You are on page 1of 6

Heart Rate Variability in Insomniacs and Matched Normal Sleepers


Objective: It was hypothesized that psychophysiological insomniacs, who have been shown to have elevated heart rate, body
temperature, and whole body metabolic rate, would also have increased low frequency and decreased high frequency power in the
spectral analysis of their heart period data. Method: Groups of 12 objectively defined insomniacs and age-, sex-, and
weight-matched controls with normal sleep were evaluated on sleep and EKG measures over a 36-hour sleep laboratory stay.
Results: Heart period was decreased (ie, heart rate was increased) and its SD was decreased in all stages of sleep in the insomniacs
compared with the controls. Spectral analysis revealed significantly increased low frequency power and decreased high frequency
power in insomniacs compared with controls across all stages of sleep. Conclusions: Because increased low frequency spectral
power is an indicator of increased sympathetic nervous system activity, these data imply that chronic insomniacs could be at
increased risk for the development of disorders, such as coronary artery disease, that are related to increased sympathetic nervous
system activity. Key words: insomnia, heart rate, heart rate variability, spectral analysis, sympathetic nervous system,
parasympathetic nervous system, sleep disorders, sleep.

performance in insomniacs with normal controls generally

EMG = electromyogram; EEG = electroencephalogram; have not found differences on tests that are sensitive to sleep
MSLT = Multiple Sleep Latency Test; Ss = subjects; loss (13, 18). On the basis of the results of these studies and
SaO 2 = oxygen saturation; AD = analog to digital; patient reports that they are fatigued or "washed out" during
REM = rapid eye movements; SWS = slow wave sleep; the day, investigators have hypothesized that standard sleep
LFP = low frequency power; HFP = high frequency and sleep loss tests are confounded in that they "simultaneous-
power. ly measure sleep need and hyperarousal, which is interfering
with sleep onset" (Ref. 15, p. 59). This concept is supported
by studies reporting significant negative correlations between
INTRODUCTION total sleep at night and MSLT values the next day (13, 15, 17).
Several studies have shown that patients with chronic Increased heart rate, body temperature, metabolic rate, and
psychophysiological insomnia show evidence of elevated secretion of corticosteroids and adrenaline are all indicators of
physiological activity during their sleep. For example, Monroe increased sympathetic nervous system activity. Spectral anal-
(1) reported a significant increase in rectal temperature, basal ysis techniques also have been used to link changes in heart
skin resistance, and phasic vasoconstrictions 30 minutes rate to underlying central nervous system activity. Spectral
before and during sleep in insomniacs compared with normal analysis has suggested that power in specific frequency bands
sleepers. It was also reported that insomniacs had increased can be related to parasympathetic and sympathetic nervous
heart rate, but this difference was not statistically significant. system activity. Specifically, relative power in high frequency
In other studies, sleep-onset insomniacs have been shown to areas, usually from 0.15 to 0.5 Hz, has been used to infer
have increased heart rate (2, 3), increased frontalis EMG (4), parasympathetic nervous system activity. Evidence has been
increased finger temperature, increased 24-hour metabolic rate published that the peak in the lower frequency component is in
(5), and more /3 and less a frequencies in their EEG (6, 7) the range of .09 to .11 with a SD of about .02 (19). To include
before sleep onset. Poor sleepers have increased secretion of both the peak and surrounding area, a range of lower frequen-
corticosteroids and adrenaline (8, 9) compared with good cies from 0.05 to 0.15 Hz typically has been related to a
sleepers in most, but not all, studies (10). combination of parasympathetic and sympathetic influences
Another line of research has examined the daytime func- (20-22). Because the lower frequency power is a combination
tioning of insomniacs to determine the existence of subjec- of sympathetic and parasympathetic effects, investigators
tively reported deficits in performance, mood, and alertness. frequently infer sympathetic nervous system activity from the
The cumulative partial sleep deprivation that should arise ratio of low (parasympathetic and sympathetic) to high (pre-
from chronic insomnia would be expected to produce daytime dominantly sympathetic) power so that parasympathetic
sleepiness or increased susceptibility to acute sleep loss in power to some extent cancels out of the ratio (21, 23, 24),
insomniacs, but studies have consistently found that insomni- leaving a better indicator of sympathetic activity. Jaffe et al.
acs are not sleepier than normal controls on MSLT (11-13) or (20), in a study of heart rate changes as a function of posture
after sleep loss (14) and may actually have longer MSLT change, found this "sympathetic index" correlated to heart rate
latencies (5, 15, 16). Studies have found that insomniacs made changes. Studies of children (25) and adults (26), using
more errors on a line tracing task (17), produced fewer spectral analysis techniques, consistently have found in-
responses in a word category test (11), and performed worse creased parasympathetic activity during NREM sleep, as
on the Romberg (balance) test (18). These results may be measured by high frequency heart rate variability and in-
interpreted as insomniacs producing worse performance on creased sympathetic activity during REM sleep, as typically
tests where too much arousal reduces steadiness or blocks measured by the ratio of lower to higher frequency heart rate
higher order associates. However, studies comparing daytime variability. These findings agree well with earlier studies of
heart rate and direct measures of nerve activity (27, 28).
In the current study, heart rate variability was examined in
From Ihe Dayton VA Hospital. Wright State University, and Kettering Medical Center psychophysiological insomniacs and matched normals with
Address reprint requests to: Michael H. Bonnet, PhD (15 IN), Department of Veterans the hypothesis that, throughout the night, insomniacs would
Affairs Medical Center, 4100 W. Third St., Dayton, OH 45428. E:mail: display increased heart rate (as indexed throughout this study
Received for publication August 13, 1997; revision received December 17, 1997. as shorter periods between heart beats), decreased variability,

