Higher heart rate and reduced heart rate variability persist during sleep in
chronic fatigue syndrome: A population-based study☆
Roumiana S. Bonevaa,⁎, Michael J. Deckerb,c, Elizabeth M. Maloneya, Jin-Mann Lina,
James F. Jonesa, Helgi G. Helgasonc, Christine M. Heimd, David B. Ryeb, William C. Reevesa
aChronic Viral Diseases Branch, National Center for Zoonotic, Vector-borne and Enteric Diseases, Centers for Disease Control and Prevention,
Atlanta, GA, USA
bDepartment of Neurology, Emory University School of Medicine, Atlanta, GA, USA
cFusion Sleep, 4265 Johns Creek Parkway, Suite A, Suwanee, Georgia 30024, USA
dDepartment of Psychiatry and Behavioral Sciences, Emory University School of Medicine, Atlanta, GA, USA
Received 22 March 2007; received in revised form 8 August 2007; accepted 20 August 2007
Autonomic nervous system (ANS) dysfunction has been suggested in patients with chronic fatigue syndrome (CFS). In this study, we
sought to determine whether increased heart rate (HR) and reduced heart rate variability (HRV) parameters observed in CFS patients during
wakefulness persist during sleep. To this end, we compared heart rate (HR) and HRVas indicators of ANS function in CFS subjects and non-
fatigued (NF) controls in a population-based, case-control study. Thirty subjects with CFS and 38 NF controls, matched for age-, sex- and
body mass index, were eligible for analysis. Main outcome measures included mean RR interval (RRI), HR, and HRV parameters derived
from overnight ECG. Plasma aldosterone and norepinephrine levels, medicines with cardiovascular effect, and reported physical activity
were examined as covariates. General Linear Models were used to assess significance of associations and adjust for potential confounders.
Compared to controls, CFS cases had significantly higher mean HR (71.4 vs 64.8 bpm), with a shorter mean RRI [840.4 (85.3) vs 925.4
(97.8) ms] (pb0.0004, each), and reduced low frequency (LF), very low frequency (VLF), and total power (TP) of HRV (pb0.02, all). CFS
cases had significantly lower plasma aldosterone (pb0.05), and tended to have higher plasma norepinephrine levels. HR correlated weakly
with plasma norepinephrine (r=0.23, p=0.05) and moderately with vitality and fatigue scores (r=−0.49 and 0.46, respectively, pb0.0001).
Limitation in moderate physical activity was strongly associated with increased HR and decreased HRV. Nevertheless, among 42 subjects
with similar physical activity limitations, CFS cases still had higher HR (71.8 bpm) than respective controls (64.9 bpm), p=0.023,
suggesting that reduced physical activity could not fully explain CFS-associated differences in HR and HRV. After adjusting for potential
confounders case-control differences in HR and TP remained significant (pb0.05). Conclusion: the presence of increased HR and reduced
HRV in CFS during sleep coupled with higher norepinephrine levels and lower plasma aldosterone suggest a state of sympathetic ANS
predominance and neuroendocrine alterations. Future research on the underlying pathophysiologic mechanisms of the association is needed.
Published by Elsevier B.V.
Keywords: Chronic fatigue syndrome; Heart rate; Heart rate variability; Aldosterone; Norepinephrine; Population-based; Case-control; SF-36; Multiple fatigue
Chronic fatigue syndrome (CFS) presents a diagnostic and
management challenge for physicians. A diagnosis of CFS is
derived from self-reported symptoms, which are not explained
by a differential diagnosis of known medical conditions.
Symptoms include clinically evaluated, unexplained, persistent
or relapsing fatigue of at least 6-months' duration, that is not
Autonomic Neuroscience: Basic and Clinical 137 (2007) 94–101
☆Disclaimer: The findings and conclusions in this report are those of the
authors and do not necessarily represent the views of the funding agency.
Names of instruments in this report are for identification purposes only and
are not in any way endorsed by the funding agency. At the time of this study,
both Drs. Decker and Rye were on the Medcare speaker's bureau.
⁎Corresponding author. Mail Stop A-15, Chronic Viral Diseases Branch,
National Center for Zoonotic, Vector-borne and Enteric Diseases, Centers
for Disease Control and Prevention, 1600 Clifton Road, N.E., Atlanta,
GA 30329, USA. Tel.: +1 404 639 3921; fax: +1 404 639 2779.
E-mail address: email@example.com (R.S. Boneva).
