Sleep apnea activity and its concomitants in a subclinical population


Material Information

Sleep apnea activity and its concomitants in a subclinical population
Physical Description:
viii, 132 leaves : ill. ; 29 cm.
Berry, David Thomas Reed, 1958-
Publication Date:


Subjects / Keywords:
Sleep Apnea Syndromes   ( mesh )
Apnea   ( mesh )
Respiration   ( mesh )
Clinical and Health Psychology thesis Ph.D   ( mesh )
Dissertations, Academic -- Clinical and Health Psychology -- UF   ( mesh )
bibliography   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph.D.)--University of Florida, 1985.
Bibliography: leaves 123-131.
Statement of Responsibility:
by David Thomas Reed Berry.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 000522493
oclc - 16877989
notis - ACU4129
System ID:

This item is only available as the following downloads:

Full Text








The author would like to thank his chairman, Dr. Wilse

B., for his moral and material support, as well as

guidance through the completion of the project. Also due

thanks are the other members of the supervisory committee,

Dr. A. J. Block, Dr. Russell M. Eauer, Dr. F. D. McGlynn,

ann Dr. Rudy Vuchinich, for their support, advice, and

contributions. Several people provided technical assistance

and are due a vote of thanks: Dan Switzer, Alex Kasterton,

Karla Smith, Alton Howard, Albert Briggs, and Edna Pearson.

The author wishes to acknowledge the moral and

financial support of his parents, Mr. and Mrs. John G.

Berry, without which the project would never have been

undertaken. Also of considerable help were the

encouragement and advice of C. i. Carswell, D. Haymaker, anc

S. Erody, all of whom deserve thanks. Cynthia Zimmerman

provided expert typing assistance for the manuscript.



ACKNOWLEDG ENTS ....... .............. ..................... ii

LIST OF TABLES ........................................... v

ABSTRACT ...............................................vii


C!E IITRODUCTION ..................................... 1

Definitions and Terminology......................2
History and Characteristics of Sleep Apnea
Syndrome ......................................
Incidence of Sleep Apnea Activity.................5
Temporal Characteristics of Apneas...............13
Variables Associated With Sleep Apnea
Syndrome and Activity.........................17
Deficits Associated With Sleep Apnea
Cardiopulmonary Deficits.....................26
Arousal Deficits (Hypersomnolence)..........28
Oxygen-Deseturation and Cognitive Deficits..30
Measurement of Deficits Found in Sleep Apnea
Measurement of Cardiopulmonary and Health
Deficits .................................. 34
Measurement of Arousal Deficits
Measurement of Hypoxia and Its Sequelae.....38
Statement of the Problem ........................46

TWO METHOD...................................... ... 48

Subjects ........................................48
Procedure .. ................... ............... 54
Statistical Procedures..........................55

THREE RESULTS................. .... ................... 56

Demographics and Incidence of Nocturnal
Respiratory Disorder..........................56


Nocturnal Respiratory and Health Variables......63
Nocturnal Respiratory and Sleep/Wake Variables..65
EEG Data.................................... 65
Daytime Sleepiness Data.....................69
Subjective Sleep Assessment.................70
Nocturnal Respiratory and Neuropsychological
Variables..................................... 74

FOUR DISCUSSION.....................................86

Demographics and Incidence of Ilocturnal
Respiratory Disorder..........................86
nocturnal Respiratory and Health Variables......90
l!octurnal Respiratory and Sleep/Wake Variables..92
EEG Data ....................................92
Daytime Sleepiness, Sleep Questionnaire,
and Sleep Log Data........................94
Nocturnal Respiratory and Neuropsychological

FIVE SUMIIARY AND CONCLUSIONS........................103



B RAk DATA TABLES................................109


EIOGRAPHICAL SKETCH.....................................132


Table Page

2-1 Interrater agreement for respiratory events
of 46 snoring males..............................51

3-1 Demographic variables from 46 snoring males.....56

3-2 Incidence of low and high levels of apnea/
hypopnea by age in 46 snoring males.............58

3-3 Significant (p<.05) Pearson correlations
between nocturnal respirEtory variables
and demographic variables in 46 snoring
males............................................ .

3-4 Means and standard deviations for selected
demographic and nocturnal respiratory
variables in 46 snoring males grouped by
level oi apnea/hypopnea.........................61

3-5 Means and standard deviations for health
related variables in 46 snoring males grouped
by level of apnea/hypopnea......................65

3-6 Means and standard deviations of selected
sleep variables from 43 snoring males............66

5-7 Meens end standard deviations of selected
sleep variables from 43 snoring males
groupec by level of apnea/hypopnea..............68

3-8 Leans and standard deviations for daytime
sleepiness variables in 46 snoring males
grouped by level of apnea/hypopnea..............69

3-9 Significant (p<.05) Pearson correlations
between nocturnal respiratory and sleep
questionnaire variables in 46 snoring males.....71

3-10 Means and standard deviations of sleep
questionnaire variables from 46 snoring
males groupec by level of apnea/hypopnea........71

3-11 Significant (p<.05) Pearson correlations
between nocturnal respiratory and sleep log
variables in 46 snoring males.....................75

3-12 Means and standard deviations of sleep log
variables in 46 snoring males grouped by
level of apnea/hypopnea......................... 73

3-13 Multivariate regression of demographic and
nocturnal respiratory variables on cognitive
scores in 46 snoring males......................76

3-14 Significant (p<.05) partial correlations,
controlling for age, between nocturnal
respiratory and cognitive variables in
46 snoring males................................ 79

3-15 Means and standard deviations of cognitive
scores in 46 snoring males grouped by
level of apnea/hypopnea.........................80

3-16 Multivariate regression of sleep and
nocturnal respiratory scores on cognitive
variables in 43 snoring males...................85

B-1 Demographic and medical variables from 46
snoring males...................................109

B-2 Daytime sleepiness variables from 46 snoring
males........................................... 111

B-3 Sleep questionnaire variables from 46 snoring

8-4 Seven day sleep log means from 46 snoring
males ..........................................115

B-5 Neuropsychological variables from 46 snoring
males........................................... 117

B-6 Respiratory variables from 46 snoring males.....119

B-7 Sleep variables from 46 snoring males...........121

Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy




August 1985

Chairman: Wilse B. Webb
Major Department: Clinical Psychology

Sleep Apnea Syndrome (SAS) is a nocturnal respiratory

disorder with serious consequences. Patients with SAS

experience apneas (pauses in breathing of 10 or more

seconds) and hypopneas (declines in the amplitude of

breathing accompanied by oxygen desaturation) while

asleep. Recently, it has become apparent that apneas and

hypopneas may occur in otherwise apparently normal

subjects. A sample selected to display a wide range of

sleep disordered breathing (heavy snoring males in good

health and over 30 years of age) received detailed

measurement of their nocturnal respiration, sleep,

cognitive skills, and health in an attempt to clarify the

impact of apneas and hypopneas in a subclinical population.

Forty-six heavy snoring males in good health with a

mean age of 50 years and a mean weight of 190 pounds

comprised the sample. During an experimental evening they


received testing and filled out various questionnaires

before sleeping overnight with their sleep and breathing

continuously recorded. Sixty-two per cent of these subjects

experienced at least one event, while 13% had more than 5

events per hour of sleep. Most events occurred in light

slow wave or REM sleep and were linked to obesity. Subjects

with high levels of apnea/hypopnea (more than 5 per hour)

experienced significantly exacerbated oxygen desaturation

relative to the remaining subjects. Oxygen desaturation was

linked to higher blood pressure readings. A typical "first

night" effect on the sleep of the subjects in the lab was

noted, resulting in a lighter sleep than usual and obscuring

possible relationships between sleep and respiratory events,

which were not observed. Deteriorating overnight

respiratory indices were associated with increased

sleepiness and napping. High levels of apnea/hypopnea were

associated with declining scores on tests measuring non-

verbal intelligence, verbal and non-verbal memory,

expressive verbal fluency, and cognitive flexibility. It is

speculated that hypoxia induced changes in cerebral

cholinergic synthetic pathways underlie these changes.

It is concluded that a significant subgroup of heavy

snoring males who experience multiple apnea/hypopneas are at

increased risk of significant oxygen desaturation, daytime

sleepiness, and cognitive changes. This group may fall on a

continuum with sleep apnea syndrome patients.



Sleep apneas are respiratory pauses during sleep.

According to the Association of Sleep Disorders Centers

Nosology of Sleep Disorders (ASDC, 1979), a sleep apnea

syndrome (SAS) is a potentially lethal condition charac-

terized by multiple apneas, excessive daytime sleepiness,

alterations of consciousness, and cardiopulmonary

complications. The apnea episodes are thought to be the

causal factor behind the other deficits associated with

SAS (Guilleminault et al., 1978).

Polysomnographic recordings in various populations

have revealed the occurrence of sleep apneas in the

general population, and particularly in aging popula-

tions. Estimates of the prevalence of at least a few

sleep apneas in older subjects have ranged as high as 75%

(Carskadon and Dement, 1981b). Carskadon et al. (1980)

have suggested that sleep apneas in older subjects may be

implicated in insomnia, cardiac ailments, and even


However, certain reports have documented the presence

of multiple sleep apnea episodes in subjects without

apparent pathology (Crr et al., 1979). This evidence

obscures the role of sleep apneas in the other pathologies

noted in SAS patients. The link between apneas and patho-

logy may be more complex than was originally thought,

given its apparently benign occurrence in otherwise

asymptomatic subjects.

The present study will examine the correlates of

sleep apnea episodes in a population selected to display a

wide range of such events. The relationship between sleep

apnea indices and sleep characteristics, age, weight and a

variety of deficit measurements will be analyzed.

A review of the current data available on various

aspects of sleep apneas and sleep apnea syndromes will be

presented as background for the study. The diagnosis,

history, incidence, and characteristics of sleep apneas

and sleep apnea syndrome will be reviewed, followed by a

discussion of variables associated with sleep apneas.

Lastly, the deficits commonly found in sleep apnea syn-

drome will be detailed, along with measurements thought to

be sensitive to these deficits.

Definitions and Terminology

Sleep apneas have been defined as cessations of air-

flow at the mouth and nose which last for 10 or more

seconds (Guilleminault et al., 1976). In a sleep apnea

syndrome (SAS), multiple sleep apneas are accompanied by

snoring, excessive daytime sleepiness, cardiopulmonary

complications, and altered states of consciousness. The

ease of quantification of sleep apneas has led to their

use as an important tool in diagnosing SAS. Guilleminault

et al. (1976) first suggested that the occurrence of 30 or

more apneas in 7 or more hours of sleep was diagnostic of

SAS, while Guilleminault et al. (1978) proposed the use of

an apnea index (AI; number of apneas/number of hours

sleep) in excess of 5 as a cut-off score to diagnose the

syndrome. Later, Carskadon et al. (1980) combined apneas

and hypopneas (a reduction in airflow of 50% or more at

the mouth and nose) to form an apnea + hypopnea index

(AHI) with 5 utilized as a diagnostic level of sleep apnea

syndrome. As can be appreciated from the preceding dis-

cussion, the diagnostic criteria for SAS have undergone a

rapid evolution, with remarkably little validation work


History and Characteristics of Sleep Apnea Syndrome

Sleep Apnea Syndrome falls within the larger category

of Sleep Disordered Breathing (Block, 1980). Respiratory

disorders within this category involve changes in

breathing patterns during sleep. Syndromes described

within this category include Pickwickian Syndrome; Sleep

Apnea Syndrome; Chronic Obstructive Lung Disease; and

syndromes of Primary and Secondary Alveolar

Hypoventilation (Block, 1980; Guilleminault et al.,


The first disorder described in this group was the

Pickwickian Syndrome. These patients were initially

identified by a chronic daytime hypersomnolence, obesity,

Cheyne-Stokes breathing, hypercapnia and hypoxemia. This

syndrome typically results in complications such as right

sided congestive heart failure, pulmonary hypertension,

and peripheral edema (Block, 1980). An exacerbation of

disordered breathing (apnea) as well as hypercapnia and

hypoxemia (often severe) occurs with sleep onset. Treat-

ment consists of oxygen therapy, respiratory stimulants,

and maintenance of wakefulness.

The discovery of a sleep induced exacerbation of the

symptoms accompanying Pickwickian Syndrome spurred a

systematic investigation of sleep respiratory events

(Gastaut et al., 1965). The new emphasis on noctural

respiration led to a series of investigations which docu-

mented sleep induced respiratory dysrhythmias in non-obese

subjects (Lugarasi et al., 1966). A new group of sleep

disordered breathers designated as Sleep Apnea Syndrome

emerged, with several features which distinguished them

from Pickwickians. Among these differences were weight

(i.e., SAS patients were not necessarily obese), daytime

respiratory control (i.e., SAS patients were typically not

hypercapnic during waking as were Pickwickians), sex

(i.e., while AO; of Pickwickians were female, very few

premenapausal females have been identified with SAS), and

age (i.e, SAS patients are typically younger than

Pickwickians)(Block, 1980).

Two large samples of SAS patients have been reported

in the literature, allowing characterization of typical

SAS patients. Guilleminault et al. (1976) reported on a

series of 50 patients diagnosed with SAS over a period of

six years. Age ranged from 28-62 with a mean age of

45.5. The sample consisted of 48 males and 2 females.

Thirty-nine were referred on the basis of excessive day-

time sleepiness, while 7 were referred for loud snoring

and abnormal movements reported by their spouses. Sixty

percent of the patients were more than 15% above ideal

weight. Other symptoms included hypnagogic halluci-

nations, automatic behaviors, intellectual deterioration,

personality changes, impotence, morning headaches, and

hypertension. Weitzman, Kahn and Pollak (1980) report

data on a group of 10 SAS patients seen in their clinic.

These subjects displayed the following characteristics:

age between 38 and 47, obesity, male sex, nicotine depen-

dence, hypertension, complaints of excessive daytime

sleepiness, and the presence of serious cardiac arrhyth-

mias. (This group was selected from a larger sample of 38

patients.) Thus, pathological symptomatology associated

with a diagnosed SAS include excessive daytime somnolence,

cardiopulmonary complications and cognitive/intellectual

changes; a "typical" SAS patient is a middle aged male

who snores and is obese.

Incidence of Sleep Apnea Activity

The incidence of sleep apnea activity and sleep apnea

syndrome have been frequently confuse. While sleep

apneas have been an important diagnostic tool for identi-

fying sleep apnea syndrome, they have not been

definitively demonstrated to be the necessary and suffi-

cient cause of the syndrome. Thus, for the purposes of

the following review, the focus will be on occurrence and

level of sleep apnea activity in both sleep apnea syndrome

subjects and other populations. The full range of sleep

apnea activity will be reflected by noting the presence of

sleep apnea activity, and an indication of the quantita-

tive level of the activity will be presented in the apnea

index (AI) or the apnea + hypopnea index (AHI). Recall

that an AI of 5 has been frequently utilized as a cut-off

score for identifying SAS patients.

