ABC of Sleep Medicine (ABC Series)

Chapter 2

Diagnosing Sleep Disorders


·        If available, a reliable sleep history is generally more useful than sleep investigations in diagnosing sleep disorders

·        Cyclical dips in blood oxygen saturations overnight form the basis of diagnosing sleep apnoea syndromes and can be demonstrated by simple home oximetry tests in most cases

·        Sophisticated home ambulatory recording equipment can be used to confirm sleep apnoea if there is diagnostic doubt

·        Full overnight polysomnography provides a clear objective snapshot of a subject's sleep quality but always needs to be analysed in a clinical or symptom-based context

·        Accurately measuring objective levels of sleepiness and wakefulness is very difficult, partly due to uncertainties over the normal range

·        The multiple sleep latency test (MSLT) and maintenance of wakefulness test (MWT) are considered ‘gold standards’ for measuring or quantifying levels of sleepiness but are very sensitive to procedural variations and protocols

·        Actigraphy is a simple technique that gives a useful approximation of a subject's sleep–wake cycle over a period of weeks by continuously measuring limb movements

Largely due to the perceived mysterious nature of sleep and the difficulties in investigating it, many clinicians lack confidence when diagnosing sleep disorders. This partly reflects the wide spectrum between normal and abnormal but also reflects the relatively low profile of sleep medicine in most educational curricula.

The most widely accepted international classification of sleep disorders (ICSD) was last revised in 2005 and outlines eight major categories (Box 2.1). Although a useful and exhaustive guide, it is written from a largely American perspective that it not always applicable or appropriate to clinical practice in other countries. Furthermore, it is clear that the relatively new field of sleep medicine continues to evolve rapidly and that many uncertainties and ambiguities remain.

Box 2.1 The eight main categories of sleep disorder as defined by the International Classification of Sleep Disorders (ICSD) manual (2nd edition)

1. Insomnia

2. Sleep-related breathing disorders

3. Hypersomnias of central origin, not due to breathing disorders or other causes of disturbed sleep

4. Circadian rhythm sleep disorders

5. Parasomnias

6. Sleep-related movement disorders

7. Isolated symptoms, possible normal variants and unresolved issues

8. Other sleep disorders

A common misperception amongst non-specialists is that sophisticated investigative techniques are invariably needed for accurately diagnosing sleep disorders. However, a directed and reliable sleep history is almost invariably more useful than sleep tests (Table 2.1). Clearly, difficulties may arise if helpful or essential collaborative information from a bed partner or family member is not available. A self-completed sleep diary for at least two weeks can be a useful addition to history taking (Figure 5.2 shows an example).

Table 2.1 Some important sleep-related symptoms and their implications.

Sleep-related symptom


Short sleep time (<6 hours)

Short sleeper (constitutional); all types of insomnia; consider depression; circadian rhythm disorder (especially delayed sleep phase syndrome [DSPS])

Irregular sleep times

Social or work related; circadian rhythm disorder

Delay in falling asleep

Sleep-onset insomnia, look for secondary cause(s)

Difficulty waking in morning

Sleep deprivation; sleep inertia (sleep ‘drunkenness’); idiopathic hypersomnolence

Restless during night

Frequent arousals (look for causes of secondary sleep-maintenance insomnia); obstructive sleep apnoea; periodic limb movements during sleep; rarely epileptic phenomena

Complex movements during sleep

Parasomnias; nocturnal epilepsy may need to be considered


Simple or primary snoring with no adverse consequences to the subject; upper airway resistance syndrome; obstructive sleep apnoea

Nocturnal choking

Obstructive sleep apnoea; gastro-oesophageal reflux; vocal cord adduction causing stridor; panic attacks

Unrefreshing sleep

Sleep restriction; all causes of secondary sleep-maintenance insomnia, including conditions associated with hyper-arousal such as fibromyalgia

Daily early morning headache

Carbon dioxide retention; obstructive sleep apnoea

Daytime naps

Insufficient sleep; all causes of secondary insomnia; narcolepsy; idiopathic hypersomnia