610 Psychosomatic Medicine 60:610-615 (1998)

Copyright © 1998 by Ihe. American Psychosomatic Society

and increased low frequency spectral power characteristic of to have EEG sleep latencies greater than 30 minutes on both
increased sympathetic nervous system activity. laboratory nights or to have a sleep efficiency of less than 85% on
both nights. Subjects with an apnea/hypopnea index greater than 10
METHOD or a periodic leg movement arousal index greater than 10 were
Subjects To qualify for the study as a normal, individuals were required to
Ss were required to be healthy, 18- to 50-year-old men and report normal sleep as described earlier and to have EEG sleep
women. Potential Ss were solicited from sleep center referrals and latencies of less than 30 minutes on both laboratory nights and to
from advertisements in the local newspapers for participants in sleep have a sleep efficiency greater than 90% on both nights (and not to
research. have sleep apnea or leg movements as defined above). Twelve
Insomniacs. Individuals completing a screening questionnaire normal participants met the above requirements and could be
indicating a) that they had a sleep problem; b) that it took them 45 matched with one of the insomniacs on the basis of age, weight, and
minutes or more to fall asleep at least 4 nights each week or that they sex. Therefore, the results reported in this study reflect the data
were awake for 60 minutes or more each night after falling asleep for obtained from two groups of 12 age-, sex-, and weight-matched
at least 4 nights each week; and c) that this condition had existed for individuals (normals and insomniacs).
at least 1 year were considered for participation in the study (see
common exclusions below). EKG Data Collection. Beginning 30 minutes before lights out on
Normals. Normals were required to indicate normal sleep on their the second night in the laboratory, EKG data were digitized by a
screening questionnaire. They were also required to report a sleep National Instruments NB-MIO-16 AD Board sampling at a rate of
latency of less than 30 minutes and less than 30 minutes of wake time 500 samples per second. A time code was digitized by a second
during the night. Normal subjects additionally were required to match channel on the AD board and also printed out on the polygraph paper
a qualified insomniac by sex, age (within 5 years), weight (within 25 to allow epoch by epoch matching of digitized EKG with sleep stages
pounds), and general time in bed characteristics. and events. The EKG and time code data were collected by Lab View
Exclusions. Potential normal or insomniac subjects who indicated 3.0 software running on a Macintosh II computer and stored to
excessive caffeine consumption (more than 250 mg of caffeine per optical disk. Data collection was continuous throughout the sleep
day), who were using psychoactive medication or drugs, or who had period. After awakening at their normal time of arising in the
completed a drug or alcohol abuse program within the previous year morning, Ss were asked to stand up for a final 10-minute EKG
were excluded. Ss with a history of depression or psychiatric recording period.
hospitalization within the previous year were excluded. Potential Ss After collection, the EKG and time data were visualized and
who had histories strongly suggestive of circadian desynchrony (eg, checked for artifacts with the Labview software and output to a
shift workers), sleep apnea, or periodic leg movements were ex- separate peak detection program used to construct the tachogram and
cluded. Of selected subjects, two were taking medication when associated time code. The tachogram data was again examined for
initially contacted. One subject had taken Synthroid (Boots Pharma- artifacts and converted for input into the BMDP Tl (Spectral
ceuticals, Inc., Lincolnshire, IL) for several years for an underactive Analysis) Program. Five-minute sequential time periods were iden-
thyroid and was continued on standard medication throughout the tified from the sleep record with the following constraints: a) Data
study. One subject had taken a benzodiazepine for sleep approxi- from the presleep baseline waking period and postsleep standing data