A review of the literature suggests that at least

three sampling strategies have been used for studying the

incidence of sleep epnea activity. Subjects have been

selected on the basis of one of the following three

criteria: 1) absence of sleep or respiratory complaints,

2) presence of sleep or respiratory complaints, or 3)

sampling without regard to these variables.

Because diagnosis of SAS is often dependent on number

of sleep apneas (or level of SAA), most researchers report

number of apneas observed, although some report only on

number of subjects who exceed one of the criteria for

SAS. A review of studies with samples of 20 or greater

will be presented with careful note of the sampling

strategy utilized, as well as the dependent measure of


One sampling strategy which has been pursued is to

exclude subjects who complain of sleep or respiratory

difficulties. Screening for sleep complaints varies

widely. Guilleminault et al. (1978) studied a sample of

20 normalss" (presumably noncomplaining subjects). These

subjects ranged in age from 40-60 years. Males showed a

mean number of apneas of 7 (range 1-25), while females

showed a mean number of apneas of 2 (range 0-5). Thus,

none of these noncomplaining subjects presented a clinical

syndrome of sleep apnea, although sleep apneas did occur

apparently without pathology. Block et al. (1979) studied

49 subjects (M = 30, F = 19). Males in the sample had a

mean age of 38, while females had a mean age of 29. Any

potential subjects who complained of breathing difficul-

ties or sleep disturbances were excluded from the study.

Twelve males (40%) had episodes of sleep apnea while only

3 females (15%) suffered as well. For those subjects,

males had a mean of 4.2 episodes, while females had a mean

of 3. Within this sample, the number of apneas was not

significantly correlated with increasing age, although

occurrence of oxygen desaturation (a concomittant of

disordered breathing) was positively correlated with

age. Additionally, although the mean number of sleep

apneas was not significantly different between sexes (M =

4.2, F = 3), the more sensitive measure of apneas with

desaturation was (M = 3, F = 0). Webb (1974) studied 2C

males with a mean age of 44. Subjects were excluded from

the study if they reported any "serious" sleep com-

plaint. Two of the subjects reported consuming hypnotics

for sleep induction, indicating that the criterion used in

this study was less stringent than that of Block et al.

(1979) report. Nine of the subjects (45%) suffered at

least one episode of sleep apnea with a mean number of

episodes of 2.5. All of the subjects with episodes of

sleep apnea were older than the mean age of the sample,

leading Webb to conclude that apnea is related to age.

Bixler et al. (1982) constructed an age stratified patient

representative sample of 100 subjects (M = 41, F = 59).

In perhaps the most rigorous screening for sleep com-

plaints in this group of studies, Bixler et al. eliminated

any subject with a sleep complaint or chronic medication

usage. For the entire sample, 6 males (14.6%) and 5

females (10.25) had at least one episode of sleep apnea.

For subjects under 30, 3.3%; for those between 30 and 50,

12.8e; and those 50-74, 19.4; suffered episodes of sleep

apnea. For the subsample over 60, 25% had sleep apnea

activity with 8.5 episodes for males and 10.5 episodes for

females. In this sample, there was a nonsignificant trend

toward more apnea activity in males and a significant

positive correlation between age and sleep apnea.

block et al. (1980) report data on 20 post menapausal

women with a mean age of 59. Eight women (40%) had epi-

sodes of sleep apnea, with a mean number of episodes of

5. There was a significant correlation between apnea and


For noncomplaining subjects then, males suffer a

higher incidence of apnea activity in virtually every

study which compares sex effects. For males, incidences

range widely from a low of 14.6% (Bixler et al., 1980) to

a high of 45% (Webb, 1974); while females range from a

low of 10.2% (Bixler et al., 1982) to a high of 40% (Block

et al., 1980). These differences are probably due at

least in part to age differences in male samples, as well

as age differences in females (only as they reflect mena-

pausal status). Reported number of apneas range from 2.5

(Webb, 1974) to 8.5 (Bixler et al., 1982) in males and

from 2 (Guilleminault et al., 1978) to 10.5 (Bixler et

al., 1982) in females. Many studies report a positive

correlation between number of apneas and increasing age.

A second sampling technique which has been utilized

has been studying subjects with sleep and breathing com-

plaints. Kales et al. (1982) studied 200 subjects with

primary complaints of insomnia. There were 82 males and

118 females in the sample with a mean age of 42.3. Sleep

apnea activity was noted in 10.5% of this group (M =

13.4%, F = 8.5%) with a mean number of 11.2 episodes for

males and 6.6 for females. Subjects with sleep apnea

activity were significantly older (46.8) than those with-

out (40.6). Ancoli-Israel et al. (1981) recruited 24

subjects whose answers to questionnaires provoked

suspicion of insomnia or nighttime breathing and muscular

events. Eleven males (7 = 72.5) and 13 females (R = 68.5)

were studied. All of these subjects, save one male and

one female, suffered at least one episode of sleep

apnea. For those subjects with at least one episode of

sleep apnea, males had a mean of 53 episodes of apnea,

while females had a mean of 32 episodes. In contrast to

the Kales et al. study, 6 males had a high level of sleep

apnea (AI>5), while 3 females met this criterion. A rela-

tionship between age and apnea was not reported in this

sample, possibly because of restricted range. These two

studies of sleep dist'r'bed subjects provide conflicting

data. That is, the insomniac alone sample did not demon-

strate an apnea incidence which was substantially

different from a noncomplaining sample (10.5%), while the

sample recruited for sleep induction, respiratory, and

muscular differences showed very discrepant apnea inci-

dences (nearly 100%). One potential explanation involves

age, as the Ancoli-Israel sample is substantially older

than Kales', although the aging samples have not reported

this level of apnea before. An alternative explanation is

that sampling insomniacs does not generate a high number

of apneas, whereas aging sleep and respiratory disturbed

subjects show a very high incidence of apnea. It would

seem then that the relevant "complaint" would involve

those reported by Ancoli-Israel et al. which are in

addition to those noted by Kales et al. These are (as

nearly as can be judged) questions surrounding


The final category of incidence studies involves

those which do not consider sleep complaints in their

sampling. As in the other studies, these vary in their

generalizability. Carskadon and Dement (1981b) recruited

40 aging subjects, including 18 males (R = 72.7) and 22

females (R = 74). Their exclusion criteria are somewhat

problematic, involving excluding subjects who "spon-

taneously complain of sleep problems." Thus, it is

unclear what this is a representative sample of, specifi-

cally whether the distribution of those not spontaneously

complaining of sleep disturbances may be different from

those who actually do not suffer from sleep distur-

bances. These considerations aside, Carskadon and Dement

report that 16 males (88%) and 16 females (72%) demon-

strated at least one sleep apnea episode, while 8 males

(44%) and 7 females (31.8%) have high levels of sleep

apnea (AI>5). Males with apnea showed a mean number of

51, while females showed a mean number of 34. Sex

differences appear to have fallen out at this age, while

the restricted range of this extreme age sample does not

lend itself to a correlational analysis of age versus

apnea. In another study of limited generalizability,

Kreiss et al. (1982) randomly selected inpatients on a

Veterans Administration medical ward. These patients, all

presumably male, numbered 26. Further data reported were

sketchy, but 7 (27%) met the sleep apnea syndrome

criterion of >30 episodes of sleep apnea. The fact that

these subjects were patients on a medical ward obviously

restricts generalizability of these findings. Finally, in

the only truly random sampling reported, Ancoli-Israel et

al. (1982) present initial data on randomly selected sub-

jects. Forty subjects (sex not reported) with a mean age

of 71.5 were studied. Thirteen (32.5%) met the sleep

apnea syndrome criterion of 30 or more apneas. The sleep

apnea subgroup was older, 74.2, than the full sample.

The data collected from relatively random sampling

are primarily based on aging populations (>60). The data

from these studies suggest that between 32.5% (Ancoli-

Israel et al., 1981) and 37% (Carskadon and Dement, 1981b)

of aging subjects suffer high levels of sleep apnea

(AI>5). An even larger percentage, perhaps greater than

75%, suffer at least some episodes of sleep apnea. A

major deficiency in these randomly sampled studies is a

lack of data on younger subjects.

Thus data collected under different sampling proce-

dures yield variable incidence rates of sleep apnea and

AI>S. Subjects without sleep or respiratory complaints

have a very low incidence (perhaps 0) of AI>5, between 14%

and 455 of males surveyed had at least some apnea, while

females ranged between 10% and 40% experiencing some

apnea. Sampling subjects with sleep or respiratory com-

plaints suggested that subjects suffering respiratory

difficulties had very high incidences of apnea (almost

100% for both males and females) with increase of inci-

dence of AI>5 (55% for males, 23% for females). Sampling

without regard to sleep or respiratory difficulties is

largely confined to aging populations. Incidences of

apnea in these aging populations would seem to be rather

high (88% males, 72% females) while incidence of AI>5 seem

to be in the range of 32%-37%.

It is clear that different populations (as defined by

sampling strategies) suffer different incidences of sleep

apneas and AI>5. The clinician dealing with sleep dis-

orders would be wise to consider the acturial base rates

of apnea and AI>5 provided by these data. It is apparent

that sex, age, and respiratory status are significantly

related to the incidence rates of SAA and SAS. These

factors must be considered in any study of apnea and asso-

ciated deficits.

Temporal Characteristics of Apneas

The temporal characteristics of sleep apneas repre-

sent a potentially important variable. Above it was shown

that a wide variability in number of sleep apneas occurs,

and the arbitrary nature of the division between SAA and

SAS subjects was noted. Therefore, it is of interest to

determine whether the duration of sleep apneas provides a

more reliable distinction between the two groups. Another

variable of interest involves the placement of apneas

within sleep, as Guilleminault et al. (1976) noted that

normals might experience apneas during REM without

pathological consequence. Therefore, data available on

these two issues will be reviewed, including only studies

with a sample size great than 10 (with one exception).

Reports on normal subjects below 30 years of age are

somewhat rare, although some do exist. Bixler et al.

(1982) included a group aged 18-29 years (M = 13, F =

17). One subject in this group had sleep apneas (12) with

a mean duration of 15 seconds. Bixler et al. noted that

most apneas occurred in Stages 1 and REN sleep. Part of

the sample of Block et al. (1979), the females, had a mean

age of 29. Three of these subjects showed apneas with a

mean duration of 14 seconds. Again, a disproportionate

number of apneas occurred in Stages 1 and REM. Thus in

normal subjects under 30, apneas are somewhat rare, with a

mean duration of approximately 15 seconds when they do

occur. Most apneas occur in light and REM sleep.

More data are available on normal subjects between 30

and 60. Guilleminault et al. (1978) do not report on

duration of apneas in their sample aged 40-60, but they

note that apneas occurred only in Stages 1 and REM.

Bixler et al. (1982) reported that, in their 50-44 year

old group, 5 subjects experienced sleep apnea with a mean

duration of 13.1 seconds. A disproportionate number ol

apneas occurred in Stages 1 and REM. Block et al. (1979)

studied 30 males with a mean age of 38, with 12

experiencing sleep apneas. These apneas had a mean

duration of 20 seconds and occurred most frequently

(30/60) in Stages 1 and REM. Eight of 20 aging females,

in the report by Block et al. (1980), had sleep apneas

with a mean duration of 19 seconds. Sixty-two percent of

apnea episodes occurred in Stages 1, 2 or REM. Thus in

middle aged normal subjects, incidence of sleep apnea

appears to rise, with a mean duration of between 13 and 20

seconds, perhaps slightly longer than in younger groups.

Again, apneas occurred mostly in light (Stages 1 and 2)

and REM sleep.

Duration data on aging normals are limited.

Carskadon and Dement's (1981b) study of aging normals (M =

72, F = 74) does not report duration of apnea events.

Similar, Ancoli-Israel et al. (1982) do not provide this

information on their aged normals (R = 71.5). Bixler et

al. (1982) noted a high incidence of apneas in their sub-

jects over 60, but failed to report a mean duration for

this group. Given the sketchy details reported for this

group, it is difficult to draw any conclusions about

changes in temporal duration or placement of apneas in

aging normals.

To summarize the temporal data in normals, apneas

appear somewhat rarely in normals below 30, with a mean

duration of 15 seconds. Incidence of apneas increase

between 30 and 60, with a wider variation in mean dura-

tions reported (13-20 seconds). Apneas occur most

frequently in light and REM sleep in normals of all ages.

Temporal data on subjects diagnosed with sleep apnea

syndrome (>30 or AI>5) primarily include patients between

the ages of 30 and 60. The reason for the lack of older

SAS subjects is unclear, particularly in light of data

reported above, demonstrating very high percentages of SAS

in certain groups of aging subjects.

Garay et al. (1981) studied 13 sleep apnea syndrome

subjects. They divided these 12 males and 1 female into

two groups, based on their daytime resting CO2 levels.

Subjects who were eucapnic during the day had a mean apnea

duration of 16 seconds, while those who were hypercapnic

had a mean duration of 17 seconds. No data on distribu-

tion relative to sleep stages were reported.

Guilleminault et al. (1978) studied 50 predominantly male

(R age 48) sleep apnea syndrome patients. Apneas in this

population averaged 22 seconds in duration. Apneas were

longest in REM, and few occurred in Stages 3 or 4 sleep.

Fujita et al. (1981) studied 12 male SAS patients with a

mean age of 43. Apnea episodes averaged 23 seconds in

duration. Weitzman et al. (1980) studied 10 SAS males

whose age ranged from 38-57. The mean duration of apneas

in this group was 30 seconds. All apneas occurred in

light or REM sleep. Orr et al. (1979) studied 4 hyper-

somnolent males diagnosed with SAS. A control group of 4

nonhypersomnolent subjects was matched on number of ob-

structions per minute. Mean age was 57 for asymptomatic

and 42 for symptomatic subjects. Asymptomatic subjects

had apneas with a mean duration of 25.9 seconds, while

symptomatic subjects had a mean duration of 20.8 seconds.

Several characteristics emerge from the group of

reports on SAS subjects. There is wide variability in

duration of apneas, with reported means ranging from 16-30

seconds. The data available on placement of apneas in

sleep suggest that in SAS subjects most apneas occur in

light and REM sleep.

Comparison of data on normal and SAS subjects reveals

substantial overlap between mean duration of apneas in SAS

subjects (13-20 seconds) and SAS subjects (16-30

seconds). Particularly interesting was the Orr et al.

(1979) study which demonstrated a lower mean duration of

apneas for hypersomnolent subjects than for nonhypersomno-

lent subjects. Data on temporal placement of apneas show

most events occurring in light or REM sleep for both

normal and SAS subjects. Taken together, these data

suggest that substantial overlap exists on temporal

variables between normal and SAS subjects indicating a

limited utility for diagnostic purposes.