Weakness, localised or general, with emotion

Cataplexy, invariably in association with narcolepsy

Pre-sleep apprehension

Anxiety; psychophysiological insomnia; fear of event during sleep (nightmares, parasomnias)

Pre-sleep leg discomfort and movements

Restless legs syndrome and associated periodic limb movements

This chapter will focus on diagnostic investigations used to confirm clinically suspected sleep disorders. It should be emphasised that information from sleep testing is not always diagnostic in itself and invariably needs to be taken in context with a patient's sleep–wake symptoms of concern. It is not rare for minor abnormalities picked up on sleep tests to be misleading or misinterpreted.

Tests for sleep-related breathing disorders

Before embarking on formal treatment for a sleep-related breathing disorder such as obstructive sleep apnoea (OSA), it is mandatory to provide some objective evidence to confirm the diagnosis.

Many clinics specialising in sleep apnoea will only accept patients for a diagnostic work-up if there are associated indications of daytime hypersomnolence. The most commonly accepted simple screen for assessing levels of somnolence is the Epworth scale. In this subjective assessment, a subject has to rate the chances of falling asleep (between 0 and 3) over the previous few weeks in eight routine situations (Box 2.2). A score of over 10/24 on the scale is often taken as the arbitrary criterion for accepting referrals to a sleep apnoea clinic.

Box 2.2 The eight stem situations from the Epworth scale in which a subject is asked to assess their propensity to sleep

·        Sitting and reading

·        Watching TV

·        Sitting inactive in a public place (e.g. theatre or meeting)

·        Sitting as a passenger in a car for an hour without a break

·        Lying down to rest in the afternoon when circumstances permit

·        Sitting and talking to someone

·        Sitting quietly after lunch without alcohol

·        Sitting in a car while stopped for a few minutes in traffic

Patient rates each item as

0 (would never doze) to

3 (high chance of dozing)

ESS total score:

0 → 24

ESS = Epworth sleep scale

The Epworth scale is easy to administer but highly subjective, with a significant range of normality in the general population (Figure 2.1). Furthermore, some of the questions may not be valid or appropriate for certain individuals, including children.

Figure 2.1 Daytime sleepiness in a population-based sample. In this representative sample of the general population, almost 20% scored 11 or over on the Epworth scale, indicating possible excessive daytime sleepiness. Severe OSA patients or those with narcolepsy would be expected to score 15 or over.



Ambulatory oximetry is a simple and inexpensive screening tool for OSA that can be undertaken in a subject's home. A finger probe measures oxyhaemoglobin saturation during the recording period of one or more nights along with pulse rate (Figure 2.2). In clear cut cases of sleep apnoea, a characteristic pattern of saturation dips with associated pulse rises is seen (Figure 2.3a and 2.3b). A desaturation index of significant oxygen dips per hour measuring more than 4% is often taken as a useful guide of OSA severity.

Figure 2.2 An example of a finger probe used in home oximetry.


Figure 2.3 (a) An example of an overnight oximetry recording showing numerous cyclical dips in oxygen concentration through the night (red lines) and significant pulse rate variability (blue lines); this indicates severe sleep apnoea. (b) A normal overnight oximeter recording for comparison.


Published data imply considerable range in both the specificity (40–100%) and sensitivity (25–99%) of oximetry, reflecting the many factors that may influence the quality or reliability of the test (Table 2.2). If there is strong clinical suspicion and inconclusive data from simple oximetry, more sophisticated testing is usually appropriate.

Table 2.2 Factors influencing the diagnostic yield of ambulatory oximetry in OSA.