mately 3 nights per week until applying for the study. This subject were blocked and separated from the sleep period data, b) Five-
was removed from medication for 2 weeks before participation. Two minute blocks were chosen, whenever possible, to contain a single
subjects had seen a psychiatrist in their lifetime. Except for the sleep stage; this included, on occasion, shortening or lengthening a
subject removed from the benzodiazepine, no subject had used block by up to 1 minute or skipping 1 to 2 minutes of recording, c)
psychoactive medication for the 3 months before the study. When a block contained a mixture of sleep stages, the "lightest" stage
Subjects meeting the above criteria were invited to participate in was assigned (wake < stage 1 < stage 2 < REM < SWS). Whenever
the study after completing an informed consent and 2 hours of possible stages 2 and REM were not included in the same block. In
acclimatization to the laboratory with practice on computer tests and practice, all subjects had many pure blocks of stage wake, stage 2,
questionnaires to be used in the study. and REM and about 100 total blocks per night. Four Insomnia
Design. After practice, subjects were scheduled to spend 2 nights patients and four Normals did not have sufficient slow wave sleep for
and the intervening day in the laboratory. On both nights, a standard any blocks to be labeled as SWS.
clinical polysomnogram, including two eye channels, central and Analyses. Data were compared by sleep stage of block as defined
occipital EEG channels, chin and leg EMG channels, EKG, airflow, above. Two spectral variability measures were examined: a) LFP was
and chest movements, was performed (29, 30). On the first night, defined as the. total spectral power in the 0.05 to 0.15 Hz frequency
SaO2 was also recorded. On the second night, metabolic measures band divided by the total spectral power in the 0.15 to 0.50 Hz
and a time code were recorded instead of SaO2. frequency band; b) HFP was defined as the total spectral power in the
All subjects were assigned their own room for the course of the 0.15 to 0.50 Hz frequency band divided by the total spectral power
study. Each room contained a standard hospital bed and furniture, (0.0-0.50 Hz). Linear trends were removed from the data using the
including a desk with an Apple IIGS computer. Subjects participated Detrend option of the BMDP IT program, and the spectral analysis
in the study in groups of one to two individuals. Subjects completed was performed using the Rectangular option for the shape of the
tests and questionnaires at their individual computer workstation in periodogram weighting function (31). The simple mean and SD for
their room under technician observation via video monitors. Meals heart period were also calculated for the 5-minute blocks defined
and breaks were scheduled in another area of the laboratory, which above. All reported results in the text will refer to statistically
was also within the observation area of the technician. Ss performed significant differences (p < .05) except where noted otherwise. In
computer tests, completed an MMPI and a sleep history, and were fed analysis of variance in which interaction F values were nonsignifi-
the same daily menu of food prepared at the laboratory during the cant, the interaction variance was pooled with the error variance to
day. Caffeinated beverages were not available. Subjects usually did test main effects. Where indicated by significant F values from
not leave the laboratory during this day and did not engage in any analysis of variance, pairwise comparisons, using degrees of freedom
activity more vigorous than walking to the bathroom. After the defined by the Huynh-Feldt correction, were made using the New-
second night in the laboratory, subjects completed some brief man-Keuls procedure at the p < .05 level.
computer tests and were allowed to leave. Subjects were matched by age, weight, and sex specifically to
Subject Groups. To qualify for the study as an insomniac, allow within subject pair comparisons. As such, paired ( tests or
individuals were required to report insomnia as described earlier and repeated measures analyses of variance were performed on the data.