Variables Associated With Sleep Apnea
Syndrome and Activity

It was noted above that several variables have been

suggested as associated with increased incidences of sleep

apneas and SAS. These variables represent potential

"markers" of subpopulations which might experience higher

incidences of sleep apnea activity. These primary

associated variables include sex, age, snoring and

obesity. The data available on these variables and their

relationships to sleep apnea activity will be reviewed,

followed by a brief discussion of their relevance to

experimental methodology.

Regarding sex variables, as previously noted,

Guilleminault et al. (1978) found 96% of their SAS

patients to be males. In the Block et al. (1979) study of

nonsleep disturbed subjects, 40% of the males in the

sample had at least one episode of apnea, while only 15"

of the females so suffered. Bixler et al. (1982) noted a

trend toward more sleep apnea activity in males. Block et

al. (1980) studied 20 post menapausal females and found

apnea activity in 40% of them, leading them to suggest

that some factor associated with menapause (e.g., pro-

gesterone) might somehow "protect" premenapausal females

from apneas, and that this factor along with its benign

influence was lost with menapause. This explanation pre-

dicted the low level of apnea activity in younger females,

as well as the heightened apnea activity of aging

females. While the data of Carsakdon and Dement (1981b)

show aged females to be close to aged males in percentage

exhibiting sleep apnea activity, a report by Smallwood et

al. (1983) found no evidence of sleep apnea activity in

their 6 elderly (post menapausal) females, while elderly

males showed significant levels of apnea activity. The

limitations of the sample size in the Smallwood et al.

report might explain this discrepancy as representing a

sampling bias. Taken together, this evidence strongly

supports the notion that in young and middle aged popula-

tions, sleep apnea is a male phenomenon. Additionally,

evidence points toward a closing of the gap between

females and age matched males as the former group passes


A second factor which has been implicated in sleep

apnea activity is age. Webb (1974) found that his sub-

jects exhibiting sleep apnea activity were significantly

older than his full sample. This was also the case for

the subjects of Kales et al. (1982) and Ancoli-Israel et

al. (1982). Smallwood et al. (1983) found that sleep

apnea was an age dependent phenomenon with male subjects

over 50 exhibiting significantly more apneas than those

under 30. Other evidence includes that of Bixler et al.

(1982) who found a positive correlation between age and

apnea in their full sample, as did Block et al. (1980) in

their post menapausal females. It seems clear from these

data then that increasing age is a factor predisposing for

sleep apnea activity.

Snoring has been closely associated with sleep apnea

(Lugaresi et al., 1982). This may be due to a causal

relationship or a single common pathway. Lugaresi et al.

propose the latter explanation. Specifically, they

suggest that both snoring and SAS are due to a sleep

induced stenosis of the upper airway. With this premise

in mind, they conducted a survey of 1000 subjects on the

incidence of snoring. Chronic snorers included 31% of the

males and 19% of the females. In an elaboration of this

study, Lugaresi et al. (1980) questioned 5713 indivi-

duals. In this sample, 24% of the males and 13% of the

females were chronic snorers. Snoring increased with age,

and by 60 years of age 60% of the males and 40% of the

females snored. Of note was the observation that hyper-

tension occurred more frequently in snorers than non-

snorers, suggesting that the former was more at risk for

cardiovascular complications.

Virtually every study of SAS patients notes the pre-

sence of snoring in all these subjects (Sullivan and Issa,

1980; Block, 1980; Coverdale et al., 1980). This is

true for both predominantly obstructive, and predominately

central SAS subjects (Guilleminault et al., 1978). How-

ever, the presence of heavy snoring in asymptomatic

(nonhypersomnolent) apneas is also well documented (Orr et

al., 1979; Fisher et al., 1978; Berry and Block, 1983),

as well as in otherwise normal subjects. These data may

be taken as evidence for snoring as a necessary, but not

sufficient, condition for the occurrence of sleep apnea.

The role of body weight in SAS is less clear.

Recall that Pickwickian patients are invariably obese, as

well as hypercapnic. Guilleminault et al. (1978) stress

that SAS may occur in nonobese as well as obese sub-

jects. However, Block (1980) states that he has found SAS

subjects to be uniformly obese. A review of the

literature is in order to clarify this point. First

normal, then obese, then SAS groups will be reviewed.

Bixler et al. (1982) found that their subgroup of

subjects with sleep apnea activity weighed significantly

more than the remainder of the sample. Kales et al.

(1982) found that within their sample of 200 insomniacs

and 100 normals, a significant positive correlation

emerged between a weight:height ratio and apnea acti-

vity. Block et al. (1979) found a positive correlation

between apneas, oxygen desaturation, and weight within

their normal males, while females did not demonstrate this

relationship. Using an older female sample, Block et al.

(1980) found a positive correlation between weight and

oxygen desaturation, but not apnea. Thus in normal sub-

jects, there appears to be a positive relationship between

weight and apnea activity, although this may be restricted

to males.

A research approach which bears on this question

involves studying obese subjects alone. Two studies on

successive referrals for gastric bypass surgery were

noted. Sicklesteel et al. (1981) found that of 19 suc-

cessive referrals, all 14 males exhibited sleep apneas,

while only 5 of the females showed evidence of oxygen

desaturation alone. Harmon et al. (1981) studied gastric

bypass patients and found 5 of 6 males to suffer apnea, as

opposed to none of the females. This evidence provides

further evidence for a role of increasing weight in apnea.

The data on symptomatic SAS subjects are more com-

plex. Several studies (Sullivan and Issa, 1980; beitzman

et al., 1980; Garay et al., 1981; Zwillich et al., 1982)

report that all their SAS subjects were obese, although

definitions of obesity either vary widely or remain un-

specified. In contrast, Guilleminault et al. (1978)

reported that only 60% of their SAS sample was greater

than 15% above ideal weight. Coverdale et al. (1980)

indicate that only two of their subjects proved to be

obese, according to their definition (>125% of ideal

weight). These reports suggest that most, although not

all, SAS subjects are obese.

Several investigators have sought to compare weight

in symptomatic (somnolent) and asymptomatic subjects with

enough sleep apnea episodes to be diagnosed as SAS. Orr

et al. (1979) compared 4 symptomatic SAS subjects with

matched asymptomatic subjects. In this sample, symptoma-

tic SAS subjects were heavier than asymptomatic sub-

jects. Examination of the data of Berry and Block (1963)

indicates that their symptomatic subjects were heavier

than their asymptomatic subjects (118 kg vs. 91 kg).

Standing in contrast to these data is a study reported by

Fisher et al. (1978), which included 19 subjects referred

for SAS evaluation. Based on number of apneas per night,

Fisher et al. divided their subjects into 3 groups. Group

I had >100 apneas per night, Group II had >30 but <100

apneas per night, while Group III had less than 20 apneas

per night. Thus Groups I and II were diagnosable as SAS

(>30), while Group III was not. Obesity was present in

Groups I and II, but in only 3 subjects in Group III.

However, these differences were not statistically signifi-


Taken together, this evidence suggests a positive

relationship between weight and apnea in the general popu-

lation. However, SAS patients are not invariably obese.

Further, weight loss fails to result in a decrease in

apneas in many SAS patients. Thus the relationship

between obesity and full blown SAS is less than clear.

The preceding pages have investigated the currently

understood role of several variables (sex, age, snoring

and weight) in sleep apnea activity. Presently available

data suggest that in young and middle aged subjects, apnea

is a predominantly male phenomenon, although post mena-

pausal females may close the gap. Increasing age is re-

lated to sleep apnea activity. Snoring, which occurs in a

significant percentage of the general population, is

thought to be a necessary but not sufficient condition for

SAS. Weight was shown to have a positive relationship to

sleep apnea activity in the general population, although

its presence or role in SAS is equivocable.

These conclusions have important implications for

experimental designs directed at understanding sleep apnea

activity and its correlates in a subclinical population.

It is clear from the earlier review of apnea incidence

that sampling from the general population would result in

a relatively low population of individuals with sleep

apnea, an inefficient approach from the perspective of the

limited resources available for research. Thus selection

variables must be considered with one eye on efficiency

and the other on protecting the generalizability of

results. Selection criteria should also aim toward eluci-

dating the relative roles and possible interactions of

these variables in sleep apnea. Therefore, selection

variables should have as straight-forward a relationship

as possible with sleep apnea, seeking to answer more ques-

tions than they raise.

With these caveats in mind, the variables available

as selection criterion will be reviewed and conclusions

drawn about their relative merits. The first variable

considered, sex, has been relatively well investigated.

It is clear that sleep apnea activity is primarily a male

circumstance. Therefore, the selection of female subjects

is contraindicated from the standpoint of efficiency. Age

has been demonstrated to exhibit a positive relationship

with sleep apnea, indicating selection of older sub-

jects. Snoring is another variable with a close relation-

ship to sleep apnea. The apparently common mechanism of

snoring and sleep apnea (stenosis of the upper airway)

indicates that possibly very few subjects without snoring

suffer sleep apneas. A significant proportion of snorers

might be expected to suffer apneas, while few nonsnorers

might have apnea. The last variable, weight, while having

a positive correlation with apnea activity in the general

population, is of uncertain pertinence to SAS. Addi-

tionally, changes in weight may occur without changes in

apnea activity. Therefore, this variable has a less than

explicit connection to apnea activity and violates the

second principle adopted to consider these variables.

This review of available data suggests that sampling

aging, snoring, males would generate a high proportion of

sleep apnea activity. External validity for this idea is

drawn from its similarity to the "typical" SAS patient

described earlier. Extant data suggest that it may also

prove useful to utilize weight measures as a covariate, in

an attempt to control its apparent influence on apnea


Deficits Associated With Sleep Apnea Syndrome

Sleep Apnea Syndrome is thought to result in several

deficits. These include cardiopulmonary complications,

hypersomnolence, hypoxia and intellectual changes. Some

researchers have suggested that these deficits may be

present, in an attenuated form, in subclinical apnea

(Ancoli-Israel et al., 1981; Carskadon and Dement,

1981b). A discussion of the evidence for the presence of

these deficits will be presented, followed by proposals

for appropriate measures of these variables.

Cardiopulmonary Deficits

Cardiopulmonary complications have been reported as

concomittents of SAS (Guilleminault et al., 1978).

Coccagna et al. (1972) reported high incidences of hyper-

tension and congestive heart failure in their SAS sub-

jects. Schroeder et al. (1978) studied 22 SAS patients

with a mean age of 47. Six subjects had waking systemic

hypertension, and several had waking cardiac abnorma-

lities. During sleep studies, 20 patients developed

significant rises in systemic arterial pressure which

cycled with episodes of apnea, while 21 patients developed

pulmonary arterial hypertension. Tracheostomy reversed

all these abnormalities, strongly implicating the apneas

in their etiology. Tilkian et al. (1978) studied 25 male

SAS subjects with a mean age of 44. Twenty-four of the 25

showed sleeping sinus arrhythmias, while 9 developed more

serious symptoms such as asystole. Seventeen of these

patients received tracheostomy, which abolished all of

these abnormalities. Experimental occlusion of the

tracheostomy during sleep in 6 of these patients resulted

in a return of the cardiopulmonary symptoms. Zwillich et

al. (1982) studied six consecutive male SAS patients. In

all patients, bradycardia accompanied any apnea with

significant desaturation. Additionally, a significant

correlation was noted between degree of desaturation and

severity of bradycardia. Bradycardia was abolished or

attentuated during administration of 02 enriched air.

Zwillich et al. proposed that bradycardia during apnea

results from increased vagal tone mediated by carotid body

chemoreceptors. Fujita et al. (1981) reported on 12 male

SAS patients. All reportedly suffered cardiac arrhythmias

during sleep. The body of data suggests that SAS patients

frequently suffer from cardiac abnormalities such as

hypertension, arrhythmias, and bradycardia.

A second body of data exists which indicates that SAS

patients do not invariably have cardiopulmonary compli-

cations. Coverdale et al. (1980) studied 14 patients, of

whom 8 were diagnosed with SAS. Only 2 of these patients

had evidence of corpulmongle, while 2 more had systemic

hypertension. None of these patients demonstrated severe

brady or tachycardia during sleep. In Orr et al. (1979),

8 patients, all of whom were technically diagnosable as

having SAS, all symptomatic (somnolent) patients had evi-

dence of right sided heart failure as well as frequent

arrhythmias during apneas. Asymptomatic subjects had none

of these symptoms. Kreiss et al. (1982) studied 26

Veterans Administration ward patients and found 7 to have

a SAS. The SAS patients had significantly more signs of

congestive heart failure, but not hypertension or

angina. Ancoli-Israel et al. (1981) sampled 24 elderly

subjects with complaints of sleeping respiratory ois-

orders. Nine of these subjects were diagnosable as SAS,

but this group did not have significantly more heart

disease or hypertension. Thus a group of SAS patients

without evidence of heart disease exists.

Taken together, these data indicate that cardiac

complications frequently, but not inevitably, accompany a

sleep apnea syndrome. That a large number of apneas may

be present without a significant increase in cardio-

pulmonary complications is demonstrated by the Orr et al.

and Ancoli-Israel et al. studies. As the relationship

between these complications and a full blown SAS is

unclear, the possibility of cardiopulmonary sequelae from

a subclinical level of apneas is even more uncertain.

However, an assessment of cardiopulmonary status is

indicated by the frequent association of cardiopulmonary

complications with sleep apnea syndrome.

Arousal Deficits (Hypersomnolence)

Excessive Daytime Sleepiness (EDS), or hypersomno-

lence, has been described as the single most common result

of SAS (Dement et al., 1978). Indeed, patients who meet

the criterion of number of apnea episodes and exhibit EDL

are described as symptomatic, while those with number of

apneas alone are labeled asymptomatic (Orr et al.,

1979). As an aside, this distinction illustrates again a

central issue of SAS. If patients exhibit the number of

apneas qualifying them for SAS, but no other pathology,

are they suffering from a pathological process? Dement et

al. (1978) believe that they are. They suggest that most,

if not all, patients with a clinically elevated number of

apneas actually suffer LDS, but that it is masked by two

factors. Dement et al. noted that SAS patients may deny

somnolence while literally falling asleep before the

clinician. They speculate that this results from a re-

sponse bias against acknowledging illness or a change in

subjective frame of reference about just what alertness

is. Thus EDS may be denied by patients for reasons which

may be outside the clinician's control or knowledge.

Alternatively, Block et al. (1979) propose that asympto-

matic SAS patients are suffering a subclinical level of

fallout from their frequent apneas. As years pass, in-

creasing weight or cumulative effects of desaturation

eventually lead to a full blown hypersomnolence--Sleep

Apnea Syndrome. A final possibility is that nonhyper-

somnolent subjects with frequent apneas exhibit a com-

pletely benign process totally unrelated to that found in

symptomatic SAS patients. It would seem that delineation

of the level of EDS in subjects with subclinical levels of

apnea may help clarify this issue. This notion will be

elaborated below.