1566-0702/$ - see front matter. Published by Elsevier B.V.
substantially alleviated by rest, and results in substantial
reduction in previous levels of activity. In addition, CFS
diagnosis requires the presence of at least 4 of 8 case-defining
significant impairment in memory or concentration, headache,
muscle pain, joint pain, sore throat and tender lymph nodes)
(Fukuda et al., 1994). There are no pathognomonic physical
signs or diagnostic laboratory abnormalities. The pathophysi-
treatment. Current therapy is directed toward relieving
Disorders of the autonomic nervous system (ANS) share
many clinical features of CFS and may accountfor some cases
(Jones et al., 2005). Indexes of heart rate variability (HRV)
derived from a continuous electrocardiographic (ECG)
recording can be used as indicators of the overall output of
the sympathetic and parasympathetic branches of the ANS to
the sino-atrial node of the heart. Several studies have found
to controls (Freeman and Komaroff, 1997; De Becker et al.,
1998; Stewart et al., 1998; Stewart, 2000; LaManca et al.,
1999; Soetekouw etal.,1999; Winkler etal.,2004; Yamamoto
studies on HRV in CFS included patients from referral clinics
and, therefore, have some recruitment bias. Studies examining
heart rate and HRVin CFS in a population-based fashion have
not been conducted. Perhaps of greater significance, all of the
preceding studies were conducted during wakefulness, and
usually under different challenges (standing, head-up tilt test,
forced/paced breathing, treadmill, Valsalva maneuver). There-
induced stress responses, anxiety, etc. cannot be excluded.
Many CFS subjects receive medications (Jones et al., 2003)
that may affect HR and HRV. Existing studies either excluded
subjects receiving cardiovascular- or ANS-acting medications
or, conversely, did not discuss medication use at all and,
therefore, did not control for such. Finally, none of the studies
included long-term ECG recordings at rest or during sleep,
when environmental stressors are removed or assessment of
baseline plasma norepinephrine. In this study, we used a
population-based, case-control study design to determine
whether previously described reduced HRV in CFS subjects
persists during sleep.
We hypothesized that decreased HRV observed in CFS
reflects a perturbation in autonomic function that persists
3.1. Study design
This was a population-based, case-control study, approved
by the Centers for Disease Control and Prevention (CDC) and
Emory University Institutional Review Boards. All subjects
gave informed consent. Two hundred twenty seven adults
from the general population of Wichita, Kansas, who were
identified during a 4-year CFS surveillance study as either
having CFS or being non-fatigued (NF), participated in a 2-
day in-hospital study conducted between January–July 2003.
This study was described in detail elsewhere (Reyes et al.,
2003; Heim et al., 2006; Reeves et al., 2006). Briefly,
participants answered several questionnaires that measured
symptoms, functioning and fatigue [CDC Symptom Invento-
ry, Short Form Health Survey (SF-36), Multidimensional
Fatigue Inventory (MFI)]. They completed the Diagnostic
Interview Schedule (DIS) (Robbins et al., 1995) and
underwent medical history, physical examination, and
laboratory testing, which were all used by a physician panel
to identify conditions exclusionary for CFS.
As a result of these evaluations there were: (i) 58 subjects
who met the 1994 CDC clinical case definition for CFS;
(ii) 55 non-fatigued (NF) controls randomly selected and
sex-, race-, age- and BMI-matched to CFS cases; (iii) 59
persons with medically unexplained fatigue of 6 months or
longer but having “insufficient” number of symptoms to
meet the CFS case definition (called ISF). Other 27 persons
with CFS and 28 persons with ISF who had melancholic
depression, as determined by DIS, were not included in our
analyses. Remaining participants were subsequently reclas-
sified based on cut off values from the SF-36, MFI and the
CDC Symptom Inventory questionnaires, which resulted in
43 who met criteria for CFS, 61 classified as ISF, and 60 NF
controls (Heim et al., 2006; Reeves et al., 2005).
3.2. Data collection
Demographic information, including age, sex, and race/
ethnicity was obtained by telephone interview prior to the
clinic appointment and confirmed at time of exam. Race was
self-selected by subjects from either: White, African Ameri-
can, Native American/Alaskan, Asian/Asian Pacific or mixed,
and weight were measured at clinic. Body mass index (BMI)
was computed as the ratio of weight in kilograms to height in
meters squared. Participants were instructed to bring to clinic
counter). A nurse recorded the name, dose, and time of
administration of every medication as reported by the patient.
Patients continued their medications during the clinic stay.
Blood and urine for standard laboratory and for endocrine
or day 3 ofthe clinicstay(day 1 being the lateafternoon when
patients were admitted). Clinical laboratory tests were
performed in licensed commercial laboratories (Quest Diag-
nostics and Esoterix Inc). Blood for aldosterone and
norepinephrine (NE) was drawn in the morning, upon
95R.S. Boneva et al. / Autonomic Neuroscience: Basic and Clinical 137 (2007) 94–101
wakening. For aldosterone, 1 ml of serum was separated and
EDTA tube and, immediately after collection, centrifuged in a
cold (4 °C) centrifuge. One milliliter of plasma was separated
and frozen immediately. Samples were shipped frozen to
Esoterix Laboratory Services for testing. Aldosterone levels
were measured by radioimmunoassay after solvent extraction.
Norepinephrine plasma levels were determined by high
performance liquid chromatography.
We calculated osmolality by the formula (1.89⁎sodium
[mmol/l])+(1.38⁎potassium [mmol/l])+(1.08⁎glucose [mg/
dl]/18)+(1.03⁎BUN [mg/dl]/28)+7.45, which has been
shown to generate the closest values to measured osmolality
(Bhagat et al., 1984).