Psychosomatic Medicine 60:610-615 (1998) 611


RESULTS Heart Period Spectral Analysis Data

Demographic data from the 12 pairs of normals and Examination of the low frequency power ratio data re-
insomniacs can be found in Table 1. The groups did not differ vealed that these data were nonnormally distributed. A log
in age or weight. As expected by selection criteria, insomniacs transformation (of the data plus 1), therefore, was used to
reported significantly longer sleep latencies and shorter total normalize the data. Figures 3 and 4 present mean LFP and
sleep per night on their subjective sleep screening forms. HFP ratios from stage 1, stage 2, REM, and wake during the
Results from the performance tests will not be reported in this night for Insomniacs and Normals. Analysis of variance
paper. indicated that the interaction F value for Group by Stage was
Objective and subjective sleep data from the two groups for not significant for either LFP or HFP analyses (F(3,33) =
the EKG data collection night are presented in Table 2. It can 1.407, NS and F(3,33) = 1.766, NS). There was a significant
be seen from Table 2 that sleep was significantly different in main effect for Group for both LFP (F(l,32) = 12.93, p <
the two groups based on the inclusion criteria. Total sleep time .001) and HFP (F(l,44) = 12.21, p < .001). These results, as
was significantly shorter and wake time during the night was seen in the figures, indicated that LFP was increased and HFP
greater in the insomniacs. Decreased total sleep time in the was decreased in insomniacs, compared with normals across
insomniac group came primarily from decreased stage 2, all sleep stages. There was a significant main effect for Stage
although the 1% decrease in stage 2 was significant only at the for both LFP (F(l,32) = 12.62, p < .001) and HFP (F(l,44)
p < O.I level. The insomniacs also gave subjective reports of = 12.21, p < .001). Pairwise comparisons showed that
significantly longer latencies and decreased total sleep time as low-frequency ratio power was significantly greater when
compared with the normals on the EKG data collection night. subjects were awake than when they were in stage I or REM
and that stage 1 and REM values, which did not differ, were
Heart Period and Heart Period SD Analyses greater than those for stage 2. In a similar manner, pairwise
The interval between heart beats (in ms) and the SD of the comparisons showed that high-frequency ratio power was
intervals for the 5-minute periods selected for spectral analysis significantly less when subjects were awake than when they
were analyzed for comparison purposes. Analysis of variance were in stage 1 or REM, and that stage 1 and REM values,
with a term for Group (Insomniac vs. Normal) and Sleep Stage which did not differ, were less than those for stage 2.
(stage wake, stage 1, stage 2, and REM) was performed. The Only eight of the Insomnia Ss and eight of the Normal Ss
data are presented in Figures I and 2. Analysis of variance had sufficient SWS for any 5-minute period to be. classified as
indicated that the interaction F value for Group by Stage was SWS. Of these Ss, only four subject pairs could be formed. A
not significant for either the intervals or for the SD of the t test for LFP revealed t3 = 2.78, p < .07 (mean values of 1.53
intervals (F(3,33) = 0.63, NS and F(3,33) = 1.87, NS). There vs. 65 for insomniacs vs. normals), and the t test for HFP
was a significant main effect for Group for both beat-to-beat showed t3 = 3.69, p < .05 (mean values of 0.289 vs. 0.598 for
interval (F(l,44) = 44.67, p < .001) and the interval SD insomniacs vs. normals).
(F(l,47) = 56.16, p < .001). The intervals were reliably
longer (ie, heart rate was lower) and the variability was greater DISCUSSION
in the normal Ss, compared with the insomniacs. There was a Several previous studies have shown increased heart rate in
significant main effect for Sleep Stage for both beat-to-beat insomniacs as compared with normals during the night.
interval (F(l,65) = 9.13, p < .001) and the interval SD Monroe (1) reported a nonsignificant 3.9 bpm increase in
(/7(l,69) = 30.57, p < .001). Newman-Keuls pairwise com- insomniacs, Stepanski (3) reported a significant 4.4 bpm
parisons indicated that the interval between beats was shorter increase in insomniacs, and Haynes (2) reported a significant
during wake and REM, which did not differ from each other, 4.6 bpm increase in insomniacs compared with normal sleep-
than in stage I and stage 2, which also did not differ from each ers. These reported differences agree well with the data from
other. Variability was least in stage 2 and increased signifi- the present groups where the overall mean difference in heart
cantly with each stage from stage 2 to REM to stage 1 to wake. period translated to a heart rate difference of 6.9 bpm. In
Only eight of the Insomnia Ss and eight of the Normal Ss general, the previous studies, with the exception of the Haynes
had sufficient SWS for any 5-minute period to be classified as study, which reported EKG only from wake periods, could be
SWS. Of these Ss, only four subject pairs could be formed. A criticized because a comparison of nocturnal heart rates would
t test for interval between beats revealed f3 = 4.515, p = .02 be influenced by the sleep stage composition of the nights;
(mean values of 0.84 vs. 1.095 for insomniacs vs. normals), that is, heart rate is higher when Ss are awake, and insomniacs,
and the t test for variability showed t3 = .095, NS (mean who are awake for longer periods of time, therefore, would be
values of 0.048 vs. 0.067 for insomniacs Vs normals). expected to have higher average heart rates solely on basis of
increased wake time. The present study, which has compared
heart periods within matched EEG time segments, provides
TABLE 1. Demographic Data"
similar results with the sleep stage controlled. It is also
Insomniacs Normals ; Value p Value
unlikely that the heart period differences are driven by the
microstructure of sleep, because insomnia and normal groups
Number 12 12 had similar brief EEG arousal indices in the current study.
Age 31.2(6.8) 29.1 (5.2) 1.900 NS In the current study, traditional heart period mean and SD
Weight (lbs.) 161 (16) 167 (20) -1.593 NS
Subjective latency (min) 62 (25) 18 (10) 5.363 .000
values were calculated from the same time intervals that were
Subjective total sleep (hr) 5.3(1.7) 7.4(1.3) -2.861 .015 used in the spectral analysis to give a better idea of the relative
Length of insomnia (yr) 6.3 sensitivity of the traditional and spectral measures. It is,
therefore, notable that significance levels for all of the
Values are mean (SD). reported variables are similar; that is, if one were to choose

