Is the Measurement of Sleepiness the Holy Grail of Sleep Medicine?

Donald L. Bliwise

Emory University Medical School, Atlanta, Georgia

	ARTICLE

TOP ARTICLE REFERENCES

Sleepiness is a curious construct. Any clinician evaluating sleep disorders patients invariably encounters it. Yet sleepiness is familiar to all of us at some time. That a medically relevant symptom should and can be distinguished from a ubiquitous, commonplace occurrence is not a new notion. How that distinction is to be made and differentiating the commonplace from a marker of disease, by virtue either of its frequency, its intensity, its timing, or perhaps other qualitative properties, only adds allure to the conundrum of the measurement of sleepiness. Traditionally, the value of an index of disease is ascertained through the properties of sensitivity/specificity and predictive value. But what happens when the process of measurement and even, perhaps, the construct itself are not the subject of consensus?

There are no shortage of methods introduced worldwide that have sought to quantify sleepiness. Self-reports represent one approach and include a panoply of developed scales ranging from the Stanford Sleepiness Scale, Epworth Sleepiness Scale, Sleep-Wake Activity Inventory, Karolinska Sleepiness Scale, and a variety of 100-mm linear scales. Physiological measurements represent another option that, at least in principle, should be more amenable to objective measurement and free from reporting biases. The Multiple Sleep Latency Test (MSLT) and Maintenance of Wakefulness Test (MWT) represent two of the most widely-accepted measurements. Other less commonly used methods but with great potential for ambulatory application are extent of eyelid closure and blink speed. Purely behavioral approaches, such as simple repetitive performance tasks (e.g., Wilkinson Addition Test), have a long history to detect sleepiness in experimental situations, whereas more recent behavioral measures rely upon computerized administration, have considerable face validity for lapses during driving (1, 2), and may be of great clinical relevance.

In the midst of this plethora of options of quantifying sleepiness, the current issue of this journal (pp. 1619-1625) (3) highlights a head-to-head comparison of two promising measures: the PVT (4), a test examining manual reaction time in response to a randomly occurring flash, and the OSLER test (5), a test requiring subjects to press a button in response to a light regularly appearing every 1 of 3 s. Before describing the authors' interpretation of their data, it is important to step back and ask perhaps a more fundamental question: what exactly do we seek from a measure of sleepiness? When will we know when the perennial search is over? The answer may well vary from center to center, from clinician to clinician, and from researcher to researcher, but a few basic tenets of measurement must hold. The measure must be reliable (at least within the temporal confines for which this is possible, see below); it must reflect some element of functional impairment; it must validly measure the underlying brain process(es) it purports to measure; it should have divergent validity (i.e., so as to distinguish it from other related, yet conceptually different markers, such as depression and fatigue); it must reflect the effects of treatment designed to ameliorate its occurrence; and last, but certainly not least, it must be expeditiously administered and economically viablea tall order indeed to measure a phenomenon that most of us experience every night before going to bed.

With this as background, Priest and coworkers (3) appear to break new ground along the seemingly endless journey for establishing measurement of sleepinessor do they? The OSLER test is expeditiously administered, does not require electrophysiological monitoring, and according to Priest and coworkers, provides as valid a result with only three administrations during the day as it does with four. The OSLER test may be exquisitely sensitive to lapses in attention even apart from altered EEG (theta activity), although four consecutive misses on the OSLER test also appear to be a good indicator of EEG-defined microsleeps. Priest and coworkers (3) also imply that the OSLER test may be a better test of vigilance than the PVT, but here they are simply wrong. Reanalysis of their data presented in the web site indicates that both the PVT and OSLER reflect the effects of sleep deprivation, but effect sizes for the PVT pre/postsleep deprivation were, in fact, larger (d = 2.04) than those calculated from the OSLER results (d = 1.55). Examination of the raw data suggests why that should be the case. Because the OSLER test sessions are truncated at 40 min, this essentially provides a ceiling effect and limits the variance in the test results. The PVT, by contrast, employs no ceiling or floor and allows for a larger variance in scores. Thus, to the aforementioned list can be added one additional, seemingly mundane, requirement for a measurement of sleepiness that is largely statistical in naturethe procedure in question should generate metrics that maximize, rather than truncate or censor, variance.