Polysomnography (PSG) was performed with the Embla
N-7000 diagnostic system (Medcare, Iceland), which con-
sists of a small (approximately 10×7.5×15 cm) digital
amplifier that interfaces with a standard Pentium-based
computer for real time data collection and review. It has 40-
channel capability with all channels fully configurable for
sensitivities, sampling rates and filters. We employed a
sampling rate of 200 Hz. The system employs a Windows
platform and uses proprietary software (Somnologica). The
Embla N-7000 (Medcare) relies upon standard gold electro-
des for recording of electroencephalogram (EEG), electro-
occulogram (EOG), and electromyogram (EMG) for sleep
staging and appreciation of sleep structure. Respiration was
recorded with standard sensors for airflow, respiratory effort,
body position, and snoring sounds (Pro-Tech, Inc., Woodwin-
ville, WA). A pressure transducer (Pro-tech, model PTAF2)
attached to a cannula worn in the nasal nares was used to aid
in identification of episodes of elevated upper airway
resistance. Built-in oximetry provided continuous co-moni-
toring of arterial oxygen saturation via a standard non-
invasive finger probe. The following signals were recorded:
central (C3-A2//C4-A1) EEG, left and right monopolar EOG,
surface mentalis EMG, ECG (lead II), respiratory airflow
(recorded with both thermistors and pressure transducers),
respiratory effort (rib cage and abdomen), and 4 separate
Impedance at the time of evening hook-up will be less than
5000 Ω on all EMG channels. Subjects had continuous PSG
recordings during night 1 and night 2 (from ∼22:30 h to
7:00 h the following morning). Recorded data were scanned
visually and artifacts were manually removed. ECG record-
ings from the patients' second night were analyzed as the first
night was considered an ‘adaptation night’.
The mean RRI [in milliseconds (ms)] and mean HR [bpm]
(HR=60,000 ms divided by RRI) were calculated from the
mean cycle length of all normal RRI (i.e., abnormal QRS
complexes were excluded from calculations). The following
time domain heart rate variability parameters (Task force,
1996) were derived from the recordings: SDNN — the
standard deviation of the N–N interval (the N–N interval
refers to ‘normal-to-normal’ RRI); SDANN — the standard
deviation of the average N–N intervals calculated over short
(5-minute) periods; SDNN index — the mean of the 5-minute
standard deviations of N–N intervals (calculated over the
whole duration of the recording), which measures the var-
iability due to cycles shorter than 5 min; RMSSD — the
square root of the mean squared differences of successive N–
differing by more than 50 ms; pNN50 — the proportion
derived by dividing NN50 by the total number of all N–N
intervals. The following frequency domain measures were
calculated: power in low frequency range (0.04–0.15 Hz) (LF
[ms2]); power in the very low frequency range (≤0.004 Hz)
(VLF [ms2]); power in the high frequency range (0.15–
0.4 Hz) (HF [ms2]); total power, measured as the variance of
the N–N interval over the temporal segment (TP [ms2]). We
also analyzed HR measurements obtained for 29 of the 30
CFS and all NF subjects using data collected at daytime
baseline, after resting for 30 min in supine position (Jones
et al., 2005).
Standard PSG variables were analyzed previously and
sleep data have been published separately (Reeves et al.,
2006). Briefly, there were no major differences in sleep
architecture between CFS subjects and NF controls with
the exception of a significantly higher mean frequency of
obstructive apnea per hour in persons with CFS. Other
characteristics of sleep architecture did not differ significantly
between persons with CFS and controls (Reeves et al., 2006).
Patients with respiratory disturbance index (RDI) ≥10 events
per hour, indicative of sleep hypopnea/apnea, were not
included in our analyses because said events, per se, elicit
ANS response (Ferini-Strambi et al., 1992).
4. Statistical analysis
We used a t-test to compare continuous variables such as
age, BMI, HR and HRV parameters. Case-control differences
in categorical variables (e.g., sex, medication use) were
analyzed using chi-square or Fisher's exact test. For con-
tinuous variables that did not pass the normality distribution
tests (Kolmogorov–Smirnov and Levin's tests), logarithmic
transformation was used toachieve normal distribution for the
purposes of the analysis. When log transformation did not
result in normal distribution, Wilcoxon rank test was used to
compare values. Spearman correlation coefficients were used
Distribution of demographic characteristics and body mass index (BMI)
among CFS cases and non-fatigued (NF) controls
CharacteristicsCFS (n=30)NF (n=38)p-value
Age in years (mean±SD)
Sex [n (%)]
Race [n (%)]
96 R.S. Boneva et al. / Autonomic Neuroscience: Basic and Clinical 137 (2007) 94–101
to measure correlations of scores from the SF-36 and MFI
subscales with HR and HRV parameters.
independent variable of case (CFS) — control (NF) status. To
rule out spurious associations between CFS and HR/HRV
and to assess effects of other factors, we controlled for several
sets of covariates by including them in the regression model.
