Polysomnography (Sleep Study): Which Tests to Order Alongside

What a sleep study measures, and what it does not

Polysomnography is an in-lab, technician-attended overnight test. It differs from a home sleep apnea test (HSAT), which records fewer channels (typically airflow, effort, and oximetry) and is used for a narrower population with a high pretest probability of moderate-to-severe OSA. Full PSG adds EEG, EOG, and chin EMG, which allows staging of sleep (N1, N2, N3, REM) and detection of arousals, periodic limb movements, and some parasympathetic and cardiac events that a home test cannot capture.

The core output of PSG is the apnea-hypopnea index (AHI), the number of apneas and hypopneas per hour of sleep. Under American Academy of Sleep Medicine scoring conventions, an AHI of 5 to 14 events per hour is generally described as mild OSA, 15 to 29 as moderate, and 30 or higher as severe. These thresholds are widely used in sleep medicine, though exact scoring rules have been revised over time and a treating clinician's report should be read alongside the specific criteria that lab used.

What PSG does not do is explain why sleep is fragmented at a hormonal, metabolic, or nutritional level. A patient can have a normal AHI and still have significant thyroid dysfunction, iron-deficiency-related restless legs, or an elevated HbA1c that a sleep study alone will never surface. That gap is the reason blood work is usually ordered around the same visit.

Should thyroid testing accompany every sleep study?

Hypothyroidism and OSA have a reported bidirectional relationship: an underactive thyroid can promote upper-airway narrowing through soft-tissue changes and blunted ventilatory drive, and untreated OSA is associated with metabolic changes that can complicate thyroid interpretation. Published work, including a meta-analysis pooling multiple observational cohorts, has reported that both subclinical and overt hypothyroidism are associated with increased odds of OSA, though the source material for this article listed a specific pooled odds ratio and sample size that could not be verified against the underlying paper and should be checked before it is quoted to a patient or used in a clinical note.

A reasonable practical position: TSH and free T4 are inexpensive, low-burden tests, and checking them alongside a diagnostic sleep study is a low-risk way to catch a treatable contributor to fatigue, weight gain, and airway narrowing. Free T3 adds value mainly when TSH is normal but clinical suspicion remains high, or when a patient is on biotin supplements, which can distort some immunoassay-based thyroid results.

Should testosterone be checked before or after OSA treatment?

Sleep fragmentation, particularly loss of REM sleep, is reported to blunt the nocturnal testosterone pulse that normally peaks in the early morning hours. Multiple observational studies have found lower average testosterone in men with more severe OSA compared with men without it, though the exact magnitude of difference reported in the source draft for this article (a specific milligram-per-deciliter gap tied to a named cohort) could not be verified and should not be repeated as a precise figure without checking the primary paper.

This is the key sequencing question for men with both low testosterone and untreated moderate-to-severe OSA: treat the sleep apnea first, then re-measure testosterone, rather than diagnosing hypogonadism and starting testosterone replacement therapy (TRT) at the same visit as the OSA diagnosis. This sequencing is consistent with Endocrine Society guidance on excluding reversible causes of low testosterone, including untreated OSA, before committing a patient to long-term TRT. TRT itself can worsen OSA in some patients, which is an additional reason not to start both treatments simultaneously without re-testing in between.

If testosterone testing is ordered, total testosterone, free testosterone (preferably by equilibrium dialysis, or calculated from SHBG), and SHBG should be drawn together. SHBG is relevant because obesity, common in OSA, lowers SHBG and can make total testosterone look falsely reassuring while bioavailable testosterone is actually low. Morning blood draws, generally before 10 a.m., are recommended because testosterone has a strong diurnal pattern.

A note on attribution: an earlier version of this article included quotations attributed to named physicians discussing testosterone and sleep apnea sequencing, and a claim describing an internal HealthRX.com cohort of 312 men. Neither the quotations nor the internal cohort data could be verified for this revision, so both have been removed. The sequencing recommendation above is retained because it is consistent with published Endocrine Society guidance, not because of the removed quotes or data.

Does OSA change the case for checking HbA1c and lipids?

OSA and insulin resistance are linked through a plausible mechanism: intermittent hypoxia triggers sympathetic activation, raises hepatic glucose output, and can impair pancreatic beta-cell function over time. Cohort studies, including large population-based samples, have reported higher rates of prevalent diabetes and dyslipidemia among people with more severe OSA compared with those with little or no sleep-disordered breathing, independent of body weight in adjusted analyses. Several of the exact effect sizes cited in the original source material (a specific adjusted odds ratio, a specific milligram-per-deciliter change in triglycerides and HDL) could not be traced to a verifiable source during this review and are presented here only in general, non-numeric terms until they can be checked.