The somnolence deficit which is noted in full blown

SAS patients is commonly reported. Dement et al. (1978)

describe impairments in continuous performance of

virtually any activity. Sleep Apnea Syndrome patients

reportedly fall asleep at outdoor stadiums, in front of

classes, and while treating patients. Excessive Daytime

Somnolence is described in every SAS patient reported in

experimental protocols by Weitzman et al. (1980), Fujita

et al. (1981), Zwillich et al. (1982), Garay et al. (1981)

and Sullivan and Issa (1980). At the same time, subjects

wno are asymptomatic for EDS but with the necessary number

of apneas to be diagnosed as SAS are described by Orr et

al. (1979) and Smirne et al. (1980) as well as others.

However, it should be noted that the latter studies relied

on global reports of somnolence, rather than quantified

data. Recalling the criticism of Dement et al. (1978) of

this technique as vulnerable to subjective bias, these

asymptomatic patients may have unreported or undetected


Excessive Daytime Somnolence appears to be frequently

concomittant with Sleep Apnea Syndrome. Thus its possible

occurrence in subclinical populations seems worthy of


Oxygen Desaturation and Cognitive Deficits

One deficit which appears in SAS patients is oxygen

desaturation which accompanies apneas. When apnea begins,

arterial oxygen saturation begins to fall, and may con-

tinue falling throughout the event. An early technique

utilized in studying oxygen saturation levels involved

periodic blood samples drawn from arterial sources.

Birchfield et al. (1958) studied 11 normal males with a

mean age of 23. Daytime oxygen saturation in these sub-

jects averaged 96.5%. These values fell during sleep to

95.3%. Later, Orr et al. (1979) studied 8 males who had

been matched on number of apnea events. Asymptomatic

(nonhypersomnolent) subjects fell from a baseline 02

saturation of 80 mm (Hg) to a mean maximum desaturation of

54 mm (Hg) during sleep. Symptomatic subjects fell from a

baseline of 54 mm (Hg) to a mean maximum desaturation of

35 mm (Hg) while asleep. Orr et al. suggested that the

more severe desaturation which accompanied apneas in

symptomatic subjects underlaid their somnolence and other

complications. However, these conclusions are vulnerable

to certain methodological criticisms, as Block (1980)

points out. Specifically, periodic sampling may well fail

to detect the multiple brief desaturations associated with

SAS, rendering this technique potentially insensitive to

important events.

In an attempt to provide more representative data on

desaturation, Block et al. (1979) utilized an ear oximeter

in studying saturation in normal subjects. An ear oxi-

meter provided a continuous, accurate readout of moment to

moment arterial oxygen levels. Thirty males (mean age 38)

and 19 females (mean age 27) were studied. Of the 30

males, seventeen suffered at least one episode of desatu-

ration, falling from a mean saturation of 95% to a maximum

desaturation of 84% during sleep. In contrast, no epi-

sodes of desaturation were noted in the sample of premena-

pausal females (baseline 02 saturation: 96%). The oxygen

desaturations of the males were always found in asso-

ciation with breathing abnormalities or snoring. A stay

of oxygen saturation in 20 post menapausal females was

reported by Block et al. (1980). Eleven of these women

exhibited desaturation with several subjects desaturating

to less than 85%. Dolly and Block (1982) studied a group

of 17 males and 3 females with a mean age of 49 years.

These normal subjects dropped from a mean baseline satura-

tion of 96.3% to a mean maximum desaturation of 88.3%.

This evidence indicates that desaturations of 10-127 are

common in normal males and post menapausal females while

premenapausal females rarely, if ever, desaturate.

Oxygen saturation levels in SAS patients are not

commonly reported. Using an ear oximeter, Garay et al.

(1981) studied 11 SAS patients who were divided into 6

daytime eucapnics and 7 daytime hypercapnics. Eucapnics

fell from a baseline saturation of 95% to a mean maximum

desaturation of 75.5%, while hypercapnics fell from a

baseline saturation of 89% to a maximum desaturation of

62.8% (calculated from tabled data). These data were

collected during daytime "nap" studies, however, and hence

is somewhat questionable. Berry and Block (1983) recorded

9 heavy snoring males who suffered many apneac events.

Four subjects were somnolent, while 5 were not. Symptoma-

tic subjects fell from a baseline of 95% to a mean maximum

desaturation of 44%, while asymptomatic subjects desetu-

rated from a baseline of 94% to a mean maximum desatura-

tion of 85%.

Data on desaturation are interesting from several

perspectives. First of all, Sleep Apnea Syndrome subjects

seem to desaturate more heavily than normals, with Berry

and Block's symptomatic group desaturating up to 50% and

the subjects of Garay et al. desaturating about 22%,

compared with normal desaturation levels of 10-12%.

Secondly, amount of desaturation distinguished symptomatic

(hypersomnolent) from asymptomatic subjects in the data of

both Orr et al. and Berry and Block. Additionally, two

points relevant to methodology emerge. Measures of oxygen

desaturation seem crucial to understanding sleep apnea and

its effects, and ear oximetry appears to provide a proven,

valid measure of this variable.

Weitzman (1979) has suggested that the pathological

process in SAS is the hypoxia of desaturation in apneac

episodes. While cognitive and intellectual changes in SAS

have been only anecdotally reported (Guilleminault et al.,

1978), hypoxia does seem a plausible potential cause of

these changes. Thus oxygen desaturation appears to be a

reliable consequence of sleep apnea syndrome, while cogni-

tive/intellectual changes are putative sequelae.

In summary, a review of deficits found in sleep apnea

syndrome subjects suggests that cardiopulmonary complica-

tions, excessive daytime sleepiness, oxygen desaturation

and cognitive/intellectual changes are frequently asso-

ciated with this syndrome. An assessment of these

variables in a subclinical population seems indicated by

these data. Below a discussion of measurement issues and

proposal assessment of these variables appears.

Measurement of Deficits Found in Sleep Apnea Syndrome

Above it was noted that cardiopulmonary complica-

tions, excessive daytime sleepiness, nocturnal oxygen

desaturation and cognitive/intellectual changes are common

deficits of sleep apnea syndrome. A discussion of

measurement issues and proposed measurement devices for

these variables will be presented below.

Measurement of Cardiopulmonary and Health Deficits

Cardiopulmonary deficits are thought to be frequent

sequelae of sleep apnea syndrome. Thus an assessment of

blood pressure as well as self reports of hypertension and

heart trouble seems indicated in subclinical apnea sub-

jects. Additionally, a general survey of health status,

such as that provided by the Cornell Medical Incex (CiI;

Broadman et al., 1949) would also screen for other health

deficits. The CMI provides an overall score indicating

number of symptoms endorsed, as well as subscales on

several symptom categories. Separate examination of sub-

scales on respiratory, cardiopulmonary, and neurological

subscales seems appropriate. Assessment of these

variables provides a broad health screening, as well as a

detailed analysis of symptoms found in sleep apnea syn-


Measurement of Arousal Deficits (Hypersomnolence)

Excessive daytime sleepiness is a common sequel of

sleep apnea syndrome. Measurements of the sleep/wake

cycle of subclinical apnea patients thus seems indi-

cated. A sleep questionnaire assessing trait sleep habits

seems in order, as well as 1 week sleep logs following the

experimental night. The sleep questionnaire and sleep

logs have been used in ongoing research in W.B. Webb's

studies of aging and sleep and seem an adequate assessment

of sleep patterns of subclinical apnea subjects. Addi-

tionally, electroencephalographic recordings of sleep

during the experimental night seem in order using a stan-

dard recording and scoring system (Agnaw and Webb,

1972). The assessment of daytime sleepiness is relatively

new, only recently being the subject of investigation.

William Dement and associates began to develop methods of

quantifying somnolence in the early 1970s. The first

attempt produced a subjective rating scale, the Stanford

Sleepiness Scale (SSS; Hoddes et al., 1973). The SSS was

a 7 point Likert scale with endpoints anchored on 1) feel

active and vital; and 7) almost in reverie, sleep onset

soon; with 5 points with statements reflecting various

degress of somnolence in between. Patients rated their

introspective level of sleepiness on this scale.

Unfortunately, the SSS proved to be vulnerable to exactly

the same difficulties as a global rating scale used on SAS

patients, i.e., response bias and altered frame of

reference (Dement et al., 1978).

It was clear then that an objective measure of day-

time somnolence was needed. With this goal in mind,

Carskadon and Dement (1977) introduced the Multiple Sleep

Latency Test (MSLT). This technique involved measuring

the objective sleep latency during multiple nap attempts

throughout the day. In a typical paradigm, a subject

would be wired to an EEG and attempt to fall asleep at 2

hour intervals through the day. The latency to sleep in

each nap was measured by determining the latency to the

first epoch of Stage 1 sleep. In order to prevent signi-

ficant amounts of sleep accumulating during testing, sub-

jects were awakened after an epoch of sleep was

observed. If no sleep was noted, subjects were discon-

nected after 20 minutes and allowed to resume other acti-

vities. Carskadon and Dement believe that this measures

underlying physiological sleep tendency. This underlying

tendency is thought to be modulated by alerting stimuli in

the environment (Carskadon and Dement, 1982a). The vali-

dity of the test was established on normal subjects, who

exhibited a biphasic curve of sleep tendency throughout a

24 hour period. Extension of sleep over typical lengths

in normals (who are thought to be moderately chronically

sleep deprived) saw a significant rise in average sleep

latencies on the MSLT (Carskadon and Dement, 1979b), while

chronic sleep restriction (5 hrs sleep per night) caused

significant declines in average sleep latencies (Carskadon

and Dement, 1981a). Finally, total sleep deprivation led

to a drastic decline in MSLT values, with averages

plummeting below one minute (Carskadon and Dement,

1979a). Mean MSLT values showed recovery to baseline

values after a full night's sleep for partially (5 hrs)

deprived subjects, and recovery after two nights of ad lib

sleep for totally sleep deprived subjects (Carskadon and

Dement, 1979; 1981a). This body of evidence suggests

that the MSLT is a reliable and valid index of sleepiness

in normal subjects.

With its internal validity established, the MSLT was

applied to somnolent populations. Dement et al. (1978) as

well as Richardson et al. (1978) demonstrated that SAS and

narcoleptic patients have significantly shorter latencies

to sleep on the MSLT than normal subjects do. Hartse et

al. (1979) found SAS subjects to have an average latency

to Stage 1 of 2.5 minutes, while narcoleptics had a mean

of 3.2 minutes. Both these latencies were significantly

lower than those of a miscellaneous control group whose

latencies were about 11.2 minutes. Hartse et al. (1980)

found latencies to Stage 1 sleep as follows for various

patient groups: Normals 12.2 m, Insomniacs 16.2 m, SAS

2.5 m, and Narcoleptics 3.2 m. The SAS and narcoleptic

patients had significantly shorter latencies to Stage 1

sleep than insomniacs or normals. Roth et al. (1980)

compared 10 SAS patients with 10 age matched controls.

The normals averaged 20.4 m to Stage 1 sleep while the SAS

patients averaged 1.9 m. Zorrick et al. (1982) found mean

MSLT latencies to Stage 1 sleep of 3.1 m for SAS subjects

and 2.9 m for narcoleptics, significantly shorter than for

patients with psychiatric disorders.

The chief difficulty remaining in judging the MSLT

"reliable and valid" is external validation demonstrating

that exceeding a certain mean latency is correlated with

other deficits of performance. In some ways, low laten-

cies on the MSLT may suffer the same limitations as sleep

deprivation protocols; that is, demonstrating serious

deficits is surprisingly difficult. At the same time, few

would deny the subjective unpleasantness of sleep depriva-

tion, which may be applicable to low MSLT scores. Thus

the MSLT appears indicated in assessing excessive daytime

sleepiness while the Stanford Sleepiness Scale (SSS) pro-

vides an index of the subjective experience of the sub-

ject. Application of both these measures to a subclinical

apnea population would seem to adequately assess daytime

sleepiness in these subjects.

Measurement of Hypoxia and Its Sequelae

Noctural desaturation is frequently noted in SAS

patients. Therefore, it seems in order to examine the

known sequelae of hypoxia of other etiologies in an

attempt to determine variables or behaviors likely to be

sensitive to the hypoxia found in SAS and SAA. Research

on hypoxia varies systematically along two dimensions,

human vs. animal subjects and acute vs. chronic hypoxia.

First, the data available on animals will be reviewed,

followed by that extant on human subjects. Finally,

behavioral variables sensitive to changes in these

variables will be summarized.

The effects of chronic hypoxia on animals have not

been exhaustively studied (Davis, 1975). One project

which did examine this variable involved exposure of rats

to up to 36 hours of 10% oxygen. Davis (1975) observed

the behavior of these animals and concluded that they

appeared normal. Although hyperventilation was noted, no

permanent effects of the hypoxia were found. Hanbauer et

al. (1981) examined the effects of 10% 02 on rats exposed

for up to 4 weeks. They found that relatively short term

(2 days) hypoxia resulted in an increase in dopamine (DA)

content in the carotid body. An increase in norepine-

phrine (NE) content was noted after 1 week of exposure.

As the carotid bodies are implicated in the control of

respiration, the authors concluded that different

mechanisms are operative in the adaptation to short and

long term hypoxia. Unfortunately, little other data have

been reported on the chronic effects of hypoxia on more

complex functioning. This is particularly problematic as

it are these functions which are thought to be affected in

humans exposed to hypoxia.

Certain aspects of acute hypoxia in rats have been

carefully examined. Gibson et al. (1981) reviewed the

evidence which has been collected on neurotransmitter

alterations in hypoxia. In brief, mild to moderate

hypoxia, cerebral levels of ATP are normal, implying that

energy supplies of neurons are not disturbed. In con-

trast, the turnover of certain neurotransmitters is

altered. Gibson et al. note that oxygen is an important

substrate for the synthesis of several neurotrans-

mitters. When rats are exposed to acute hypoxia, syn-

thesis of dopamine and serotonin is decreased, while

absolute levels of tryosine, tryptophan, catecholamines,

and serotonin remain constant. Continuing hypoxia leads

to an adaptation resulting in normal turnover of these

neurotransmitters. Gibson and Blass (1976) examined the

effect of acute hypoxia on central acetylcholine (ACh).

As in dopamine (DA) and serotonin (5-HT), acute hypoxia

results in a decrease in synthesis in ACh, but not in the

absolute level of ACh. Thus there is a substantial body

of evidence indicating changes in central neurotransmitter

levels in hypoxia. Generalizing from the animal data to

human subjects, Gibson and Duffy (1981) suggest that these

changes underlie the behavioral deficits of hypoxia in

humans (which will be reviewed below).