These were: (a) age, sex and BMI (the pre-clinical match
factors) — to account for breaking the matched pairs due to
reclassification and exclusions after the clinical evaluation; (b)
medications with cardiovascular effect (antidepressants, anti-
hypertensives (ACE inhibitors and diuretics), beta blockers,
calcium channel blockers, and sympathetomimetics) — to
evaluate possible medication effect on HR and HRV (Task
force, 1996); (c) sleep parameters that were significantly
apnea episodes per hour (Reeves et al., 2006), periodic leg
movements (PLM) with arousal, and number of central apnea
events per hour); (d) plasma baseline levels of norepinephrine
and aldosterone (for 6.5% of subjects with aldosterone values
b1 ng/dl (lower limit of detection), the value of 0.5 was
imputed); and (e) physical activity level — as it can also
influence HR and HRV, because by definition CFS subjects
have reduced physical activity. Therefore, in addition to the
above analyses, we also conducted a separate restricted
analysis of the association between HR/HRV and CFS only
among cases and controls who reported the same level of
moderate physical activity in the physical function section of
Statistical significance of two-tailed tests was determined
by an alpha level of 0.05. SAS Version 9.0, was used to
conduct all analyses (SAS Institute, Cary, N.C).
There were 43 subjects classified as CFS, 61 subjects with
ISF and 60 non-fatigued (NF) controls who did not have
exclusionarymedical or psychiatric conditions as determined
by review of DIS data, medical history, physical exam, and
laboratory test results (see Reyes et al., 2003; Heim et al.,
2006; Reeves et al., 2006). After excluding subjects with
RDI score ≥10 or inadequate ECG recording, there were 30
of 43 (69.8%) CFS cases, 44of 61 (72.1%) subjects with ISF,
and 38 of 60 (63.3%) NF controls eligible for analysis.
Comparisons focus on the 30 CFS cases and 38 NF controls.
CFS subjects and NF controls were similar with respect to
their distributions of age, sex, race and BMI (Table 1). Use of
beta blockers, calcium channel blockers, anti-hypertensive
medications, and antidepressants did not differ significantly
between CFS cases and NF controls (Table 2).
Frequency of medication use among CFS patients and non-fatigued (NF)
Type of medication CFSNFp value
n (%)n (%)n (%)n (%)
aIncludesdiuretics,ACEinhibitors, angiotensinII receptorblockers;does
not include beta blockers or calcium channel blockers.
Distribution of mean values for RR interval, heart rate, and heart rate variability parameters among CFS cases and non-fatigued controls
HRV parameterCFS (n=30)
NF (n=38)p value
Unadjusted Adjusted for age, sex,
Adjusted for age, sex, BMI
and specific medications†
RR interval (ms)
HRV triangular index
⁎SD, standard deviation; The p-value for differences between CFS and NF were derived from general linear models.†Medications: (a) antidepressants; (b) beta
blockers; (c) calcium channel blockers; (ah) anti-hypertensives.
0.98 (a, b)
0.27 (a, b)
0.04 (a, ah)
0.01 (a, c)
97R.S. Boneva et al. / Autonomic Neuroscience: Basic and Clinical 137 (2007) 94–101
We compared HRV parameters between CFS and NF
groups. Although the mean RRI (and heart rate) were within
normal limits for all study subjects, the mean RRI in CFS
patients was significantly shorter (840.4 ms) than in the NF
controls (925.0 ms) (p=0.0004, Table 3). A corollary finding
was that CFS subjects had a higher mean HR (72.1±7.6)
compared to controls (65.5±7.0), (p=0.0004, Table 3).
Similarly, during wakefulness, after a 30-minute period in
recumbent position, CFS cases had a significantly higher
mean HR at baseline (79.2±9.6 bpm), compared to the NF
controls (72.2±8.7, n=36) (p=0.003). Mean values for LF,
VLF, and TP were also significantly lower in cases than
controls (Table 4, p=0.02, p=0.006, p=0.003, respectively).
The remaining HRV parameters were also lower in CFS
subjects than in NF controls, but the differences did not reach
statistical significance (Table 3).
In our study sample, women had significantly lower
mean LF, VLF and TP compared to men (Table 4); HRV
parameters did not vary significantly by age or BMI. To
adjust for potential confounding that may account for the
association between CFS and reduced HRV, we examined
these covariates by including them in the statistical model.
Controlling for sex, age, and BMI did not alter the asso-
ciation of CFS with shorter RRI (i.e., higher HR), LF, VLF,
and TP (Table 3).
Users of antidepressants had significantly faster mean
HR, shorter mean RRI, and lower SDANN, HF power, and
LF power (data not shown, all pb0.05). Users of beta
blockers had significantly lower mean SDANN (p=0.03)
compared to non-users. After controlling for medication use
(antidepressants alone, or antidepressants and an additional
cardiovascular medication, where indicated) CFS remained
significantly associated with higher HR (shorter RRI), and
lower LF, VLF and total power (pb0.01, p=0.04, p=0.01,
pb0.01, respectively) (Table 3).
We further examined whether additional covariates (serum
osmolality, plasma aldosterone and norepinephrine levels)
accountedfor the relationshipbetweenCFS andreduced HRV
parameters. Osmolality did not differ between CFS and NF
controls (283.4±3.7 vs 282.9±2.9 mmol/l, respectively) and
was not considered in further analyses. CFS subjects had
controls (7.07±4.00 ng/dl), p=0.012. Mean norepinephrine
levels in CFS subjects were higher than in NF (449.7±264.4
vs 409.5±210.0, respectively) but not statistically different.