Psychosomatic Medicine 60:610-615 (1998)


TABLE 2. Sleep Parameters"

Normals p Value

EEG parameters
Total sleep (min) 382 (68) 445 (17) -3.537 .005
Stage 1 (%) 15.0 (8.4) 12.9(5.9) 1.092 NS
Stage 2 (%) 42.9(13) 50.2(12) -1.927 .1
Stage 3 (%) 4.6 (3.6) 4.8 (3.8) -0.15 NS
Stage 4 (%) 4.0 (7.4) 5.7 (7.7) -0.65 NS
REM (%) 18.0 (5.0) 20.2 (3.9) -1.488 NS
Wake (min) 69 (57) 27 (13) 2.644 .02
Sleep latency (min) 16.4(15) 6.6 (5.5) 2.108 .1
Sleep efficiency 82 (11) 93 (3.2) -3.721 .003
REM latency (min) 81 (28) 68 (14) 1.301 NS
Arousal index 18.8(14) 19.4(22) -0.121 NS
Subjective report data
Bed time 2421 (156) 2346 (133) 0.733 NS
Wake time 0812 (135) 0748 (132) 0.560 NS
Sleep latency 60 (37) 22 (14) 3.796 .003
Wake 77 (107) 18 (21) 1.831 .1
Total sleep 5.7 (.91) 7.4 (.48) -6.461 .000

' Values are mean (SD).