Other variables have been examined in acute hypoxia

in animals. Annau (1972) found that rats exposed to brief

hypoxia exhibited decreased appetite, thirst, and self

stimulation rates. Annau observed that brief hypoxia

depressed all behavioral variables. Gellhorn (1951) found

that brief hypoxia of 4-7.5% 02 abolished cortical re-

sponses to auditory stimulation. Sara (1974) trained rats

in active one way avoidance tasks. Immediately after

training, subjects were exposed to 3.5-4.0% oxygen. Rats

tested at 1 and 3 hours post hypoxia avoided correctly,

while those tested at 6 and 24 hours did not, leading Sara

to suggest that hypoxia resulted in a memory retrieval

deficit. Freides and Allweise (1978) also trained rats in

one way avoidance and exposed them to 2.0% hypoxia imme-

diately following training. These animals avoided at 12

and 200 minutes post hypoxia, but not at 90 minutes.

Friedes and Allweise suggested that a sensitive period for

hypoxia effects on training existed for 15 minutes after

learning. Tauber and Allweise (1975) followed the same

training/hypoxia paradigm and found avoidance impaired at

2.5 but not 4.5 hours. They suggest that hypoxia inter-

feres with a medium term memory. While some of these data

are conflicting, it appears that acute hypoxia depresses

certain behavioral variables and interferes with aspects

of memory processes.

While caution is in order when generalizing from

animals to humans, several seemingly relevant points

emerge from the review. Behaviorally, simple activities

may be spared, although some depression of appetitive

activities may occur, while more complex behavior depen-

dent on memory processes may be altered. Of particular

interest are the demonstrated alterations in

neurotransmitter turnover seen in hypoxia. It has been

suggested that these subtle alterations may underlie the

changes in complex behavior seen in hypoxic humans.

Data on human performance after hypoxia have neces-

sarily been reliant on fortuitous or weak manipulations.

Although this may limit the usefulness of these data, it

is at the very least highly suggestive of variables sensi-

tive to hypoxia. Van Liere and Stickney (1963) note that

acute hypoxia leads to confusion, headache, drowsiness,

weakness and incoordination. After effects include head-

ache, nausea and emotional liability. Richardson et al.

(1959) found that acute, fatal hypoxia from surgery

results in damage to cortical neurons, particularly in

layers III and IV. Susceptible subcortical neurons in-

clude those of the corpus striatum and cerebellum. Plum

et al. (1962) describe a syndrome of delayed postanoxic

encephalopathy. After the acute coma resolves in 4-5

days, a brief period of normal functioning returns.

Increasing irritability and confusion then emerge, with

loss of coordination and memory, and diminished attention

span. Typically, the lesion spares grey matter while

affecting white matter. Devereaux and Partnow (1975)

describe a patient who recovered from a delayed acute

encephalopathy. Although IQ was relatively spared, severe

dysarthria emerged. Relevant points which emerge from

studies of acute hypoxia in humans suggest that sublethal

levels of hypoxia affect coordination, memory, attention

span, and may result in headache, drowsiness, and nausea.

The effects of milder, chronic hypoxia are summarized

by Gibson et al. (1981). At the oxygen pressure equiva-

lent of 5,000 feet, impaired dark adaptation is noted. At

10,000 feet decreased concentration, hyperventilation, and

short term memory deficits are noted. By 15,000 feet,

euphoria and loss of both coordination and critical judge-

ment are observed. Christenson et al. (1977) found signi-

ficantly less visual signals detected at 17% 02 than at

21,%, suggesting reduced vigilance. McFarland (1937)

studied men in ascents to high altitudes and concluded

that subjects with slower rates of ascent performed better

on unspecified psychological tests. West (1984) reports

data collected on an expedition to Mt. Everest. Results

of a neuropsychological battery indicated deficits in

finger tapping, verbal fluency, verbal learning, short

term memory, and expressive language. Thus the general

affects of chronic mild hypoxia are thought to be deficits

in dark adaptation, short term memory, critical judgement,

motor coordination, verbal fluency, and vigilance.

One group of chronically hypoxia patients, those with

chronic obstructive lung disease (COLD), has been rela-

tively well studied with sensitive neuropsychological

instruments and is deserving of a more painstaking

review. Krop et al. (1973) studied COLD patients before

and after they received oxygen supplements. The WAIS,

Wechsler Memory Scale (WKS), Bender-Gestalt, Bender

Gestalt Interference Procedure, Facial Recognition and

Finger Tapping tests were administered to COLD and control

subjects. Subjects receiving supplementation improved

significantly more than controls on WAIS Full Scale IQ,

Performance IQ, Wechsler Memory Quotient, both Bender

procedures and Finger Tapping. Grant et al. (1982)

studied another group of COLD patients. They administered

the Halstead-Reitan Battery, Aphasia Screening Test,

Trailmaking Test, Russell modification of the Wechsler

Memory Scale, Tactile memory test, and the WAIS. In COLD

subjects, deficits were found relative to controls on all

variables except Reitan Rhythm, Aphasia Screening and WNS

Logical stories. Grant et al. concluded that COLD

patients exhibited deficits on a global impairment rating,

attention, abstracting ability, complex perceptual motor

tasks, simple sensory and motor tasks, and memory tasks.

Low but significant negative correlations were reported

between several measures of oxygen saturation and both

Halstead Impairment Index and Global Impairment Rating.

Grant et al. speculated that floor effects resulted in low

variability in PA02 levels among COLD patients and kept

these correlations low. They further suggested that the

neuropsychological deficits observed in these patients

resulted from the oxygen want found in these subjects.

Thus data available on humans indicate that dark

vision, memory, motor coordination, critical judgement,

verbal fluency, and signal detection are affected by acute

hypoxia. Animal data suggest that sensitivity of memory

processes to hypoxia, while data from chronic hypoxia

humans indicate that attention, vigilance, abstracting

ability, complex perceptual motor skills, as well as motor

and sensory abilities are also impaired by hypoxia. While

none of these findings are unequivocally directly

applicable to the brief, acute, desaturation occurring

during apnea, they are, at the very least, indicative of

abilities sensitive to hypoxia in humans.

A neuropsychological battery sensitive to these

abilities includes Wechsler Adult Intelligence Scale

(WAIS; Wechsler, 1955); Wechsler Memory Scale (WMS;

Wechsler, 1945) including the Russell delayed memory

aspects (Russell, 1975); Rey-Osterreith Complex Figure

(Osterreith, 1944); Hooper Visual Organization Test

(Hooper, 1958); Verbal Fluency Test (Halstead, 1947);

Wisconsin Card Sort (Berg, 1948); and the Finger Tapping

Test (Halstead, 1947). Administration of this test

battery provides a broad screening of neuropsychological

functioning, with a particular emphasis on those abilities

known to be sensitive to hypoxia. Application of this

battery to subclinical apnea subjects provides a sensitive

indicator of possible hypoxic brain dysfunction.

In summary, alterations of sleep/wake cycles appear

to be present in sleep apnea syndrome patients. Thus

assessment of these variables in a subclinical population

is indicated. An adequate description of the sleep/wake

cycle would seem to include sleep questionnaire, sleep

log, overnight EEG, multiple sleep latency test, as well

as several Stanford Sleepiness Scale ratings.

Statement of the Problem

Sleep Apnea Syndrome is a clinical entity with

seemingly serious consequences. Although the nocturnal

apneas found in this illness are thought to be the cause

of the associated deficits, the identification of

seemingly asymptomatic subjects with a number of apneas

comparable to those found in SAS throws doubt onto this

formulation. Despite this evidence, the recent discovery

tnat apneas occur in significant numbers in the general

population has led to speculation that this subclinical

level of apnea may be the cause of various pathological

processes. At present, there is little evidence to

support or refute this notion.

The present report evaluates this issue through a

detailed study of a group of subjects thought to be at

risk for sleep apneas. Aging, heavy snoring males were

recruited, and a comprehensive analysis of their respira-

tory, health, neuropsychological, and sleep/wake cycle

status was carried out. Thus potential relationships

between respiratory disturbances and variables known to be

disrupted in sleep apnea syndrome patients could be

evaluated. This approach also allowed exploration of the


possible risk factors associated with subclinical sleep




Heavy snoring males were recruited via newspaper

advertisements, notices, and phone calls to a list of

aging subjects reported by Webb (1982). Participants were

required to be male, self reported heavy snorers, over 30

years of age, and in general good health (i.e., self

described healthy and normal and not under the active care

of a physician for illness). Subjects with a history of

head trauma or alcoholism were excluded. Potential

subjects who met these criteria were offered a chance to

participate in a "snoring study" in which they would

complete testing during an evening, sleep overnight in a

lab with several physiological parameters recorded, and

complete one week's worth of "sleep logs," for a payment

of fifty dollars. Volunteers were scheduled on a first

come, first served basis. A total of 60 subjects were

studied, with 9 cropped because of technical problems with

their overnight recordings and 5 excluded because of

health problems. Thus 46 subjects represented the final

sample. These 46 subjects had a mean age of 49.9 (sd

13.1) and represented a wide cross section of the

population, ranging from the unemployed to university

professors. All subjects completed and signed an informed

consent agreement.


During the evening testing, all subjects received a

blood pressure check with a standard hospital cuff. The

measurement was made from a seated position. Overnight

recordings were made in a quiet, darkened room with the

subject sleeping on a standard hospital bed. All noc-

turnal physiological recordings were routed through a

Grass Polygraph (model 7D). Electrodes for electro-

encephalographic recordings were mounted in three pairs

using sites F2/F8, PZ/T6, RE/LE from the international

10/20 system of placement. These electrodes were affixed

to the scalp with collodion soaked gauze pads dried with

an air hose. Respiratory measures included measurement of

chest and abdomen wall movement derived from impedence

measures monitored from surface electrodes (3M Ag/AgC1

#2246) placed on the lower chest and just above the

navel. The signal from these electrodes was passed

through an impedence convertor and onto the polygraph.

Oral and nasal airflow was recorded by thumistors (TCT-1R

Transducer-Grass) clipped to the respective orifices.

These signals were also routed through the polygraph.

Electrocardiograms were monitored from modified bipolar

chest leads (MCL2). Finally, blood oxygen saturation was

continuously monitored from a probe attached to the ear

lobe, and analyzed by a Biox ear oximeter (Ilodel IIA BTA

Co). All physiological variables were recorded on the

polygraph's chart paper which was run at a speed of 10



Two classes of independent variables are identified,

with the first being thumistor variables (those derived

primarily from the nose and mouth thumistors). These

included number of apneas and hypopneas, mean duration of

apneas and hypopneas, mean low oxygen saturation in apneas

and hypopneas, mean oxygen saturation change in apneas and

hypopneas, and total seconds spent in apneas and hypo-

pneas. Thumistor variables were separately scored by two

trained raters, whose agreement appears in Table 2-1. An

overall agreement of 85% was achieved. Disagreements were

resolved by consensus scoring. The second class of

independent variables included oxygen saturation variables

(derived from a minute by minute rating of highest and

lowest oxygen saturation from the overnight record).

These variables included mean highest saturation, mean

lowest saturation, and number of desaturations of 4 ana

10% or more. In selecting these scoring criteria, several

considerations were relevant. Initially, a system with

reasonable interrater reliability was necessary. While

published criteria for scoring apneas were not problem-

atic, reliable scoring of hypopneas was not achieved until

a desaturation criterion of 10% was introduced. In

selecting measures of oxygen saturation, a mean high and

Table 2-1. Interrater agreement for respiratory events of
46 snoring males.

Consensus Initial Disagreements
Scoring Agreements #1 #2

(178) -
91 90 4 6
78 76 1 1
61 52 3 8
57 56 9 2
32 29 3 3
28 27 0 2
25 25 0 0
19 16 4 0
17 17 0 0
15 16 1 0
13 13 0 0
7 7 0 0
S44 0 0
4 4 0 0
3 2 0 1
3 3 0 0
3 0 7 2
3 3 0 0
3 3 0 0
1 1 0 0
1 1 0 0
1 1 0 0
1 1 0 1
1 0 0 1
1 1 0 0
1 1 0 0
1 1 0 0

474(652) 450


Scored by

added 1
added 2
added 9
added 1
added 3
added 1

Consensus added 3

Consensus added 1

Consensus added 1

Consensus added 3

Consensus added 1

37 56

Note: 28 of

the 46 subjects experienced at least 1 apnea or

low score seemed obvious. In order to assess the quan-

titative "impact" of numerous desaturations, a sum of 4%

and 10% desaturation was tabulated. These selected

parameters for thumistor and saturation variables allowed

reliable scoring and a description of several aspects of

nocturnal respiration.

Dependent measures fell into three broad classes:

measures of sleep/wake status, measures of health status,

and measures of neuropsychological status. Measures of

the sleep/wake cycle included one week's sleep logs

(derived variables included number of reports of daytime

sleepiness, hours asleep, estimated sleep latency, number

of wakenings after sleep onset, and minutes of waking

after sleep onset), a sleep questionnaire (derived

variables including hours of sleep usually obtained,

number of naps, hours napping, quality of sleep,

restfulness on awakening, depth of sleep, and amount of

daytime sleepiness), the mean of the several Stanford

Sleepiness Scale ratings, the two multiple sleep latency

tests (derived variables included mean sleep latency and

total minutes asleep in both naps) and the overnight LEG

variables derived from the scoring system of Agnew and

Webb (1972). Electroencephalographic variables included

time in bed; pure sleep time; sleep efficiency index;

number stage 0 periods; time 1 stage 0; sleep latency;

time % stages 1, 2, 3, 4 and REM; latency first hLM

period; mean REM period; mean REM cycle length; $ slow

wave sleep; and number of stage changes.

Measures of health status included height, weight,

diastolic and systolic blood pressure, as well as the

Cornell Medical Index. From the CMI, reports of hyper-

tension and heart trouble were noted, as well as the total

number of symptoms endorsed. Additionally, the scores for

the scales tapping respiratory, cardiac, and neurological

symptoms were tabulated separately.

Measures of neuropsychological status included the

WAIS Verbal and Performance IQ's, the WMS Memory Quotient,

delayed recall of the logical stories and the visual re-

production subtests of the WMS, delayed recall of the Rey

complex figure, number of words generated in the verbal

fluency test, number of correct sorts of the Wisconsin

Card Sort, mean number of taps for the right and left

handed finger tapping tests, and lastly the number of

correct trials on the Hooper test.

These neuropsychological tests were chosen with two

aims. First, an attempt was made to sample from each of

several loosely bounded areas of cognitive skill. These

included intelligence (both verbal and non-verbal),

memory, both immediate and delayed (verbal and non-

verbal), visuo-perceptual/organizational skills, language,

and frontal self-regulatory skills. A second emphasis was

on a relatively detailed analysis of areas previously

shown to be sensitive to hypoxia. These sensitive areas

included memory, visuo-organizational skills, verbal

fluency, and motor coordination. The chosen neuropsycho-

logical battery reflected these two aims. Thus the WAIS

tapped intellectual functioning; the Wechsler Memory

Scale with delayed recall of logical stories and visual

reproductions as well as digit span and Rey Figure heavily

sampled verbal and non-verbal memory; the Hooper tapped

visuo-organizational skills; Verbal fluency sampled lan-

guage skills; the Wisconsin card sort tapped frontal

self-regulatory skill; and Finger tapping sampled motor

functions. The chosen battery provided a broad screening

as well as a detailed assessment of neuropsychological

functions thought to be impaired by hypoxia.