Norepinephrine levels, however, were weakly correlated with
HR, (i.e. higher norepinephrine correlated with higher HR) in
the combined group of CFS, ISF and NF subjects (r=0.23,
Mean values and standard deviations of heart rate variability (HRV)
parameters by sex in the combined group of CFS cases and NF controls
HRV parameterFemale (n=57)Male (n=11)p
Heart rate (bpm)
R–R interval (ms)
Low frequency power (ms2)
Very low frequency power
Total power (ms2)
HRV triangular index
3378.0 (1183.0) 5135.6 (1978.4) b0.001
5999.6 (2525.5) 9133.1 (3007.0) b0.001
3523.0 (3831.8) 4149.0 (3845.7)
1537.4 (1149.3) 1870.2 (1360.3)
16256 (4173.3) b0.001
Correlation of HRV parameters with health, function, and impairment scores from the MFI and SF scales
SF-36 and MFI variablesSF and MFI score, by group Spearman coefficients of correlation between HR and HRV parameters and individual SF
or MFI scores (with respective p value)
SF-36 subscalesCFS, n=30NF, n=38rpr
⁎Correlation coefficients b0.20 and p values N0.11 are marked as NS. Higher scores in SF-36 reflect better health and function; higher MFI scores correspond to
more fatigue or impairment.
98R.S. Boneva et al. / Autonomic Neuroscience: Basic and Clinical 137 (2007) 94–101
p=0.05). In addition, in a linear regression model, HR (and
RRI, respectively) was significantly associated with a
dichotomous (high vs. low) measure of norepinephrine (high
norepinephrine≥410 pg/dl, the laboratory's upper normal
limit), p=0.039. After adjusting for norepinephrine, sex and
TP levels between CFS and NF remained statistically
significant (p=0.0004, p=0.048, p=0.014, p=0.007, respec-
tively). After adjustment for aldosterone, the case-control
difference in HR, RRI, VLF and TP remained statistically
significant (p=0.001, p=0.014, p=0.008, respectively), while
Among the examined sleep parameters, only adjusting for
mean hourly number of periodic leg movements with arousal
diminished the statistical significance of the association of
CFS with lower LF and VLF power (p=0.07 and p=0.09,
HR correlated significantly with all the subscale scores
from the SF-36 (negatively) and the MFI (positively),
indicating that a higher HR was associated with more
impairment and worse fatigue. The correlations of HR with
SF-36 subscales measuring vitality (r=−0.49), general and
physical fatigue, activity reduction (all rN0.40, pb0.005),
bodily pain, and physical function (r=−0.39, p=0.01, for
both) were stronger than the correlation with mental fatigue
(r=−0.24, p=0.049) (Table 5).
In a separate model, we explored the associations between
individual SF-36 subscales measuring physical health, as
independent variables, and each HRV parameter, as the
dependent variable. Limitation in moderate physical activity
remained the only variable significantly associated with HR
and HRV. Limitation in moderate physical activity was
significantly associated with higher HR (shorter RRI)
pb0.001, and with lower LF, VLF and TP (p=0.031,
p=0.006, and p=0.002, respectively). Finally, to establish
whether, after accounting for level of limitation in moderate
physical activity, CFS was still associated with higher HR and
lower HRV, we restricted the analysis to 36 CFS cases and 6
controls who reported being equally limited (i.e., “limited a
lot”) in moderate physical activity. Among this subset of
subjects, patients with CFS still had significantly higher mean
HR compared to controls (71.2 bpm vs. 64.9 bpm), p=0.02,
with shorter RR interval (884.0 ms vs 933.9 ms, p=0.03).
Their mean LF, VLF and TP were also lower compared to
controls but no longer statistically different (Table 6).
In separate general linear models in which we regressed
either CFS or limitation in moderate physical activity on
HRV parameters, each explained a small part of the
variations in HR (17% and 20%, respectively), LF (18%
and 19%, respectively), VLF (11% and 14%, respectively)
and TP (12.5 and 17%, respectively). A model that included
CFS, limitation in moderate physical activity, and sex as
independent variables, explained 23.6%, 27.5%, 28.5% and
35.6% of the variance in HR (RRI), LF, VLF and TP,
respectively (data not shown).
We demonstrate that subjects meeting the CFS case
definition have increased HR with reduced LF, VLF, and TP
components of HRV that persist during sleep. This is
observed in the presence of higher baseline norepinephrine
levels and lower baseline aldosterone levels in the CFS
group compared to controls.
Our novel findings concur with previous reports of
reduced HRV in CFS subjects during wakefulness, under
challenge conditions (De Becker et al., 1998; Stewart et al.,
1998; LaManca et al., 1999; Cordero et al., 1996; Duprez
et al., 1998; Yamamoto et al., 2003). Our findings further
enhance those results by demonstrating that increased HR
and diminished HRV in CFS subjects persist during sleep.