_ 0.14^

8 0.93-;
a! 0.91-
m :
« 0.89-i
2 0.87
Wake Stage 1 Stage 2 REM
Sleep Stage Wake Stage 1 Stage 2 REM
Sleep Stage
Figure 1. Heart beat-to-beat intervals in insomniacs and normals as
function of sleep stage (the bars represent the SEM). Figure 2. SDs of the mean heart beat-to-beat intervals presented in Figure I
(the bars represent the SEM).

heart period, heart period SD, LFP, or HFP, one would come
to similar conclusions about differences between stages and the collection of the data reported here, and were fed the same
between insomniacs and normals. Many studies over several meals.
years have reported consistent enough differences in heart rate As an entity, insomnia is infrequently viewed as a signif-
as a function of sleep stage that heart rate has even been icant medical problem. However, if insomnia is associated
proposed as a means of determining sleep stage (32). There- with chronic sympathetic hyperactivity, long-term conse-
fore, the reported sleep stage differences are expected. quences of the sympathetic activation associated with the
For spectral analysis parameters, recently we have reported insomnia may exist. Many risk factors for coronary heart
sleep stage differences between wake, stage 1, stage 2, and disease are related to increased sympathetic activity. For
SWS in both LFP and HFP in a group of normal sleepers (33). example, hypertension, elevated plasma insulin (34) and its
The current spectral analysis data reflect similar sleep stage- related decrease in HDL cholesterol, increase in triglyceride
related changes. and cholesterol (35), increased hematocrit (36), decreased
None of these differences between the insomniacs or plasma volume (37), increased plasma thromboglobulin (38),
normals could be based on age, sex, or weight, because increased plasma angiotensin (39), and increased cardiac
subjects were matched carefully on those variables. In addi- arrhythmias (40, 41) are all signs of increased SNS and
tion, the differences could not be based on immediate activi- decreased PNS tone. All of these factors predispose to
ties or diet because subjects remained in the laboratory over coronary disease. It remains to be seen whether any or all of
the 36-hour data collection period, including 24 hours before these factors may be abnormal in patients with chronic

Psychosomatic Medicine 60:610-615 (1998) 613


insomnia (46). Elderly individuals who reported insomnia also

I 1.3. reported both angina pectoris and cardiac arrhythmia (47)
more than twice as often as individuals without insomnia. If
insomnia is a sympathetic disorder predisposing to increased
£1.2- cardiac risk, there could be a direct pathophysiological expla-
nation of the increased mortality in insomniacs and an
1.1- implication that insomnia, like hypertension, could be a silent
and treatable indicator of cardiovascular disease.
Treatments for insomnia, such as relaxation, exercise
training, and some medications, have in common a shift away
"-0.9- from sympathetic nervous system dominance and toward
parasympathetic system activation. As such, these treatments
10.8- may provide long-term benefits over and above short-term
improvement in reported sleep quality. If additional examina-
0.7- tion of patients with chronic insomnia provides evidence of
links to coronary artery disease, more rigorous or prolonged
.9 0.6 treatment of patients by such methods may be indicated.
Wake Stage 1 Stage 2 REM
I Sleep Stage This work was supported by a Merit Review Grant from the
Department of Veterans Affairs and the Sleep-Wake Disorders
Figure 3. Ratio of low-to-high frequency spectral power across sleep stages Research Institute.
in matched groups of insomniac and normal sleepers (the bars represent the
1. Monroe LJ: Psychological and physiological differences be-
-Wt- Insomniacs
tween good and poor sleepers. J Abnorm Psycho! 72:255-264,

3 0.45-
o -•- Normals 2. Haynes SN, Adams A, Franzen M: The effects of presleep stress
on sleep-onset insomnia. J Abnorm Psychol 90:601-606, 1981

0.4- 3. Stepanski EJ, Glinn M, Fortier J, et al: Physiological reactivity in

w 0.35^ / I \
chronic insomnia. Sleep Res 18:306, 1989
4. Haynes SN, Follingstad DR, McGowen WT: Insomnia: Sleep

4 \ patterns and anxiety level. J Psychosom Res 18:69-74, 1974

— 0.3-
/ \
, / \\ \
5. Bonnet MH, Arand DL: 24-Hour metabolic rate in insomniacs
and matched normal sleepers. Sleep 18:581-588, J995