Patient factors

Body habitus

Obese subjects have decreased functional residual capacity with reduced oxygen stores; relative hypoxaemia may produce false positive data


Underlying pulmonary disease

Persistently low oxygen saturations may produce false positive results because the oxyhaemoglobin dissociation curve is steep


Patient sleep status

Oximetry does not detect sleep; if awake, false negative data may be obtained

Technical factors

Positioning of probe

Artefact is common due to incorrect positioning or factors such as nail varnish, thick or pigmented skin


Sampling rates

The oximeter can be programmed to have different ‘response times’ which can give major variations in output data

Defining abnormal results

Studies have differing criteria for abnormal levels of apnoeas and hypopnoeas

Oxygen desaturation index may be better guide; small oscillations in desaturation levels may be difficult to interpret

Oximetry is commonly used to confirm treatment outcomes for sleep apnoea at clinical follow-up.

Home ambulatory recordings with movement detectors and respiratory monitors

Increasingly, more sophisticated home recordings are undertaken, particularly if there remains diagnostic doubt after simple oximetry.

One widely available type of equipment uses additional chest muscle leads to record respiratory effort and electromyographic electrodes on limbs to monitor gross movements. Along with oxygen desaturation patterns and pulse rate changes, the additional data may confirm OSA and help to exclude central sleep apnoea (Chapter 4) as a cause for dips in oxygen saturation. If excessive limb movements are recorded at regular intervals, periodic limb movement disorder (PLMD) may be considered in the differential diagnosis.


There is little doubt that full polysomnography (PSG) remains the ‘gold standard’ investigation for sleep physicians. By measuring multiple parameters simultaneously it allows sleep to be accurately staged and its overall quality assessed with respect to variables such as limb movements and breathing. However, it is expensive, requires the input of skilled technologists and usually requires the subject to sleep in an unfamiliar environment whilst heavily monitored. The availability and, most likely, the quality of PSG vary greatly from region to region.

A typical PSG set-up will have at least the following features:

·        Two electrodes placed centrally near the vertex to record the surface electroencephalogram (EEG), vital in sleep staging;

·        Electrodes around each eye to monitor movements of the globe, especially to record REM sleep;

·        Electromyographic electrodes under the chin and on each shin to measure muscle movement and tone, essential for recording the usual lack of muscle tone in REM sleep and for measuring periodic limb movements;

·        Oximetry recording;

·        Chest and abdominal leads to record ‘strain’ and muscular effort during respiration;

·        An accelerometer to record body position and shifting of position;

·        Audio and video recording under low level or infrared lighting;

·        Electrocardiographic (ECG) monitoring.

Optional additional EEG electrodes (up to 12) may be placed over the skull if epileptic seizures are in the differential diagnosis. Nasal and mouth airflow can be assessed indirectly by thermistors that record temperature changes between inspired and expired air. In some specialised units, a small oesophageal balloon can be introduced to record intra-thoracic pressure as an accurate index of respiratory effort.

The considerable amount of data generated by PSG is usually summarised graphically (Figure 2.4). Although sophisticated software is often used to analyse sleep stages, the high levels of artefact require a trained technician to manually confirm any computerised analysis.

Figure 2.4 Overall, the trace reveals severe sleep apnoea that is primarily obstructive in nature, worse when the subject is lying supine and with some relation to REM sleep. Sleep is generally fragmented and numerous leg movements are seen, mostly in relation to respiratory events.


There are several limitations and controversial issues regarding PSG (Table 2.3).

Table 2.3 Limitations and issues regarding PSG.

Pitfalls in interpreting PSG


Artefacts during recording

There is considerable potential for technical problems to arise during overnight PSG recording; common examples are: electrode placements slipping, movement artefact obscuring other recordings, electrical interference

Ideally a technician should be present throughout the recording to monitor and minimise these issues

Normal variants

Deciding whether a given parameter is within the normal range can be difficult; the effects of age are considerable and normative data for each age group are not established; deciding whether minor EEG arousals are clinically significant is often debatable; inter-observer differences in scoring PSGs may be significant

Inadequate sleep

A subject may sleep particularly poorly in the PSG environment; complex cases are often recorded for two consecutive nights to minimise the disturbing so-called ‘first night effect’; paradoxically, some subjects with primary insomnia may sleep particularly well away from the home environment