We also demonstrate that the higher HR and lower HRV do
not appear to be due to differential use of medications
(antidepressants, beta blockers, sympathomimetics, calcium
channel blockers) known to alter HRV (Task force, 1996),
nor can be fully accounted for by CFS subjects' decreased
physical activity. In addition, this study is the first to show
that increased HR in CFS subjects correlated with standard-
ized and validated measures of fatigue and impairment (the
MFI and SF-36 instruments).
We acknowledge several limitations of our study. First,
we did not include an objective measure of physical fitness.
By default, CFS cases are limited in their physical activity.
Therefore, our study cannot definitively determine whether
the differences in HR and HRV between CFS subjects and
controls correspond to differences in physical activity and
fitness. This has also been the limitation of other studies of
heart rate variability in CFS and needs to be addressed in the
future. However, in a subset of our sample, we demonstrate
that for equal self-reported limitations in moderate physical
activity CFS subjects still had a significantly higher HR
compared to controls. Similar to the entire sample, in this
subset LF, VLF and TP in CFS cases were still lower
(numerically, though not statistically) than in controls.
Second, we may have introduced bias if the CFS cases
eliminated due to inadequate ECG recordings were more
healthy than the CFS cases who remained in our analysis.
However, we have no reason to suspect that this bias
Heart rate and frequency domain heart rate variability parameters among 42
subjects who reported being “a lot limited” in their moderate physical
CFS (n=36) NF (n=6)
p values are derived from the general linear model.
99 R.S. Boneva et al. / Autonomic Neuroscience: Basic and Clinical 137 (2007) 94–101
operated differently in cases and controls, therefore the effect
on our analysis would be to bias the results toward the null.
Third, small sample sizes may limit our ability to generalize
findings to all CFS patients but we attempted to address this
limitation by describing our sample's demographics and
BMI, demonstrating that the CFS patients in this analysis
were of similar age, sex and BMI as CFS patients described
in other studies.
We appreciate that occult coronary artery disease or heart
disease could contribute to the reduced HRV. All subjects in
our study underwent physical examination and laboratory
tests to rule out fatiguing illnesses with known cause and
subjects with fatigue due to known or suspected heart disease
or other acute or chronic condition were excluded from our
study. In addition, use of cardiovascular medications (as a
surrogate measure of cardiovascular disease) did not differ
significantly between CFS cases and NF controls.
HRV is reduced in a number of psychiatric conditions
including depression (Birkhofer et al., 2005), which is a
relatively common co-morbid condition in CFS. Subjects with
major depression with melancholic features were excluded
from the study based on detailed psychiatric interview, while
those without melancholic features were not. Even after
and increased HR with reduced HRV remained significant,
which demonstrates that CFS, per se, was associated with
lower HRVand increased HR.
It is plausible that the increased HR we observed in CFS
may reflect a homeostatic attempt to maintain adequate
cardiac output in the presence of a lower blood volume. At
least one study has reported lower blood volume in CFS
patients (Farquhar et al., 2002). While we did not measure
blood volume directly, we were able to rule out hypertonic
dehydration (hyperosmolality) in CFS by demonstrating that
mean osmolality in CFS subjects and controls were similar.
CFS cases in our study had a lower mean aldosterone level
than controls. This may be a side effect of nonsteroid anti-
inflammatory drug use (Zawada, 1982), which is common
among CFS subjects. Regardless, the mechanism underlying
lower aldosterone levels, observed in our CFS subjects,
remains unexplained at present and deserves further attention.
CFS was associated with higher norepinephrine levels
and there was a weak but significant correlation between
high norepinephrine levels and higher HR in our study.
Higher NE levels have been found to correlate negatively
with low frequency power (Kurata et al., 1997) and TP
(Eller, 2007) and our study is in agreement with those
findings as our CFS subjects had reduced LF and TP. Plasma
norepinephrine levels correlate positively with HR, espe-
cially under stress conditions when the sympathetic nervous
system dominates. Thus, higher norepinephrine in conjunc-
tion with increased HR and reduced HRV in CFS patients
may reflect an increased sympathetic output to the heart (or,
alternatively, a decreased norepinephrine clearance). Anti-
depressants, which were more frequently used by CFS
subjects, may also increase plasma norepinephrine levels
(Veith et al., 1983). However, adjusting for antidepressant
use did not change the association between CFS and HR/
HRV substantially. Higher plasma norepinephrine can also
be found in congestive heart failure (Esler and Kaye, 1998)
but, as already discussed, subjects were screened for
fatiguing illnesses with known cause at physical exam and
heart failure was an exclusionary condition.
LF power, VLF, and TP were all significantly lower in
CFS cases compared to NF controls. When we included
periodic leg movement with arousal as a covariate in the
model, the association of CFS with lower LF and VLF
became statistically insignificant (p=0.07 and p=0.09,
respectively). VLF is thought to reflect, among other things,
state of arousal (Taylor et al., 1998). Therefore, it is possible
that the reduced LF and VLF domains of HRV in CFS cases
may be partially due to more arousals episodes during sleep.
The fact that the significance of the association between
CFS and LF power diminished after including aldosterone in
the model implies that plasma aldosterone levels may
influence LF power, possibly by regulation of blood volume.