One paradox regarding the measurement of sleepiness implicitly discussed by Priest and coworkers is the uneasy tension between conceptualizing sleepiness as a trait or as a state. In the current study, a "state" notion is employed, inasmuch as the authors are examining the PVT and OSLER before and after sleep deprivation. As the authors point out, the sleepiness experienced by patients with sleep apnea or narcolepsy (i.e., as a characteristic "trait") may be different in "intensity or quality" from that experienced by the normal subjects in their protocol. They are correct in not equating the two but just how far does this go? In the general population, accumulating data suggest that sleepiness may be a trait with high heritability (6). Conversely, chronobiology teaches us that sleepiness varies by the hour, and perhaps even by the minute. Are patients with sleep apnea and narcolepsy immune to such effects? When and how is their "trait" sleepiness manifested in a "state"-like fashion? Whither sleepinessstate or trait? The answer is probably somewhere in the middle. This issue also makes consideration of reliability of measurement particularly difficult: reliability over how much time under what conditions?

Neurobiologists stress that the ability to remain awake may arise from neural substrates different from the ability to fall asleep (7), a distinction that has a procedural counterpart in the distinction between the MWT and MSLT. Ironically, as basic scientists struggle to fully examine the systems underlying these phenomena, the differentiation of results based on these procedures often assumes great practical importance in the medical/legal arena. But the MSLT and MWT, which perhaps represent the current gold chalice for measurement of sleepiness, cannot be insulated from controversy. Questions still arise regarding the number of naps required (particularly relevant for issues of reliability), the preferential use of a mean or median to characterize an individual's "trait" level, the presence of normal rapid eye movement (REM) onsets (8), accrued sleep effects on subsequent naps, the effect of stimulation between naps, the number and type of signals that best capture sleep onset, and the value of defining sleep on the basis of 30 s epochs versus recognition of microsleeps.

Perhaps the most overlooked aspect in any discussion of sleepiness is thus the fundamental nature of the brain process we are measuring. In other words, what does it mean when a waking brain verbally reports "sleepiness" or cannot rapidly generate a motor response on a proximity switch? Is sleepiness best seen as a metabolic effect (9) or are short-term changes in particular molecules most relevant (e.g., adenosine,

nitric oxide, prostaglandins, ammonia-releasing brain lipids)? Might the newly-described orexin system be the most viable candidate to examine the intrusion of sleep onto wakefulness and vice versa? It may well be that the quest to attain the best measurement of sleepiness ultimately may end only when those brain mechanisms that define such processes on a second-to-second basis are better elucidated.

	References

TOP ARTICLE REFERENCES

1. Findley L, Unverzagt M, Guchu R, Fabrizio M, Buckner J, Suratt PM. Vigilance and automobile accidents in patients with sleep apnea or narcolepsy. Chest 1995; 108: 619-624 [Abstract/Free Full Text].

2. George CF, Boudreau AC, Smiley A. Simulated driving performance in patients with obstructive sleep apnea. Am J Respir Crit Care Med 1996; 154: 175-181 [Abstract].

3. Priest B, Brichard C, Aubert G, Liistro G, Rodenstein DO. Microsleep during a simplified maintenance of wakefulness test: a validation study of the OSLER test. Am J Respir Crit Care Med 2001; 163: 1619-1625 [Abstract/Free Full Text].

4. Dinges DF, Powell JW. Microcomputer analyses of performance on a portable, simple visual RT task during sustained operations. Behav Res Methods, Instrument Comput 1985;17:652-655.

5. Bennett LS, Stradling JR, Davies RJO. A behavioural test to assess daytime sleepiness in obstructive sleep apnea. J Sleep Res 1997; 6: 142-145 . [Medline]

6. Carmelli D, Bliwise DL, Swan GE, Reed T. A genetic analysis of the Epworth Sleepiness Scale in 1560 World War II male veteran twins in the NAS-NRC Twin Registry. J Sleep Res 2001; 10: 53-58 . [Medline]

7. Jones BE. Basic mechanisms of sleep-wake states. In Kryger MH, Roth T, Dement WC, editors. Principles and practice of sleep medicine. Philadelphia: Saunders; 2000. p. 134-154.

8. Rye DB, Bliwise DL. Frequency of multiple sleep onset REM periods in "control" subjects (letter). Sleep 1997; 20: 813 [Medline].

9. Thomas M, Sing H, Belenky G, Holcomb H, Mayberg H, Dannals R, Wagner H Jr,, Thorne D, Popp K, Rowland L, Welsh A, Balwinski S, Redmond D. Neural basis of alertness and cognitive performance impairments during sleepiness. I. Effects of 24 h of sleep deprivation on waking human regional brain activity. J Sleep Res 2000; 9: 335-352 .