Subjects followed the schedule appearing below for

the experimental night:

1e00 Sign informed consent, blood pressure,

height/weight, finger tapping, verbal fluency,

subjective sleepiness (SSS-made every 1/2 hour

till bedtime)

1830 Wiring for EEG

1900 Begin MSLT I (20 m)

1930 WAIS

2000 WMS, Hooper

2030 Graphesthesia, Luria motor programs

2100 Begin MSLT II (20 m)

2130 Rey figure, Wisconsin card sort

2200 Sleep Questionnaire, Cornell Medical Index,

remaining testing

2230 Wiring for thumistor, EKG, respiratory effort,

ear oximeter

2300 Lights out, begin recording

A night technician remained with the subjects throughout

the night. Subjects usually awakened between 5 and 6 am,

and later completed the one week's sleep logs.

Statistical Procedures

Several statistical procedures were utilized in

evaluating the present data. One of the primary

approaches was correlational, considered appropriate

because of the exploratory nature of the study. Bearing

in mind the possible alpha inflation of large corre-

lational matrices, other approaches were also utilized to

add confidence. Thus the sample was stratified by level

of apnea/hypopnea, and oneway ANOVA procedures were used

to evaluate between group differences. Additionally, non-

parametric procedures were utilized where appropriate.

Positive correlational results were considered against

possible replication with other measures in the study, and

in some cases multivariate procedures were addea for

further confidence. It is conceded that correlational

approaches may carry the burden of possible spurious

results, but it is believed that the various additional

procedures outline above lend confidence to the findings.


Demographics and Incidence of
Nocturnal Respiratory Disorder

Subjects were selected to fulfill several sampling

requirements, including male sex, presence of heavy

snoring, and self report of general good health. A total

of 60 subjects were studied. Five subjects were later

excluded when questioning revealed the presence of chronic

alcoholism (2) and serious head trauma (3). Additionally

9 overnight records were rejected because of missing or

unusable data. The characteristics of the final 46

subjects are detailed in Table 3-1. These snoring males

had a mean age of approximately 50 years, a mean weight of

approximately 190 pounds, and a mean educational level of

15 years of schooling.

Table 3-1. Demographic variables from 46 snoring males.

K SD Range

Age 49.9 13.1 30-75
height (Ibs) 189.3 34.9 125-250
Weight:Height Ratio (lbs:inches) 2.7 0.5 1.9-3.6
Eaucation (years) 14.7 2.5 9-20

The respiratory records were annotated by two hour

intervals following bedtime. From a total of 652 events

scored, 150 (23%) occurred in the first two hours

following bedtime, 185 (23%) occurred in the second two

hours, 232 (36%) occurred in the third two hours, while 83

(12%) occurred in the final two hours. It should be noted

that wakeup times varied considerably, making interpre-

tation of the attenuated number of events in the final two

hours somewhat problematic.

From a total of 578 events in which reliable sleep

staging was achieved, 459 (78%) occurred in light slow-

wave sleep (stages 1-2), while 118 (20%) began in REi

sleep. Only 6 events (01%) occurred in slow wave sleep.

These precentages must be considered in light of the

relative distribution of sleep stages in these subjects:

stages 1-2 (64%), stage REM (12%), stages 3-4 (10%), stage

0 (14%). This distribution suggests that the sleep time

spent in stages, 1, 2, or REM must have provided the

overwhelming preponderance of events in these stages.

Sleep characteristics will be discussed in greater detail


As noted below, 28 of the 46 subjects experienced at

least one episode of apnea or hypopnea. Table 3-2

presents the data on occurrence of at least one

apnea/hypopnea, as well as the occurrence of a high level

of apnea/hypopnea. Almost 2/3 of the sample had at least

one apnea or hypopnea, while 13% had high levels of

Table 3-2. Incidence of low and high levels of
apnea/hypopnea by age in 46 snoring males.

Low Apnea/ High Apnea/
Age Group N Hypopnea Hypopnea

30-39 14 .71 0
40-49 7 .57 .29
50-59 11 .72 .27
60+ 14 .43 .07
Overall 46 .62 .13

apnea/hypopnea. Note that subjects with high levels of

apnea/hypopnea are included in the overall total of

subjects with at least one event for the following

analyses. The classification of "high" levels of

apnea/hypopnea utilizes the criterion of an apnea +

hypopnea index greater than 5, a frequently used clinical

cut-off score. The table also presents the frequency of

respiratory distress broken down by age. No simple age

related trend is apparent, and a X2 analysis of the two

factors of age and respiratory distress was not signi-

ficant for those with at least one event (X2=3.2; p=NS),

or for those with high levels of apnea/hypopnea (X2=5.9;


A further investigation of possible age by level of

apnea activity involved division of the sample into those

above ano below 60 years of age crossed with those above

and below 5 events/hour. The X2 value for this comparison

was also nonsignificant (X2=0.6; p=NS). Subjects were

also divided into those above ano below the mean weight:

height ratio (2.7) and crossed with those with and without

at least one event or high levels of apnea/hypopnea. Both

these non-parametric tests were also non-significant

(X2=.01; p=lS; X2=1.5; p=HS).

lablt 5-3 utilizes Pearson correlations to explore

possible relationships between demographic variables (age,

weight, weight:height ratio) and nocturnal respiratory

indices (apnea index, apnea + hypopnea index, seconds in

events, mean hiFh saturation, mean low saturation, number

of desaturations >_4, and number of desaturations >10Q).

Again, age displays little relationship with respiratory

indices, but substantial correlations between weight and

various respiratory indices are apparent.

Table 3-3. Significant (p<.05) Pearson correlations between
nocturnal respiratory variables ana demographic
variables in 46 snoring males.

Respiratory Weight:Height
Variable Ape Weieht Ratio

Number of Apneas
Lumber of Apneas+Hypopneas .265
Apnea Index
Apnea+hypopnea Index .260 .506*
Mean High Saturation -.323* -.363*
Mean Low Saturation -.397* -.460*
Mean Saturation Change .319* .373*
Number of Desaturations >4 .267 .368* .450*
lumber of Desaturations >10


In summary, non-parametric analyses failed to reveal

significant relationships between age or weight and

overnight respiration. However, Pearson correlations

suggested that increasing weight is associated with

deterioration of several noctural respiratory indices.

Because increasing apnea/hypopnea levels have been

suggested to carry increasing risks for various deficits,

the sample was stratified by level of apnea/hypopnea. The

first group included those subjects without events (No

events), those with at least one event but less than 5

events per hour (Low apnea/hypopnea), and those with 5 or

more events per hour (High apnea/hypopnea). It should be

noted that this division of subjects is slightly different

from the earlier non-parametric analyses, because in this

grouping low apnea/hypopnea and high apnea/hypopnea are

mutually exclusive, while in the earlier analyses the

group with at least one apnea/hypopnea included those with

more than 5 events per hour.

Table 3-4 utilizes oneway ANOVA procedures to compare

these groups on demographic and nocturnal respiratory

variables. There were no significant between group

differences on any of the demographic factors: age,

weight, weight:height ratio, and education [F(2,43)=2.1,

0.6,0.8,0.7; p=lIS]. Logically, both apnea and hypopnea

indices were significantly different between groups

[F(2,43)=9.6, p<.CO04; 20.5, p<.0000]. Followup

comparisons (Scheffe) indicated that the high


0 (U CU a N n m

v00 i- 0 -- .
O <0 Z NX

CO 0 CL ) o OaN

0. ..

C O*
o >> < Qr co O


O "
o 0 m IN- CU .~

aC O O ..
OU 0 JO m C


C UC** O**

SH 0)
SE -- 3 m o T- NOC~ro O^^

3J O N
Co ( I C J O-(1J'30 0.U

0 0 0L00 0 c\ 0 r
-> Q CO N 00

C 0 0

CU La a
Dc t co It
Ei > a fu cu O -H l

CU 0 Cd

v 0 L 0 I0. 0 0 0 HU
0 c >Q. H -(

9, Vm "0 u 0 cu 0.'

00. CU C w C0 OU Z o o3
Ili>10 C: CUQu )m L. (U CUE-P
,u Q) a co w w m 0 M 4J-

0. a X 4 C 00 00c
*u-H 00D0. UP)
a)o 0 z Qa o o 7H

w Q -1.0'*-H-aOO Q -41 4-j
00 c0 > a 0 OO -H-H
*H 'a CL 0 O CCQ
*4 -L +*C to (0
c a'o W0W c U C ru)C C c I

< 00 ) 400 0 C: oo o t o oo C, r

apnea/hypopnea group was significantly different from the

other two groups on these two indices. While mean

duration of apneas and hypopneas was not significantly

different between the high and low apnea groups

[F(1,25)=3.5, 0.1; p=NS], total seconds in respiratory

events was [F(2,43)=50.4, p<.0000], with followup

comparisons showing that the high apnea/hypopnea group was

significantly different from tne other two groups.

Turning to the oxygen saturation data, mean high

saturation failed to reach significance [F(2,43)=0.3,

p=NS]. However, mean low saturation, number of

desaturations >4% and >10% were all significant

[F(2,43)=11.6, p<.0001; 7.0, p<.C02; 21.4, p<.0000], and

followup comparisons showed that the group with high

levels of apnea/hypopnea was significantly different from

the other two groups.

Thus, division of the sample into levels of

apnea/hypopnea resulted in groups which were not

significantly different on demographic variables including

age, weight, weight:height ratio, and education. However,

the group with high levels of apnea/hypopnea had

significantly more nocturnal events, time in events, and

oxygen desaturation.

In summary, the present sample of 46 snoring males

appears representative of those predisposed to experience

sleep disordered breathing (middle-aged, snoring males).

Sixty-two per cent of these subjects had at least one

apnea or hypopnea, while 135 had high levels of

apnea/hypopnea. Most events occurred in light slow-wave

or REM sleep. No effect of age on level of sleep

disordered breathing was observed, although increasing

weight was correlated with nocturnal respiratory events.

When the sample was stratified by level of apnea/hypopnea

activity, subjects with high levels of events had

significantly more events and desaturations than the

remaining subjects, although no significant differences

were observed on age, weight or education.

Nocturnal Respiratory and Health Variables

Health variables including blood pressure readings

and several scores derived from the Cornell Medical Index

(overall number of symptoms endorsed, as well as number of

symptoms endorsed from the respiratory, cardiopulmonary

and neurological subscales) were evaluated initially

through a matrix of Pearson correlations which included

the respiratory variables (apnea index, apnea + hypopnea

index, seconds in events, as well as mean high and low

saturation and number of desaturations >4% and >10%).

From a total of 42 correlations, three were significant at

p<.05 or less: Systolic blood pressure and mean low

saturation (r=-.261); Systolic blood pressure and number

of desaturations >4% (r=.373); Overall number of symptoms

endorsed on the CMI and mean high saturation (r=-.254).

As a further exploration of possible relationships between

blood pressure and apnea/hypopnea activity, subjects were

asked whether they had ever been diagnosed with hyper-

tension. Thirty-three per cent of those with no events

had diagnosed hypertension, as did 33% of those with low

levels of apnea/hypopnea, while only 15% of those with

high levels of apnea/hypopnea had diagnosed hyperten-

sion. These data do not support an increasing level of

hypertension in subjects with high levels of


The subjects were grouped into the earlier delineated

levels of apnea/hypopnea. Table 3-5 presents the means

and standard deviations of health variables from the three

groups. No dramatic trends are apparent in this table,

and oneway ANOVA procedures confirmed this observation

with no significant differences for: Diastolic BP,

F(2,43)=.01, p=NS; Systolic BP, F(2,43)=.1, p=NS; CMI

overall, F(2,42)=.2, p=liS: CMI respiratory, F(2,42)=.5,

p=NS; CMI cardiopulmonary, F(2,42)=.5, p=NS; CMI

neurological, F(2,42)=2.2, p=NS.

In summary, while correlational procedures suggested

relationships between several oxygen saturation measures

and blood pressure as well as overall number of symptoms

endorsed on the CMI, oneway ANOVAs failed to reveal

significant between group differences when subjects were

stratified by level of nocturnal distress.

Table 3-5. Means and standard deviations for health related
variables in 46 snoring males grouped by level
of apnea/hypopnea.

No Apnea/ Low Apnea/ High Apnea/
Hypopnea Hypopnea Hypopnea

Systolic BPa 129.1(14.2) 128.7(12.6) 127.0(7.5)
Diagtolic BP 83.2(13.4) 83.9(11.0) 81.6(7.6)
CMIU Overall Score 14.7 (7.4) 16.1(10.5) 13.8(7.4)
CMI Respiratory Score 1.4 (1.2) 1.1 (1.8) 2.0(1.4)
CMI Cardiac Score 1.6 (1.3) 1.2 (1.2) 1.0(1.2)
CMI Neurological Score 1.2 (0.9) 1.6 (1.5) 0.8(0.8)

SBP = blood pressure.
b CMI = Cornell Medical Index.

Nocturnal Respiratory and Sleep/Wake Variables

EEG Data

Electroencephographic variables were recorded

overnight for all subjects. Records were scored by a

trained technician who utilized the system of Arnew and

Webb (1972) to determine sleep stages. In three records,

EEG tracings were judged inadequate for differential sleep

staging. However, a judgement of sleep vs. wake was made

for these subjects in order to form a basis for the

respiratory indices. These three EEG records were

excluded from the following analyses, leaving 43 records

as a base for analysis.

Fifteen variables (time in bed, pure sleep time,

sleep efficiency index, number of stage 0 periods, time E

stage 0, sleep latency, time % stages 1, 2, 3, 4, and RLM,

latency 1st REM period, mean REM period length, mean REM

cycle length, 0 slow wave sleep), reflecting multiple

aspects of the sleep of these subjects, were derived from

the sleep stage scoring. Table 3-6 presents the means ana

standard deviations for these subjects. Although subjects

spent over six hours in bed, the table indicates that not

quite 5 hours were spent asleep. The remaining variables

suggest that an increase in sleep latency, and time awake,

is responsible for the diminished sleep time, while

lighter sleep is also increased.

Table 3-i. e!eans and standard deviations of selected sleep
variables from 43 snoring males.

Mean Deviation

Time in Bed 377.3 45.1
Sleep Latency 29.2 32.9
Pure Sleep Time 288.7 75.8
Sleep Eificiency Index .76 .16
Number Stage 0 Periods 8.6 5.6
Time ? Stage 0 14.3 12.7
Time % Stage 1 3.3 2.8
Time 4 Stage 2 60.8 10.3
Time % Stage 5 3.4 2.4
Time % Stage 4 7.9 6.4
Time % Stage REN 11.5 7.2
Latency 1st HEI Period 124.7 92.2
Mean REIi Period Length 14.9 9.4
RL! Cycle Length 74.9 57.7
% Slow Wave Sleep 12.9 8.3

Note: see text for explanation of N.