Because CFS, reduced physical activity, and sex explained
only less than or approximately a third of the variance in HR
and the frequency domains of HRV, there may be additional
factors associated with CFS and with reduced HRV. Eller
(2007) found that reduced high frequency and total power
components of heart rate variability are associated with risk
factors such as waist-to-hip ratio, glycosylated hemoglobin,
and norepinephrine (Eller, 2007). Compared to their non-
(one component of which included waist-to-hip ratio)
(Maloney et al., 2006). BMI is known to be highly correlated
with waist-to-hip ratio. After we controlled for BMI in our
analyses, CFS remained significantly associated with higher
HR and lower HRV. It may be worth it, in a future study, to
further examine whether the higher HR and the lower HRVin
CFS are associated with higher glycosylated hemoglobin and
other features of metabolic syndrome.
Heart rate variability declines with age (Bonnemeier et al.,
2003). With their higher mean HR and lower heart rate
older’ than their age-, sex-, race-, and BMI-matched controls.
A large, prospective, population-based study of middle-
aged men and women (Dekker et al., 2000), found that
high heart rate and, especially, low HRV at baseline were
predictive of coronary heart disease and increased mor-
tality rates (cardiovascular and all causes) at follow-up. For
HRV, this relation could not be attributed to cardiovascular
risk factors or to underlying disease suggesting that low
HRV is an indicator of poor general health (Dekker et al.,
2000). Whether, the CFS subjects' higher HR with lower
HRV places them at increased risk of cardiovascular (and
other) morbidity and mortality remains to be seen but at
least one study suggests that such risk may exist (Jason
et al., 2006).
HR and HRV measure the overall input of the sympathetic
and parasympathetic branches of the autonomic nervous
100R.S. Boneva et al. / Autonomic Neuroscience: Basic and Clinical 137 (2007) 94–101
system to the sino-atrial node of the heart. HR and its var-
iability are viewed as a quantifiable measure of the subject's
“efforts” in response to environmental demands (Pagani and
appear to experience greater physiologic effort at rest (during
sleep) than NF controls.
In summary, our novel observations of higher HR and
lower HRV during sleep in CFS, coupled with higher
baseline plasma norepinephrine and lower aldosterone
suggest a state of sympathetic ANS predominance with
perturbed neuroendocrine activity. Future research is needed
to understand the underlying pathophysiologic mechanism
of this dysfunction.
Afari, N., Buchwald, D., 2003. Chronic fatigue syndrome: a review. Am. J.
Psychiatr. 160, 221–236.
Bhagat, C.I., Garcia-Webb, Fletcher, E., Beilby, J.P., 1984. Calculated vs
measured plasma osmolalities revisited. Clin. Chem. 30, 1703–1705.
Birkhofer, A., Schmidt, G., Forstl, H., 2005. Heart and brain — the
influence of psychiatric disorders and their therapy on the heart rate
variability. Fortschr. Neurol. Psychiatr. 73, 192–205.
Bonnemeier, H., Richardt, G., Potratz, J., et al., 2003. Circadian profile of
cardiac autonomic nervous modulation in healthy subjects: differing
effects of aging and gender on heart rate variability. J. Cardiovasc.
Electrophysiol. 14, 791–799.
Cordero, D.L., Sisto, S.A., Tapp, W.N., et al., 1996. Decreased vagal power
during treadmill walking in patients with chronic fatigue syndrome.
Clin. Auton. Res. 6, 329–333.
De Becker, P., Dendale, P., De Meirleir, K., et al., 1998. Autonomic testing
in patients with chronic fatigue syndrome. Am. J. Med. 105, 22S–26S.
Dekker, J.M., Crow, R.S., Folsom, A.R., Hannan, P.J., Liao, D., Swenne,
C.A., Schouten, E.G., 2000. Low heart rate variability in a 2-minute
rhythm strip predicts risk of coronary heart disease and mortality from
several causes: the ARIC Study. Atherosclerosis risk in communities.
Circulation 102 (11), 1239–1244 (Sep 12).
Duprez, D.A., De Buyzere, M.L., Drieghe, B., et al., 1998. Long- and short-
term blood pressure and RR-interval variability and psychosomatic
distress in chronic fatigue syndrome. Clin. Sci. (Lond). 94, 57–63.
Eller, N.H., 2007. Total power and high frequency components of heart rate
variability and risk factors for atherosclerosis. Auton. Neurosci. Basic
Clin. 131, 123–130.
Esler, M., Kaye, D., 1998. Increased sympathetic nervous system activity
and its therapeutic reduction in arterial hypertension, portal hypertension
and heart failure. J. Auton. Nerv. Syst. 72, 210–219.
Farquhar, W.B., Hunt, B.E., Taylor, J.A., et al., 2002. Blood volume and its
relation to peak O(2) consumption and physical activity in patients with
chronic fatigue. Am. J. Physiol. Heart Circ. Physiol. 282, H66–H71.
Ferini-Strambi, L., Zucconi, M., Oldani, A., Smirne, S., 1992. Heart rate
variability during sleep in snorers with and without obstructive sleep
apnea. Chest 102, 1023–1027.