HbA1c is a reasonable test to pair with a sleep study because it reflects roughly a 90-day glycemic average and is not distorted by a single night of disrupted sleep, unlike a fasting glucose drawn the same morning. The American Diabetes Association classifies prediabetes as an HbA1c of 5.7% to 6.4% and diabetes as 6.5% or above, and this cutoff comes directly from the ADA's current Standards of Care rather than from sleep medicine literature ADA Standards of Care, Diagnosis and Classification of Diabetes. A fasting lipid panel is a reasonable companion test given the reported association between OSA severity and dyslipidemia, though it should be interpreted as part of general cardiovascular risk assessment rather than as an OSA-specific biomarker.

Why check a CBC and iron studies

Chronic intermittent hypoxia can stimulate erythropoietin release and drive secondary erythrocytosis. A complete blood count is inexpensive and detects this, along with anemia, which independently worsens daytime fatigue and can be mistaken for OSA-related sleepiness. Specific prevalence figures for secondary erythrocytosis in newly diagnosed OSA patients that appeared in the original draft could not be verified and are omitted here; the general point, that hemoglobin should be checked, stands on its own without a precise percentage attached.

Restless legs syndrome (RLS) is reported to co-occur with OSA at meaningfully higher rates than in the general population, and brain iron deficiency is considered a contributing mechanism in RLS even when serum iron studies are only mildly low. Published guidance from restless legs syndrome specialty groups has recommended considering iron repletion at ferritin levels below roughly 75 ng/mL, a threshold well above the typical laboratory reference range for iron deficiency (often 12 to 15 ng/mL), because brain iron stores can be low even when standard lab cutoffs look normal. Ferritin is an acute-phase reactant, so checking CRP alongside it helps flag a falsely normal ferritin in the setting of inflammation.

Cortisol, vitamin D, and other tests to consider

Morning cortisol can be useful in patients with an elevated AHI and persistent fatigue despite treatment, as fragmented sleep and hypoxia are believed to affect the hypothalamic-pituitary-adrenal axis in some patients. A cortisol below the lower end of the reference range should prompt consideration of an ACTH stimulation test rather than a diagnosis on its own; a markedly elevated level with clinical signs of Cushing syndrome should prompt dexamethasone suppression testing. This is specialist-level interpretation and not something to self-diagnose from a single lab value.

Vitamin D deficiency is common in the general population and has been reported at elevated rates among OSA patients in observational studies. An association between low vitamin D and OSA severity has been proposed, but causality is not established, and it is plausible that shared risk factors (obesity, reduced outdoor activity) explain part of the link rather than a direct physiological effect. Checking 25-hydroxyvitamin D is reasonable because deficiency is common, inexpensive to correct, and unrelated risks of untreated deficiency (bone health, muscle function) justify testing on their own merits.

hs-CRP is a nonspecific inflammation marker. Elevated CRP in a patient with severe OSA is consistent with the systemic inflammatory changes reported in this population, but it should be interpreted as one input into cardiovascular risk assessment, not as a way to grade OSA severity.

Additional tests to consider only with a specific clinical trigger: IGF-1 if growth hormone deficiency is suspected clinically (OSA fragments slow-wave sleep, where a large share of daily GH secretion normally occurs); NT-proBNP if there are signs of right heart strain or poorly controlled hypertension, since OSA is a recognized and common contributor to resistant hypertension; and prolactin if hypersomnia persists despite a normal or only mildly abnormal AHI, since a pituitary process is a rare but important alternative explanation.

What is established, what is plausible, and what is not established

Established: PSG measures breathing events, oxygenation, sleep stages, and limb movements, and produces an AHI used to classify OSA severity. TSH/free T4 and HbA1c are inexpensive, low-risk, widely recommended screening tests with clear interpretive cutoffs from endocrine and diabetes guideline bodies. Iron-deficiency-associated RLS responds to iron repletion in patients with low ferritin, and specialty groups use a higher ferritin threshold for RLS than for classic iron-deficiency anemia.

Plausible but not fully proven for an individual patient: That correcting vitamin D deficiency, treating subclinical hypothyroidism, or repleting borderline-low ferritin will measurably improve a given patient's AHI or daytime sleepiness. These associations exist at the population level in observational data, but observational association is not the same as a guarantee of individual benefit, and some of these relationships may be confounded by shared risk factors like obesity.

Not established from the material available for this article: Precise numeric effect sizes (specific odds ratios, specific milligram-per-deciliter shifts, specific percentage improvements in AHI or HbA1c after treatment) that appeared in an earlier draft of this page could not be traced to verifiable primary sources during this review. They have been removed or converted to general statements rather than presented as confirmed facts. A qualified reviewer should check the primary literature before any of these numbers are restored to a patient-facing page.

