tool, not just a one-time diagnostic. The metrics with the most clinical meaning across drug changes are:

AHI Trend

A 25% reduction in AHI or a change across severity category (mild to normal, moderate to mild) is considered a clinically meaningful response by AASM criteria. [1] Semaglutide's 63% AHI reduction in SURMOUNT-OSA [17] far exceeds this threshold.

REM Percentage as a Drug Sensitivity Marker

Patients who show a fall in REM below 15% on any dose of an SSRI or TCA should have that finding noted on medication reconciliation. If the patient is also being evaluated for narcolepsy or RBD, the REM suppression may invalidate the MSLT (multiple sleep latency test) or obscure loss of REM atonia. [10] The AASM recommends a minimum 2-week washout of REM-suppressing agents before diagnostic MSLT in narcolepsy workups. [24]

N3 as a Recovery Marker

Slow-wave sleep is where growth hormone is secreted, where hippocampal memory consolidation occurs, and where glymphatic clearance is highest. A 2013 study of 70 adults found growth hormone secretion correlated tightly with N3 duration (r=0.67, P<0.001). [25] Tracking N3 percentage across PSG studies in patients on GABAergic drugs, or during GLP-1-driven weight loss, gives a functional readout of recovery quality that AHI alone does not capture.

Oxygen Desaturation Index as a Cardiometabolic Risk Marker

ODI predicts cardiovascular outcomes independently of AHI. The Sleep Heart Health Study (N=6,441) found an ODI >7.5 events per hour was associated with a 26% increase in incident cardiovascular events over 8.7 years. [26] Patients whose AHI normalizes on semaglutide or CPAP but whose ODI remains elevated may need supplemental oxygen or a change in CPAP pressure settings, not just documentation of AHI improvement.