Prior sleep history

The nature and amount of sleep in the days preceding a PSG recording may heavily influence the interpretation of a single overnight recording

More than one sleep disorder

Not infrequently, more than one sleep disorder may be revealed (e.g. excessive periodic limb movements and obstructive sleep apnoea); it can be difficult to decide on treatment options and which is the most clinically relevant finding

Drug effects

Many drugs may interfere with accurate interpretation of PSG recordings; common examples include the REM sleep suppressant effects of most antidepressants and the potential exaggeration of obstructive sleep apnoeas by sedatives and opiates

Protocols to guide which sleep-disordered subjects would merit from full PSG are lacking. Most would recommend PSG in a specialist centre if:

·        the condition is likely to be life-long and potentially disabling (e.g. narcolepsy);

·        there are atypical features (e.g. sudden onset of possible sleep walking in adulthood with no previous history of similar phenomena in childhood);

·        behavioural disturbances during sleep are potentially injurious (e.g. violent REM sleep behaviour disorder);

·        attempts to treat a sleep disorder have failed (e.g. persisting daytime sleepiness despite elimination of obstructive apnoeas by CPAP therapy);

·        there is likely to be more than one sleep disorder (e.g. periodic limb movements and obstructive sleep apnoea).

An example of a case where PSG recording was helpful in making a diagnosis unsuspected from the history is shown in Box 2.3.

Box 2.3 Case example in which overnight recording was helpful in establishing a diagnosis of a sleep disorder (periodic limb movements of sleep) that was not suspected from history alone

A 45-year-old man was referred for assessment of unrefreshing overnight sleep and troublesome daytime sleepiness (Epworth sleep scale score = 14). For several years, despite obtaining at least eight hours sleep every night, he awoke feeling ‘groggy’.

There were no clues from his history indicating the cause of his likely poor quality overnight sleep. He was thin and did not snore although lived alone, making the history a little unreliable. He denied restless legs syndrome but was aware his bedclothes were usually disrupted on waking in the morning.

His polysomnogram (Figure B2.3.1) revealed a degree of sleep-onset insomnia and frequent leg movements through the night, most of which were periodic in nature. The movements were associated with pulse rate rises and minor arousals on his EEG.

Figure B2.3.1 A sample of the PSG showing the overnight hypnogram (insert) and a two-minute trace with frequent and regular left leg movements (LM). Breathing parameters were all normal (not shown).


He was diagnosed with periodic limb movement disorder and responded well to a dopamine agonist drug taken before bed.

The PSG was helpful not only in showing his excessive leg movements but also demonstrating they were adversely affecting his sleep continuity and causing minor arousals from sleep.

The majority of parasomnia cases do not require PSG confirmation of the diagnosis if a clear history is obtained. Furthermore, it is rare to obtain useful information from PSG in those with chronic primary insomnia.

Objective tests of sleepiness and wakefulness


There are no truly objective tests that will directly assess a subject's levels of sleepiness. The most widely used surrogate measure is the multiple sleep latency test (MSLT). A subject is asked to lie on a bed fully clothed and encouraged to fall asleep in a sleep-inducing environment whilst monitored. The average latency to fall into stage 1 non-REM sleep is recorded for four or five nap opportunities at two-hourly intervals through the morning and early afternoon. The depth of any sleep obtained is noted and whether or not REM sleep is achieved. Each nap opportunity lasts 20 minutes.

The MSLT is very sensitive to minor procedural variations and a strict protocol is essential. In general, an average sleep latency of 10 minutes or less may be considered abnormal although there is a wide variation in control populations. Perhaps paradoxically, sleep latencies also tend to increase with age even though the elderly are generally viewed as more ‘sleepy’. Ideally, the previous night's sleep should be monitored to assess any possible contribution of sleep deprivation. Drugs potentially influencing sleep quality and quantity should be stopped, if possible, at least a week before the test.