Untreated moderate-to-severe OSA is associated with meaningfully higher long-term cardiovascular and metabolic risk in multiple large cohort studies, which is the underlying reason CPAP adherence and companion lab monitoring both matter, even though the exact multiplier of risk reported in the earlier draft of this page could not be independently confirmed here.

Building a practical paired-test order set

A reasonable panel to draw the morning after a diagnostic sleep study, fasting where possible:

Core panel, reasonable for most adults undergoing diagnostic PSG:

• TSH, free T4
• CBC with differential
• Comprehensive metabolic panel
• HbA1c
• Fasting lipid panel
• Ferritin, serum iron, TIBC
• 25-hydroxyvitamin D
• hs-CRP

Add based on symptoms, exam, or sex-specific risk:

• Total testosterone, free testosterone, SHBG in men with an elevated AHI, and in women with irregular menses or hirsutism
• Morning cortisol in patients with a moderate-to-severe AHI or unexplained persistent fatigue
• IGF-1 if growth hormone deficiency is clinically suspected
• NT-proBNP if there is hypertension, edema, or suspected right heart strain
• Prolactin if hypersomnia persists with a normal or only mildly elevated AHI
• Estradiol and FSH in perimenopausal women with new-onset OSA symptoms

Decision framework: sequencing labs and treatment around a sleep study

This framework is meant to help a clinician or informed patient decide what to order and when to re-test, given the sequencing problems described above.

Step 1: Order the core panel around the time of PSG, not before it. Draw the core panel (TSH/free T4, CBC, CMP, HbA1c, lipids, iron studies, vitamin D, hs-CRP) the morning after the study, fasting if feasible. Rationale: same-visit blood work avoids a second appointment and captures a morning hormonal snapshot close to the sleep data.

Step 2: Before ordering testosterone, ask whether OSA is already confirmed and untreated.

• If OSA is untreated and moderate-to-severe: order testosterone for context, but do not diagnose hypogonadism or start TRT yet. Re-test after a defined period of confirmed CPAP or oral appliance use.
• If OSA is mild or absent: proceed with a standard hypogonadism workup on its own timeline.
• Exception: symptoms concerning for pituitary disease (visual field changes, galactorrhea, severe headache) should not wait on a CPAP trial; escalate to prolactin testing and imaging referral instead.

Step 3: Distinguish a ferritin that needs treating for RLS from one that needs treating for anemia.

• Ferritin below roughly 75 ng/mL with limb movement symptoms or an elevated PLMS index on PSG: consider iron repletion aimed at RLS, using the RLS-specific threshold, not the general anemia cutoff.
• Ferritin low with a low MCV and no limb-movement symptoms: this points toward a separate anemia workup, which may need its own cause-finding (for example, gastrointestinal blood loss) rather than an RLS-focused approach.
• CRP should be checked alongside ferritin; an elevated CRP can make a truly low ferritin look falsely normal.

Step 4: Decide what to re-test, and when, based on what treatment actually changed.

• If CPAP was started: re-test HbA1c, testosterone (in men with baseline low levels), and hs-CRP only after a period of confirmed, adherent use, not immediately after diagnosis. Confirmed adherence (device data showing consistent nightly hours) matters more than elapsed calendar time.
• If levothyroxine was started for newly diagnosed hypothyroidism: recheck TSH per standard endocrine follow-up intervals, generally in the range of several weeks, rather than at the same time as the OSA re-test.
• If iron repletion was started: recheck ferritin after a period of consistent supplementation, and expect a slower symptom response than the ferritin number itself.

Failure mode to watch for: starting TRT and initiating OSA treatment at the same visit. This makes it impossible to know afterward whether testosterone normalized because of TRT or because of the OSA treatment, and it exposes the patient to a medication (TRT) that untreated OSA can make less safe by worsening apnea severity in some patients.

When PSG findings need urgent follow-up

A sleep study report showing severe oxygen desaturation, evidence of significant cardiac arrhythmia (such as sustained bradycardia, heart block, or atrial fibrillation captured on the single-lead ECG), or a very high AHI in a patient with cardiovascular disease warrants prompt clinical follow-up rather than waiting for routine paired labs to come back. PSG can detect these events, but it does not replace a full cardiology evaluation; an abnormal cardiac finding on a sleep study is a reason to seek that evaluation, not a diagnosis on its own.

A note on the evidence behind this page

Several numeric claims in an earlier version of this article, including specific odds ratios, milligram-per-deciliter differences, cohort sizes, physician quotations, and an internal clinical dataset, could not be verified against traceable primary sources during this review. Rather than repeat unverifiable precise figures, this revision states the general direction of each association and flags where a clinician or editor should confirm the primary literature before a specific number is used in patient-facing material. The one external reference retained with a specific link is the American Diabetes Association's current diagnostic criteria for prediabetes and diabetes, which is a stable, checkable guideline document.