X 0.25-
/ I / 6. Freedman RR, Sattler HL: Physiological and psychological
/ I/ \
\ |
factors in sleep-onset insomnia. J Abnorm Psychol 91:380-389,
0.2- 7. Freedman RR: EEG power spectra in sleep-onset insomnia.
«i 1 jT
rr Electroencephalogr Clin Neurophysiol 63:408-413, 1986
0 15- m 8. Adam K, Tomeny M, Oswald I: Physiological and psychological
Wake Stage 1 Stage 2 REM differences between good and poor sleepers. J Psychiatr Res
Sleep Stage 20:301-316, 1986
9. Johns MW, Gay TJA, Masterton JP, et al: Relationship between
Figure 4. Ratio of high frequency lo total spectral power across sleep stages sleep habits, adrenocortical activity and personality. Psychosom
in matched groups of insomniac and normal sleepers (the bars represent the
SEM). Med 33:499-508, 1971
10. Frankel B, Buchbinder R, Coursey R, et al: Sleep patterns and
psychological test characteristics of chronic primary insomniacs.
insomnia. Furthermore, sleep itself is being increasingly Sleep Res 2:149, 1973
identified as a source of cardiac risk (42). It has been known 11. Mendelson WB, Garnett D, Gillin CG, et al: The experience of
for some time that there is an increased risk of mortality insomnia and daytime and nighttime functioning. Psychiatry Res
associated with short sleep lengths. In the Alameda County 12:235-250, 1984
study (43), the age-adjusted mortality rate for men who 12. Mendelson WB, James SP, Garnett D, et al: A psychophysio-
typically slept less than 6 hours per night and often had trouble logical study of insomnia. Psychiatry Res 19:267-284, 1986
sleeping was 2.5 times the rate for 7 to 8 hour sleepers with no 13. Seidel WF, Ball S, Cohen S, et al: Daytime alertness in relation
sleep problem. Men who died from coronary heart disease to mood, performance, and nocturnal sleep in chronic insomniacs
were more likely to have had reported sleep lengths of less and noncomplaining sleepers. Sleep 7:230-238, 1984
than 4 hours than longer sleep lengths (44), although similar 14. Bonnet MH: Effect of 64 hours of sleep deprivation upon sleep
findings were reported for other diseases such as cancer. in geriatric normals and insomniacs. Neurobiol Aging 7:89-96,
Fatigue was a predictor of myocardial infarction even after 1986
adjustment for the effects of blood pressure, blood sugar, 15. Stepanski E, Zorick F, Roehrs T, et al: Daytime alertness in
antihypertensives, smoking, and age (45). In patients with patients with chronic insomnia compared with asymptomatic
documented myocardial infarction, 39% reported preceding control subjects. Sleep 11:54-60, 1988