Partial correlations, controlling for age, were

calculated between these sleep and the respiratory

variables. Out of a total of 105 correlations, only one

was significant at p<.05 or less: Number of desaturations

>4% vs REM latency (r=-.369).

Table 3-7 presents the sleep variable means for the

subjects grouped by level of apnea/hypopnea. The group

with high apnea/hypopnea appears to spend more time awake,

less time asleep, and achieve less slow wave sleep than

the others. However, none of these differences proved

significant. In fact, only one out of 15 oneway ANOVAs

achieved significance: Time in bed, F(2,40)=0.4, p=NS;

Sleep latency, F(2,40)=1.7, p-iS; Pure sleep time,

F(2,40)=0.3, p=IS; Sleep efficiency index, F(2,40)=0.2,

p=nlS; Number of stage C periods, F(2,40)=0.5, p=NS; Time

5 stage 0, F(2,40)=1.3, p=NS; Time % stage 1,

F(2,40)=1.3, p=NS; Time % stage 2, F(2,40)=0.04, p=NS;

Time % stage 3, F(2,40)=3.64, p<.04; Time % stage 4,

F(2,40)=1.32, p=NS; Time % stage REM, F(2,40)=1.5,

p=NS: Latency 1st REM, F(2,40)=0.8, p=NS; Mean REM

period length, F(2,40)=0.3, p=NS: Mean REM cycle length,

F(2,40)=0.9, p=NS; slow wave sleep, F(2,40)=2.4,

p=NS. Scheffe followup procedures for time % stage 3

revealed no significant between group comparisons.

To summarize, EEG sleep measures indicated lighter

sleep then normal in this sample, with increases in

awakenings and light slow wave sleep observed. Neither

between group comparisons nor correlational procedures

revealed significant interactions between sleep and

nocturnal respiratory variables.


0 O ^^-LnOOrnri^-

C7 1m1;N2rNtavNC

o 0c f

to . .
*r41 cu^

H nUN O O =3 CNc CO -cz
0. . .

0(U n N n C- N K C- %D Z C -' C

ID (U. c 0 X
co 0 1-J N m nL-
C O3 Q 0 * ., .

- 4q
u0 0

0. C
C. -
.cu Q

4-1 a)
au ra 'Z
o 'oQo H-t-- c )i o

cu w-

C.;F=r0 N 0t
10 N1 a
'0 00 O4- 0

000 ti Oo
)o a c0 4 -

0W > *.
a0 C 004 P 0+

0a- d 0 J2 0

0W"ICCO 0nCiQ)0 Z (U

) a C0 0C *0
i 00000000000C
ac C) ma0aaa )w

*P(U -102 0 0E) 0
1; .c, E EE F ao

n E aUCLWZ) E- E-4 E L, oZ C: H
*o~~~~~~z t: u a i im iiau u

Daytime Sleepiness Data

Included among the measures of daytime sleepiness

were the mean of the several Stanford Sleepiness Scale

ratings, and two variables calculated from the two evening

naps: mean sleep latency and total minutes asleep during

the naps. These variables were correlated with the

respiratory variables, resulting in a Pearson correlation

matrix. From these 21 correlations, two emerged signi-

ficant at p<.05 or less, both involving the mean Stanford

Sleepiness Scale ratings. Mean SSS ratings were corre-

lated with apnea index (r=.297), and apnea + hypopnea

index (r=.283).

Table 3-8 presents the means and standard deviations

for the subjects grouped by level of apnea/hypopnea. One-

way ANOVA procedures were non-significant for all these

variables [SSS: F(2,38)=2.4, p=NS; Mean sleep latency:

F(2,35)=0.4, p=NS; Total minutes asleep: F(2,35)=0.3,


Table 3-8. Means and standard deviations for daytime
sleepiness variables in 46 snoring males grouped
by level of apnea/hypopnea.

ho Apnea/ Low Apnea/ High Apnea/
Hypopnea Hypopnea Hypopnea
Stanford Sleepiness
Scale 3.2 (.7) 3.6 (.8) 4.0 (.7)
Mean Nap Latency 14.5 (5.7) 15.5 (5.3) 13.4(3.4)
Total Miinutes of
Sleep in Naps 11.3(11.9) 9.6(10.0) 13.5(6.6)

In summary, correlational procedures suggested

relationships between nocturnal respiratory variables and

subjective sleepiness, while no reliable between group

difference were noted on the nap or SSS variables.

Subjective Sleep Assessment

Two assessments of subjective sleep parameters were

completed, including the sleep questionnaire and sleep

log. Several variables were selected from the sleep

questionnaire and included: usual hours of sleep,

judgement of whether usual sleep was adequate, number of

naps per week, hours napping per week, number of nightly

wakenings, minutes of nightly wakenings, usual sleep

latency, depth of sleep, and daytime sleepiness. These

variables were correlated with the respiratory variables,

for a Pearson matrix of 63 correlations. Table 3-9

presents the significant correlations which emerged from

this matrix. Several respiratory variables were

associated with napping variables, while mean low

saturation and number of desaturations >4% were associated

with reported minutes of nightly wakenings.

Table 3-10 presents the means and standard deviations

from the subjects grouped by level of apnea/hypopnea.

Oneway ANOVA procedures revealed no significant between

group differences on the selected sleep questionnaire

variables [Usual hours sleep, F(2,43)=G.6, p=NS; Adequate

sleep, F(2,43)=0.3, p=NS; number of naps, F(2,43)=2.9,

p=NS; Hours napping, F(2,43)=0.8, p=NS; Number of

Table 3-9. Significant (p<.05) Pearson correlations between
nocturnal respiratory and sleep questionnaire
variables in 46 snoring males.

Number hours Nightly
of Naps Napping Awakening

Apnea Index
Apnea+Hypopnea Index .249
Seconds in Events .331
Pean High Saturation
Mean Low Saturation .307
Lumber of Desaturations >4% .283 -.291
number of Deseturations 10% .403 .376

Table 3-10. Means and standard deviations of sleep
questionnaire variables derived from 46 snoring
males grouped by level of apnea/hypopnea.

No Apnea/ Low Apnea/ High Apnea/
Hypopnea hypopnea Hypopnea
eean h. Sleep 7.4 (.9) 7.0(1.6) 7.5(1.0)
Adecuate Amount? 1.9 (.4) 1.8 (.4) 2.0(0)
Number of laps/Week 1.8(1.5) 2.3(2.6) 4.3(2.2)
hours Napping/Week 2.6(1.8) 3.5(2.7) 4.0(2.7)
Dumber of Nightly
lakenings 1.1 (.9) 1.5(1.3) 1.6 (.4)
Minutes Nightly
Awakenings 1.9(1.1) 1.5(1.0) 1.5 (.8)
Sleep Latency 2.0 (.9) 2.0 (.6) 2.0 (.6)
Depth of Sleep 2.4 (.8) 2.1 (.9) 3.0(C)
Daytime Sleepiness 1.6 (.6) 1.9 (.4) 1.8 (.4)

awakenings, F(2,43)=0.9, p=LS: Ilinutes of awakening,

F(2,43)=0.8, p=NS; Sleep latency, F(2,43)=0.01, p=NS;

Depth of sleep, F(2,43)=2.6, p=NS; Daytime sleepiness,

F(2,43)=1.8, p=NS].

Means were calculated from the 7 day sleep logs, and

the following variables were selected for analysis:

reports of daytime sleepiness, number of naps, minutes

napping, total bedtime, sleep latency, number of

awakenings, minutes of awakenings. These variables were

correlated with the respiratory indices to form a Pearson

correlation matrix totalling 49 correlations. Table 3-11

presents the 6 significant (p<.05) correlations which

emerged. Again, number of desaturations were associated

with reports of napping, while mean high saturation was

associated with total bedtime, sleep latency, and minutes

awake, and mean low saturation was associated with minutes

awake as well.

Table 3-12 presents the means and standard deviations

for the subjects divided into levels of apnea/hypopnea.

Oneway ANCVA procedures failed to reveal significant

between group differences [Daytime sleepiness,

F(2,35)=0.6, p=NS; Number of naps, F(2,35)=0.3, p=NS;

Minutes of napping, F(2,35)=1.3, p=NS; Total bedtime,

F(2,35)=0.7, p=NS; Sleep latency, F(2.35)=0.2, p=NS;

Number of awakenings, F(2,35)=0.2, p=NS; Minutes of

wakenings, F(2,35)=.0001, p=NS].

Table 3-11.

Significant (p<.05) Pearson correlations between
nocturnal respiratory and sleep log variables in
46 snoring males.

Number Minutes
of Total Sleep Nightly
Naps Bedtime Latency Wakenings

Apnea Index
Apnea+Hypopnea Index
Seconds in Events
Mean High Saturation .237 .250 .359
Mean Low Saturation .265
lumber of
Desaturations >4% .240
Number of
Desaturations >10% .240

Table 3-12. Means and standard deviations of sleep log
variables derived from 46 snoring males grouped
by level of apnea/hypopnea.

No Apnea/ Low Apnea/ High Apnea/
hypopnea hypopnea Hypopnea
Daytime Sleepiness 1.9 (.9) 1.9(1.1) 2.4 (.3)
Number of Naps 0.2 (.2) .3 (.4) .3 (.3)
Minutes Napping 0.0(0) .2 (.5) 0 (0)
Total Bed Time 6.5(3.1) 5.8(3.4) 7.5 (.7)
Sleep Latency 10.5(8.1) 11.1(9.8) 8.4(4.7)
Number of Wakenings 1.3(0.8) 1.3 (.9) 1.3 (.2)
Minutes making 2.0(1.4) 1.7(1.1) 1.7 (.3)

In summary, no between group differences were noted

for either sleep questionnaire or log variables when

subjects were grouped by level of apnea/hypopnea.

Correlational procedures suggested relationships between

several respiratory variables and reports of napping and

wakening after sleep onset on the sleep questionnaire.

These findings were cross validated in the correlational

analysis of sleep log data, in which reports of napping

and wakening after sleep onset were related to respiratory

variables, as were total bedtime and sleep latency.

Nocturnal Respiratory and Neuropsychological Variables

Scores derived from the cognitive/neuropsychological

battery included WAIS Performance IQ, WAIS Verbal IQ,

Wechsler Memory Quotient, delayed recall of logical

stories, delayed recall of visual reproductions, delayed

recall of Rey complex figure, digit span, Hooper score,

Wisconsin card sort score, finger tapping right and left,

and Verbal fluency.

As a preliminary analysis, these scores were compared

with selected respiratory and demographic variables in a

multivariate regression procedure. In this type of

analysis, a set of predictor scores is used to

simultaneously predict a set of criterion variables while

controlling for intercorrelations amongst the measures.

An overall test of significance, such as the Hottelings

T2, establishes the probability of obtaining the observed

relationships from chance alone. Upon obtaining a

significant overall p value, univariate regressions are

used to individually predict criterion variables with the

multiple predictors. Contribution of individual

predictors may also be examined in these univariate


The results from the multivariate regression of the

above noted cognitive scores and selected nocturnal

respiratory variables (Apnea index, Apnea + hypopnea

index, lean high saturation, Number of desaturations 14%,

Age, Weight:Height ratio) are presented in Table 3-13. An

overall Hotellings T2 score was significant (p<.003),

allowing examination of the univariate regressions.

Criterion variables which were significantly predicted by

the respiratory and demographic variables included WAIS

PIQ, VMS NQ, delayed recall of Rey figure and visual

reproductions, Hooper, and Wisconsin card sort. These

relationships reflected negative relationship between

overnight respiratory indices and neuropsychological

scores. Respiratory indices formed significant components

of the predictor combinations for all these variables save

Hooper score, suggesting relationships between nocturnal

respiratory indices and non-verbal intelligence, verbal

and visual memory, and ability to shift sets while problem


Because age is correlated with nocturnal respiratory

parameters in past reports, and also with neuropsycho-

logical scores in other samples, the various overnight

Table 3-13. Multivariate regression of demographic and
nocturnal respiratory variables on cognitive
scores in 46 snoring males.

Predictor Variables
Weight:Height Ratio
Apnea Index
Apnea+Hypopnea Index
Mean High Saturation
Number of Desaturations >4%

Criterion Variables
WAIS Verbal IQ
WAIS Performance IQ
Wechsler Memory Scale Memory
Delayed Recall Logical Stories
Delayed Recall Visual
Delayed Recall Hey Figure
Digit Span
Hooper Test
Finger Tapping Left Hand
Finger Tapping Right Hand
Verbal Fluency
Wisconsin Card Sort

Multivariate test of significance

Hotellings T2 = 4.841 (p<.003)

Univariate Regression with
WAIS Verbal IQ
WAIS Perlormance IQ
Wechsler Memory Scale
Memory Quotient
Delayed Recall
Logical Stories
Delayed Recall
Visual Reproductions
Delayed Recall Rey Figure
Digit Span
Hooper Test
Finger Tapping Right Hand
Finger Tapping Left Hand
Verbal Fluency
Wisconsin Card Sort

(6,3e) d.f.
Multiple R




Breakdown of significantly predicted criterion regressions

WAIS Performance

Weight:Height Ratio
Apnea Index
Apnea+Hypopnea Index
Mean High Saturation
Number of
Desaturations >4?'



-.3978 .024









Table 3-13-continued.

Wechsler Memory
Scale Eiemory

Delayed Recall

Delayed Recall
Rey FiFure

Hooper Test

iisconsin Card

Weight:Height Ratio
Apnea Index
Apnea+Hypopnea Index
Mean High Saturation
Number of
Desaturations >44%

Weight:Height Ratio
Apnea Index
Apnea+Hypopnea Index
Mean High Saturation
Number of
Desaturations >45,

Weight:Height Ratio
Apnea Index
Apnea+Hypopnea Index
Mean High Saturation
Number of
Desaturations >45%

l\eight:Height Ratio
Apnea Index
Apnea+Hypopnea Index
Mean High Saturation
Number of
Desaturations >,4%

Weight:Height Ratio
Apnea Index
Apnea+Hypopnea Index
Nean High Saturation
Number of
Desaturations >45





















scores were entered into a partial correlation matrix,

controlling for age, with the cognitive scores. Table

3-14 presents the resulting correlations significant at

p<.05 or less. The table incicates that every cognitive

test but digit span and Hooper correlated with at least

one overnight respiratory index. Of additional impact is

the observation that in every case of a significant

correlation, deteriorating respiratory indices are

associated with worsening neuropsychological scores.

Respiratory indices with particularly heavy loadings on

neuropsychological scores included those quantifying

apneas, (Apnea index, Apnea + hypopnea index) as well as

those more directly representing deepening hypoxia (Mean

low saturation and Number of desaturations >4%).