Freeman, R., Komaroff, A., 1997. Does the chronic fatigue syndrome
involve the autonomic nervous system? Am. J. Med. 102, 357–364.
fatigue syndrome study group. Ann. Intern. Med. 121, 953–959.
Heim, C., Wagner, D., Maloney, E.M., et al., 2006. Early adverse experience
and risk for chronic fatigue syndrome: results from a population-based
study. Arch. Gen. Psychiatry 63, 1258–1266.
Jason, L.A., Corradi, K., Gress, S., Williams, S., Torres-Harding, S., 2006.
Causes of death among patients with chronic fatigue syndrome. Health
Care Women Int. 27, 615–626.
Jones,J.F.,Nicholson, A.,Nisenbaum, R.,etal., 2005.Orthostatic instability
in persons with chronic fatigue syndrome and healthy control subjects in
the general population. Am. J. Med. 118 , 1415.e19–1415.e28.
Jones, J.F., Nisenbaum, R., Reeves, W.C., 2003. Medication use by persons
with chronic fatigue syndrome: results of a randomized telephone survey
in Wichita, Kansas. Health Qual. Life Outcomes 1, 74. doi:10.1186/
Kurata, C., Shouda, S., Mikami, T., et al., 1997. Comparison of [123I]
metaiodobenzylguanidine kinetics with heart rate variability and plasma
norepinephrine level. J. Nucl. Cardiol. 4, 515–523.
LaManca, J.J., Peckerman, A., Walker, J., et al., 1999. Cardiovascular
response during head-up tilt in chronic fatigue syndrome. Clin. Physiol.
Maloney, E.M., Gurbaxani, B.M., Jones, J.F., de Souza Coelho, L.,
Pennachin, C., Goertzel, B.N., 2006. Chronic fatigue syndrome and high
allostatic load. Pharmacogenomics 7, 467–473.
on the autonomic nervous system. Clin. Sci. (Lond). 96, 117–125.
Reeves, W.C., Wagner, D., Nisenbaum, R., et al., 2005. Chronic fatigue
syndrome—a clinically empirical approach to its definition and study.
BMC Med. 3, 19.
Reeves, W.C., Heim, C., Maloney, E., et al., 2006. Sleep characteristics of
persons withchronicfatiguesyndromeandnon-fatigued controls:results
from a population-based study. BMC Neurol. 6, 41.
Reyes, M., Nisenbaum, R., Hoaglin, D.C., et al., 2003. Prevalence and
incidence of chronic fatigue syndrome in Wichita, Kansas. Arch. Intern.
Med. 163, 1530–1536.
Schedule for DSM-IV (DIS-IV). Washington University, St. Louis, MO.
Soetekouw, P.M., Lenders, J.W., Bleijenberg, G., et al., 1999. Autonomic
function in patients with chronic fatigue syndrome. Clin. Auton. Res. 9,
Stewart, J., Weldon, A., Arlievsky, N., et al., 1998. Neurally mediated
hypotension and autonomic dysfunction measured by heart rate
variability during head-up tilt testing in children with chronic fatigue
syndrome. Clin. Auton. Res. 8, 221–230.
Stewart, J.M., 2000. Autonomic nervous system dysfunction in adolescents
with postural orthostatic tachycardia syndrome and chronic fatigue
syndromeis characterizedby attenuatedvagal baroreflexand potentiated
sympathetic vasomotion. Pediatr. Res. 48, 218–226.
Task Force of the European Society of Cardiology and the North American
Society of Pacing and Electrophysiology, 1996. Heart rate variability.
Standards of measurement, physiological interpretation, and clinical use.
Eur. Heart J. 17, 354–381.
Taylor, J.A., Carr, D.L., Myers, C.W., Eckberg, D.L., 1998. Mechanisms
underlying very-low-frequency RR-interval oscillations in humans.
Circulation 98, 547–555.
Veith,R.C., Raskind,M.A.,Barnes, R.F., Gumbrecht, G., Ritchie,J.L., Halter,
J.B., 1983. Tricyclic antidepressants and supine, standing, and exercise
plasma norepinephrine levels. Clin. Pharmacol. Ther. 33, 763–769.
Winkler, A.S., Blair, D., Marsden, J.T., et al., 2004. Autonomic function and
serum erythropoietin levels in chronic fatigue syndrome. J. Psychosom.
Res. 56, 179–183.
Yamamoto, Y., LaManca, J.J., Natelson, B.H., 2003. A measure of heart rate
variability is sensitive to orthostatic challenge in women with chronic
fatigue syndrome. Exp. Biol. Med. 228, 167–174.
Yoshiuchi, K., Quigley, K.S., Ohashi, K., et al., 2004. Use of time-frequency
analysis to investigate temporal patterns of cardiac autonomic response
during head-up tilt in chronic fatigue syndrome. Auton. Neurosci. 113,
Zawada Jr., E.T., 1982. Renal consequences of nonsteroidal antiinflamma-
tory drugs. Postgrad. Med. 71, 223–230.
101 R.S. Boneva et al. / Autonomic Neuroscience: Basic and Clinical 137 (2007) 94–101