614 Psychosomatic Medicine 60:610-615 (1998)


16. Haynes SN, Fitzgerald SG, Shute GE, et al: The utility and 31. Yeragani VK, Berger R, Pohl R, et al: Effects of yohimbine on
validity of daytime naps in the assessment of sleep-onset heart rate variability in panic disorder patients and normal
insomnia. J Behav Med 8:237-247, 1985 controls: A study of power spectral analysis of heart rate.
17. Schneider-Helmert D: Twenty-four-hour sleep-wake function J Cardiovasc Pharmacol 20:609-618, 1992
and personality patterns in chronic insomniacs and healthy 32. Welch AJ, Richardson PC: Computer sleep stage classification
controls. Sleep 10:452-462, 1987 using heart rate data. Electroencephalogr Clin Neurophysiol
18. Mendelson WB, Garnett D, Linnoila M: Do insomniacs have 34:145-152, 1973
impaired daytime functioning? Biol Psychiatry 19:1261-1264, 33. Bonnet MH, Arand DL: Heart rate variability: Sleep stage, time
1984 of night, and arousal effects. Electroencephalogr Clin Neuro-
19. Furlan R, Guzzetti S, Crivellaro W, et al: Continuous 24-hour physiol 102:390-396, 1997
assessment of the neural regulation of systemic arterial pressure 34. Deibert DC, DeFronzo RA: Epinephrine-induced insulin resis-
and RR variabilities in ambulant subjects. Circulation 81:537— tance in man. J Clin Invest 65:717-721, 1980
547, 1990 35. DeFronzo RA, Ferrannini E: Insulin resistance: A multifaceted
20. Jaffe RS, Fung DL, Behrman KH: Optimal frequency ranges for syndrome responsible for NIDDM, obesity, hypertension, dys-
extracting information on autonomic activity from the heart rate lipidemia, and atherosclerotic cardiovascular disease. Diabetes
spectogram. J Auton Nerv Syst 46:37-46, 1993 Care 14:173-194, 1991
21. Yeragani VK, Srinivasan K, Vempati S, et al: Fractal dimension 36. Julius S: Abnormalities of autonomic nervous control in human
of heart rate time series: An effective measure of autonomic hypertension. Cardiovasc Drugs Ther 8:11-20, 1994
function. J Appl Physiol 75:2429-2438, 1993 37. Cohn JN: Relationship of plasma volume changes to resistance
22. Malliani A, Pagani M, Lombardi F, et al: Cardiovascular neural and capacitance vessel effects of sympathomimetic amines and
regulation explored in the frequency domain. Circulation 84: angiotensin in man. Clin Sci 30:267-278, 1966
482-492, 1991 38. Kjeldsen SE, Gjesdal K, Eide I, et al: Increased beta-thrombo-
23. Rizzoni D, Castellano M, Beschi M, et al: Plasma norepineph- globulin in essential hypertension: Interactions between arterial
rine and spectral analysis of the heart rate during cardiopulmo- plasma adrenaline, platelet function and blood lipids. Acta Med
nary receptor stimulation in normal and hypertensive subjects. Scand 213:369-373, 1983
J Hypertens 9 (Suppl 6):S84-S85, 1991 39. Esler M, Julius S, Zweifler A, et al: Mild high-renin essential
24. Bigger JT, Fleiss JL, Steinman AB, et al: Frequency domain hypertension: Neurogenic human hypertension. N Engl J Med
measures of heart period variability and mortality after myocar- 296:405-411, 1977
dial infarction. Circulation 85:164-171, 1992 40. Schwartz PJ, Vanoli E, Stramba-Badiale M, et al: Autonomic
25. Baharav A, Kotagal S, Gibbons V, et al: Fluctuations in mechanisms and sudden death—new insights from analysis of
autonomic nervous activity during sleep displayed by power baroreceptor reflexes in conscious dogs with and without a
spectrum analysis of heart rate variability. Neurology 45:1183— myocardial infarction. Circulation 78:969-979, 1993
1187, 1995 41. Billman GE: Effect of carbachol and cyclic GMP on suscepti-
26. Vaughn BV, Quint SR, Messenheimer JA, et al: Heart period bility to ventricular fibrillation. FASEB 4:1668-1673, 1990
variability in sleep. Electroencephalogr Clin Neurophysiol 94: 42. Verrier RL, Muller JE, Hobson JA: Sleep, dreams, and sudden
155-162, 1995 death: The case for sleep as an autonomic stress test for the heart.
27. Somers VK, Phil D, Dyken ME, et al: Sympathetic-nerve Cardiovasc Res 31:181-211, 1996
activity during sleep in normal subjects. N Engl J Med 328:303- 43. Wingard DL, Berkman LF: Mortality risk associated with
307, 1993 sleeping patterns among adults. Sleep 6:102-107, 1983
28. Hornyak M, Cejnar M, Elam M, et al: Sympathetic muscle nerve 44. Kripke D, Simons R, Garfinkel L, et al: Short and long sleep and
activity during sleep in man. Brain 114:1281-1295, 1991 sleeping pills. Arch Gen Psychiatry 36:103, 1979
29. Carskadon MA: Basics for polygraphic monitoring of sleep. In 45. Appels A, Mulder P: Excess fatigue as a precursor of myocardial
Guilleminault C (ed), Sleeping and waking disorders: Indications infarction. Eur Heart J 9:758-764, 1988
and techniques. Menlo Park, Addison-Wesley, 1982, 1-16 46. Carney RM, Freedland KE, Jaffe AS: Insomnia and depression
30. Bornstein SK: Respiratory monitoring during sleep: Polysom- prior to myocardial infarction. Psychosom Med 52:603-609,
nography. In Guilleminault C (ed), Sleeping and Waking Disor- 1990
ders Indications and Techniques. Menlo Park, Addison-Wesley, 47. Asplund R: Sleep and cardiac diseases amongst elderly people.
1982 J Intern Med 236:65-71, 1994

Psychosomatic Medicine 60:610-615 (1998) 615