Patients with significant daytime sleepiness will normally have mean sleep latencies of eight minutes or considerably less. In control subjects, one night of sleep deprivation will typically produce a subsequent sleep latency of around three minutes. In narcolepsy, a latency of eight minutes or less with at least two of the naps containing REM sleep within 15 minutes would fulfil the strict diagnostic criteria.

Although considered a ‘gold standard’, the MSLT should always be interpreted with caution. The wide variation of sleep latencies obtained from MSLTs in the general population often produces both false positive and negative results.


In real-life situations, the ability to stay alert is arguably more important to measure or confirm than the ability to fall asleep easily. As with the MSLT, however, deciding an abnormal result can be difficult and normative data for tests assessing wakefulness are lacking.

In drug trials, the maintenance of wakefulness test (MWT) is the most widely used index of wakefulness. As with the MSLT, the subject is asked to recline in a potentially sleep-inducing environment with EEG and video recording. Unlike the MSLT, however, the instruction is to stay awake, usually for 40-minute sessions, subsequently repeated four times. If sleep is witnessed within 15 minutes or so, despite attempts to stay awake, the result may be considered abnormal.

A less labour-intensive test of wakefulness is the Oxford Sleep Resistance (OSLER) test. Usually for four 40-minute sessions, a subject has to respond manually on a button to a regularly flashing light. Lack of any motor response for seven consecutive three-second intervals is recorded as a sleep episode. Shorter periods of unresponsiveness indicate impaired vigilance possibly secondary to ‘microsleeps’. There appear to be good correlations between sleep latencies obtained by the OSLER test and the MWT.


Actigraphs are small devices resembling a wristwatch that record small movements, usually at the wrist, when a subject is presumed to be awake (Figure 2.5). Recordings can be made over a period of weeks and a profile obtained of the sleep'wake cycle (Figure 2.6).

Figure 2.5 An example of a typical actigraph. The person is wearing two varieties of a commonly used device.


Figure 2.6 An example of a normal actigraph recording taken over one week. Time of day is shown at the top of the graph. The lines indicate the levels of arm movement and imply a regular waking time of 06:30 with lessening activity levels during the evening. Overnight, movement is minimal, between 23:00 and 06:30.


A large amount of data is gathered and modern devices will also record parameters such as audio signals, light levels and pulse rate.

Using movement as a surrogate measure for wakefulness is clearly prone to providing false positive and negative data but actigraphy is an inexpensive and unobtrusive technique. It is most useful in confirming diary information on sleep–wake cycles and also in investigating possible disorders of circadian timing (Figure 2.7).

Figure 2.7 An abnormal actigraph recording in a subject with a rare abnormality of circadian timing. It demonstrates a tendency to sleep an hour later each day through the two weeks of recording. This helps to diagnose non-24-hour sleep phase syndrome in which a subject has an internal clock that runs on approximately 25 hours a day rather than 24 hours. This is most commonly seen in individuals who are blind from birth and unable to entrain their clock mechanisms to the day–night cycle via pathways from the retina to the hypothalamus.


Further reading

American Academy of Sleep Medicine (2005) International classification of sleep disorders: diagnostic and coding manual, 2nd edition. American Academy of Sleep Medicine, Westchester, IL.

Ancoli-Israel, S., Cole, R., Alessi, C. et al. (2003)The role of actigraphy in the study of sleep and circadian rhythms. Sleep26, 343–392.

Carskadon, M.A., Dement, W.C., Mitler, M.M. et al. (1986) Guidelines for the Multiple Sleep Latency Test (MSLT): A standard measure of sleepiness. Sleep9, 519–524.

Keenan, S.A. (1992) Polysomnography: technical aspects in adolescents and adults. J Clin Neurophysiol9, 21–31.

Raymond, B., Cayton, R.M. and Chappell, M.J. (2003) Combined index of heart rate variability and oximetry in screening for the sleep apnoea/hypopnoea syndrome. J Sleep Res12, 53–61.

If you find an error or have any questions, please email us at Thank you!