Subjects were grouped by level of nocturnal

respiratory disorder (no apneas/hypopneas, subclinical

level of apnea/hypopnea, and clinically significant level

of apnea/hypopnea). Means and standard deviations for the

neuropsychological scores from these groups are presented

in Table 3-15. The table suggests a general deterioration

of cognitive scores as level of nocturnal respiratory

distress increases. Oneway AIIOVA procedures were applied

to the neuropsychological scores from each test. Signifi-

cant between group differences emerged for WAIS PIQ

[F(2,43)=6.9, p<.002] delayed recall of logical stores

[F[2,43)=4.0, p<.03] delayed recall of visual reproduc-

tions [F(2,45)=3.4, p<.04], delayed recall of Hey complex

-OO 00 0

0 0
0 r- ?N1

I f I I I I I

C6 CO CUD N .f

*-" U-C- 0 0 L 0N 3

0 C

S4-) *
C4E o c
0 N 0 0 N o ,

0 I0 I .

0 <;
O C.
1 0 > ZC


41 r r- o
O -P ( C X *

r-O co 0 9 z
-P C

0U o 0

Q0 0 c O w
*Uc- t C

.H0 CO to
4OCO C40 F- C C0

e 33 a c
C "O 2i C >, a CUUflC
c] CU a 4-1 0 C a) w i

rC o) -r ai 4 W ., M aU
a:ECUE W0 0CU C: CrC

>CO 0 0. O C H- C

E cC 0 0 0 00 C. O *C -C .> *
ar) H 0 H .1 a H > a

Table 3-15. teans and standard deviations of cognitive
variables in 46 snoring males groupec by level
of apnea/hypopnea.


WAIS Verbal IQ
WAIS Performance IQ
Weschler Memory
Scale memory
Delayed Recall
Logical Stories
Delayed Recall
Digit Span
Delayed Recall
Rey Figure
Hooper Test
.isconsin Card Sort
Finger Tapping
Right Hand
Finger Tapping
Left Hard
Verbal Fluency

No Apnea/ Low Apnea/ High Apnea/
Hypopnea Hypopnea Hypopnea



111.5 (9.6)

124.7(14.3) 117.1(19.0) 107.8(18.7)

9.9 (2.4) 8.6 (3.2) 5.8 (4.0)a

10.5 (5.7) 10.4 (3.4) (.1 (4.7)
6.8 (0.8) 6.8 (1.2) t.6 (.8)

25.6 (5.4)
25.2 (2.6)
4.1 (1.2)

23.2 (6.0)
25.5 (4.8)
3.4 (1.5)

15.1 (8.4)a,b
22.8 (4.1)

62.6 (8.5) 57.5(10.4) 56.1 (8.3)

59.1 (9.0) 54.0 (9.7) 54.8(15.1)
14.6 (4.5) 13.1 (4.2) 9.3 (2.6)e

a Significantly different from no sleep apnea/hypopnea.
b Significantly different from low sleep apnea/hypopnea.

figure [F[2,45)=6.5, p<.OO3], and Verbal fluency,

[F[2,43)=3.6, p<.04]. Followup comparisons (Scheffe)

showed the group with high apnea/hypopnea to be

significantly impaired relative to those with no sleep

apnea/hypopnea on WAIS PIQ, delayed recall of logical

stories and Rey complex figure as well as verbal

fluency. Additionally, the high apnea/hypopnea group was

impaired relative to those with subclinical apnea/hypopnea

on aelayea recall of the Rey complex figure. These

findings are bolstered by the lack of significant between

group differences on age, weight, or education (see Table

3-15), all of which are potential alternative explanations

of the between group differences.

As a check, an analysis was run to evaluate the

relative contribution of nocturnal respiratory variables

and sleep variables reflecting fragmentation in predicting

cognitive/neuropsychological scores. A multivariate

regression of these scores was carried out. Demographic

and respiratory scores included age, apnea index, apnea +

hypopnea index, and number of desaturations >4%. A subset

of the earlier noted sleep variables was chosen to reflect

measures thought to be most responsive to sleep

fragmentation. These included sleep efficiency, number of

stage O periods, time % of stage 0, time % of stage 2, and

number of stage changes. These respiratory and sleep

parameters were used to predict the 5 cognitive scores

which were earlier shown to be predicted by nocturnal

respiratory scores: WAIS PIQ, Wechsler Memory Quotient,

delayed recall of visual reproductions and Hey figure, and

Wisconsin card sort. Table 3-16 presents the data from

this analysis. The Hotellings T2 was significant,

(p<.001) with delayed recall of visual reproductions and

Rey figure significantly predicted by the predictor

scores. Inspection of the individual regressions for

these two measures discloses that age, apnea index, and

number of desaturations >4% all formed significant

components of the prediction of delayed recall of visual

reproductions, while sleep efficiency was the sole

significant predictor of the Key figure. These data seem

supportive of an independent contribution of both

nocturnal respiratory variables and sleep variables at a

limited level, to the prediction of cognitive/neuro-

psychological scores.

In summary, various statistical procedures indicate

relationships between nocturnal respiratory measures and

non-verbal intelligence, verbal and non-verbal memory,

expressive verbal fluency, and ability to shift cognitive

set while problem solving. Division of subjects by level

of nocturnal respiratory distress indicated that subjects

with high amounts of apnea/hypopnea were impaired relative

to controls on measures of non-verbal intelligence, verbal

and non-verbal memory, and expressive verbal fluency, as

well as being impaired relative to subjects with

subclinical levels of apnea/hypopnea on measures of

Multivariate regression of sleep and nocturnal
respiratory on cognitive variables in 43
snoring males.

Predictor Variables

Apnea Index
Apnea+Hypopnea Index
Number of Desaturations >4%
Sleep Efficiency
Number Stage 0 Periods
Time % Stage 0
Time C Stage 2
Lumber of Stage Changes

Criterion Variables

WAIS Performance IQ
Wechsler Memory Scale Memory
Delayed Recall Visual
Delayed Recall Rey Figure
Verbal Fluency
Wisconsin Card Sort

Sultivariate test of significance

Hotellings T2 = 4.063

Univariate Regression
WAIS Performance IQ
Vechsler Memory Scale
Memory Quotient
Delayed Recall Visual
Delayed Recall Rey
Verbal Fluency
Wisconsin Card Sort


with (9,33) d.f.
Multiple R




Breakdown of significantly predicted criterion regressions
Criterion Predictor Beta Sig.
Delayed Recall Age -.3596 .020
Visual Apnea Index -.3218 .021
Reproduction Apnea+Hypopnea Index .2529 .218
Number of
Desaturations >4% -.3778 .041
Sleep Efficiency .2303 .408
Number of Stage 0 Periods .3260 .209
Time % Stage 0 -.1136 .731
Time % Stage 2 .0005 .998
Number of Stage Changes -.2235 .391

Table 3-16.





Si .




Table -1I-continuec.

Delayed recall
Key Figure

Apnea Index
Apnea+Hypopnea Index
Lumber of
Desaturations >4%
Sleep Efficiency
lumber of Stage 0 Periods
Time S Stage 0
Time % Stage 2
Number of Stage Changes


-. 054




non-verbal memory. Addition of variables reflecting sleep

fragmentation to the analysis indicated that both

nocturnal respiratory and sleep variables make significant

and independent contributions to prediction of

cognitive/neuropsychological scores.


Demographics and Incidence of
Nocturnal Respiratory Disorder

The present sample of 46 snoring males was selected

as a group likely to display a wide range of sleep

disordered breathing. These subjects, with a mean age of

50 years and a mean weight of 190 pounds, reflect a mostly

middle-aged to older male population which is moderately

obese. The demographic characteristics of this sample are

similar to those of the "typical" sleep apnea syndrome

patient (Guilleminault et al., 1978), although the

requirement of general good health insured that subjects

with a classic SAS were excluded. This sample may be

considered representative of subjects predisposed to sleep

disordered breathing, although at no more than a

subclinical level. Therefore it seems valid to consider

various risk factors as they relate to nocturnal


In the present sample, 62% experienced at least one

apnea or hypopnea, while 13% suffered higher levels of

apnea/hypopnea. Only two other reports have described

samples selected for snoring or respiratory complaints.

Ancoli-Israel et al. (1981) examined 24 subjects with

complaints of respiratory difficulties, daytime

sleepiness, and muscular events (nocturnal myoclonus).

Further, these subjects had a mean age of almost 70 years,

indicating significant sampling and age differences from

the present study. Ninety-three per cent of these

subjects had at least one event, while 25% had high

numbers of apnea/hypopnea. These rates are much higher

than those found in the present sample. This is probably

attributable to the sampling differences noted above.

However, a recent report by Miles and Simmons (1984)

described apneas in a series of 190 patients referred for

complaints of heavy snoring. In this sample, 73% had at

least one event, while 9% had severe levels of apnea.

These data are quite comparable to the prevalences

observed in the present sample, 62% for at least one event

and 13i for high levels of apnea/hypopnea, and lend some

confidence to the present finding.

Thus the overall frequency of sleep disordered

breathing in middle-aged subjects with heavy snoring

appears to be between 62 and 75% for at least one event,

and between 9 and 135 for more severe levels of apnea.

The relatively close concordance of the two reports on

heavy snoring subjects is reassuring.

Factors which were found to be related to the

occurrence of sleep apneas in the previous review included

male sex, complaints of snoring, increasing age, and

obesity. The first two factors could not be examined in

the present sample because of their use as selection

criteria. A wide range of age was achieved in this

sample, but neither non-parametric nor correlational

procedures revealed significant relationships between

increasing age and sleep disordered breathing. Although

several previous reports have noted relationships between

increasing age and sleep disordered breathing, typically

this relationship is moderate for apnea (Block et al.,

1979; Carskadon et al., 1980; Bixler et al., 1982).

Additionally, no other correlational data are available in

subjects selected for sleep and breathing complaints. It

may be speculated that the effect of snoring masks the

effect of age in this group. Alternatively, some quirk of

sampling may have generated this effect. Correlational

procedures, however, revealed substantial relationships

between weight and sleep disordered breathing. This

finding is consistent with several past reports. To

summarize, the present data support relationships between

increasing weight and nocturnal events, while

relationships between age and sleep disordered breathing

are not found in this sample of heavy snoring males,

perhaps because the effect of age is masked by the effect

of snoring.

The temporal characteristics of the apneic events

assessed by two hour interval suggested an increase in

events between 3 and 5 am, although differences in sleep

onset time and wakeup time render any interpretation of

these data somewhat tenuous. This finding needs

replication before any conclusions may be drawn.

Ninety-eight per cent of apneic events occurred in

light slow wave and REM sleep, an observation which is not

surprising given that subjects spent nearly 80% of their

sleep in these stages. The preponderance of events in

these stages of sleep is consistent with the bulk of past

reports in this area (e.g, Bixler et al., 1982; Block et

al., 1979; Krieger et al., 1985).

When subjects were divided into groups by level of

apnea/hypopnea, significant differences emerged on oxygen

desaturation variables. Specifically, comparisons

indicated that the group with a high level of

apnea/hypopnea events experienced deeper desaturations

than the remaining two groups. This difference occurred

despite a lack of differences on age and weight among the

three groups, and appears to implicate the respiratory

events in the genesis of the desaturations. This is

consistent with an earlier report by McGinty et al. (1982)

who demonstrated that subjects with significant levels of

sleep disordered breathing were impaired on oxygen

desaturation variables, relative to subjects without

respiratory events. Subjects with heavy snoring and

multiple nocturnal apneic respiratory events appear to be

at risk for nocturnal oxygen desaturation.

In summary, the present sample of 46 snoring males

appears representative of those predisposed to experience

sleep disordered breathing (middle-aged, snoring males).

A prevalence of 62k was observed for occurrence of at

least one event, while 13% had high numbers of

apnea/hypopnea, data which are consistent with an earlier

report drawn from heavy snoring males. Most events

occurred in light slow wave or REM sleep. No effect of

age on level of sleep disordered breathing was observed,

although increasing weight was linked to nocturnal

respiratory events. When the sample was divided by level

of apnea/hypopnea, subjects with high numbers of

apnea/hypopnea suffered significantly exacerbated oxygen

desaturation, although no significant differences were

noted for weight or age among the groups.

ijocturnal Respiratory and Health Variables

The finding of only very limited relationships

between nocturnal respiratory events and health variables

must be tempered by an acknowledgement of possible bias

against such findings in this sample. Specifically, the

sample was selected to be normal and healthy, a criterion

which undoubtably restricted the range of possible health

pathology, and hence probably attenuated the correlational

findings. Additionally, the self-report health checklist,

the Cornell Medical Index, cannot be regarded as a

powerful measure of health pathology. In a very real

sense then, the odds were against finding reliable

relationships between nocturnal respiratory and health


In light of these caveats, it is of interest that two

measures of desaturation, mean low saturation, and number

of desaturations >04 were associated with increasing blood

pressure readings. While a report by Guilleminault et al.

(1978) noted an increasing prevalence of hypertension in

subjects diagnosed with sleep apnea syndrome, a

relationship between desaturation and hypertension has not

been previously reported. This finding was observed in

spite of the fact that subjects prescribed hypertensives

were not medication free for the pressure reading.

Additionally, a check of self reports of diagnosed

hypertension showed six subjects in the no event group as

having diagnosed hypertension, as opposed to six in the

low apnea/hypopnea group, and one in the high

apnea/hypopnea group, data which did not link

apnea/hypopnea, per se, with hypertension in these

subjects. The isolated finding of a relationship between

mean high saturation and overall number of symptoms

endorsed on the CMI is more puzzling and does not follow

any logical analysis.

In summary, the present study's assessment of health

variables must be regarded as somewhat limited. In spite

of this, an association between two measures of desatura-

tion and a measure of blood pressure was noted. It is

speculated that blood pressure may be more responsive to

nocturnal oxygen saturation tnan to the apneic events


Nocturnal Respiratory and Sleep/Wake Variables

EEG Data

The overnight sleep in the laboratory for these

snoring subjects was noteworthy for a general shift

towards lighter sleep than usual. A longer sleep latency,

more awakenings, more light sleep, and less slow wave and

REM sleep than normal are apparent. These findings

parallel the report by Block et al. (1979) who found

shifts to lighter sleep and more awakenings than in a

sample of normal subjects. These changes were said to

result from a first night in a strange bed and the

discomfort of the multiple clips and sensors used to

monitor respiration. Recording montages were identical

for the present subjects, and these two factors

undoubtably influenced the sleep of the present subjects

as well.

An intriguing observation was a lack of covariation

between respiratory and EEG sleep variables derived from

the recording night. This finding is counterintuitive, as

one would expect arousals from the respiratory events to

influence the sleep characteristics of subjects with

events. However, correlational procedures failed to

support this notion. It may be that a ceiling effect was

exerted on sleep measures sensitive to disturbance, one

that resulted from the first night effect compounded by

the discomfort of the recording procedure. The already

elevated disturbance sensitive sleep variables might not

Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EOGSOKC06_8RP790 INGEST_TIME 2012-09-24T13:31:42Z PACKAGE AA00011824_00001