Sex effects on heart rate variability in fibromyalgia and Gulf War illness




To investigate autonomic abnormalities in male and female fibromyalgia (FM) and Gulf War illness (GWI) patients by comparing heart rate variability (HRV) with that of age- and sex-matched healthy controls.


Subjects included 26 (19 women, 7 men) with FM, 11 (6 men, 5 women) with GWI, and 36 (18 men,18 women) healthy controls. HRV was determined from Holter recordings obtained in the Clinical Research Center. Analysis of variance compared 24-hour, daytime, and nighttime HRV by sex within groups and by group within sex.


In women with FM or GWI, HRV was significantly lower than in men with FM or GWI. HRV was similar in male and female controls. When HRV was compared by group within sex, HRV was significantly decreased in women with FM or GWI and no significant differences were seen for men with these conditions.


Decreased HRV in FM and GWI appears to be sex dependent. Results suggest that different mechanisms may be operative in symptom expression in men and women with this spectrum of illness.


Fibromyalgia syndrome (FM) is an incompletely understood rheumatic disorder estimated to affect 1–4% of the general population in industrialized countries (1). Patients with FM experience chronic widespread pain, as well as a number of other symptoms. Most investigators in the field view fibromyalgia as the result of aberrant functioning of the central nervous system, especially systems involving pain processing (2). There are also abnormalities that have been identified in elements of the human stress response, including both the hypothalamic–pituitary–adrenal (HPA) axis and the autonomic nervous system (3, 4).

A number of other conditions overlap with FM, including such systemic conditions as chronic fatigue syndrome and such regional syndromes as irritable bowel syndrome, temporomandibular disorder, etc. There is significant evidence that many of these disorders are characterized by the same underlying mechanisms operative in FM, leading to the suggestion that a whole host of conditions seen commonly in the population may in fact represent different portions of a continuum. This spectrum of illness, termed chronic multisymptom illnesses (CMI) by the Centers for Disease Control and Prevention, may also include the set of symptoms experienced by certain individuals after returning from deployment to the 1991 Gulf War (5–9). However, very few mechanistic studies have been performed to determine whether the ill Gulf War veterans may also demonstrate some of the same abnormalities in pain processing, HPA, and autonomic function as seen in people with FM.

Analysis of the fluctuations in heart rate (heart rate variability [HRV]) provide a window into the autonomic nervous system and can be used to test the hypothesis that autonomic function is abnormal in CMI patients. HRV can be calculated in both the time domain using statistical methods and in the frequency domain using fast Fourier transforms or autoregressive techniques. Different HRV indices provide different perspectives on autonomic function, and abnormalities in HRV reflect abnormalities in cardiac autonomic nervous system function. Prior studies have documented abnormalities in HRV among FM patients, but there is a lack of data on HRV in men with this syndrome and no data on HRV in patients with Gulf War illness (GWI) (10–12).

In the current study, we tested whether autonomic function was similarly abnormal in male and female CMI patients by comparing HRV in male and female patients with FM or GWI and controls, all of approximately the same age. In addition to including these different cohorts, another strength of this study is that subjects spent the entire Holter monitoring period in the clinical research center, performing the identical set of tests and procedures.



Participants in this study (n = 81) were admitted to the Georgetown University General Clinical Research Center (GCRC) for up to 2 days of testing. The protocol and consent form were reviewed and approved by the Georgetown University Institutional Review Board. All subjects provided written informed consent prior to participation. While in the GCRC, subjects participated in a number of research activities. Table 1 is a list of those activities and the approximate times of day associated with them. Holter recordings were made during the first 24 hours of the stay. Subjects with FM (n = 29, 8 men, 21 women, aged 40 ± 9 years, range 24–58 years) and GWI (n = 13, 8 men, 5 women, aged 43 ± 7 years, range 32–53 years) were recruited from the rheumatology clinics at Georgetown University and from outside referrals. At the time of the study, subjects were required to have a diagnosis of FM in accordance with the American College of Rheumatology criteria (13). Veterans were considered to have GWI if they were deployed to the Gulf War theater of operations between August 1990 and August 1991 and reported at least 2 of the following 3 symptoms beginning after August 1990: 1) fatigue that limited usual activity; 2) musculoskeletal pain involving 2 or more regions of the body; and 3) cognitive symptoms including memory, concentration, or attention difficulties (14). Additionally, these symptoms must have been present at the time of screening. Healthy controls (n = 39, 20 men, 19 women, aged 37 ± 9 years, range 23–59 years) without any chronic somatic symptoms were recruited from the community.

Table 1. Activities in which subjects participated over the 2-day time period in the GCRC*
  • *

    GCRC = General Clinical Research Center; IV = intravenous.

Day 0Arrive at GCRC
 6:00 PMQuestionnaires, familiarization
 8:00 PMHolter monitor placed and IV catheter inserted
Day 1 
 6:30 AMPreawakening blood sample
 8:00 AMBlood collection
 8:30 AMPain testing (90 minutes)
 10:00 AM30-minute rest and recovery period
 10:30 AMCognitive testing (45 minutes)
 12:00 PMLunch
 2:30 PMHandgrip test
 3:30 PMExercise testing
 5:00 PMRemove Holter monitor, draw blood samples, discharge subject

Besides meeting the case criteria for FM or GWI, subjects and controls had to meet specific inclusion and exclusion criteria. Age at the time of participation was limited to 18–60 years. Certain medications, including any antidepressant, tricyclic compound, benzodiazepine, antiinflammatory, or antipsychotic medication, could not have been taken within 2 weeks of the study. Individuals with disorders, other than FM or GWI, known to affect the HPA axes, the autonomic nervous system, or pain nociception were also excluded. These disorders included current substance abuse, a weight of 30% heavier or 20% lighter than predicted by standard body mass index tables, hypertension, diabetes, and known coronary artery disease. Individuals were encouraged to refrain from smoking prior to and during their study participation, and caffeine consumption was prohibited 3 days prior to and during their study participation; however, no objective measures of caffeine or tobacco consumption were made.

Electrocardiogram analysis.

Continuous electrocardiogram recordings were obtained for 24 hours using Holter recorders (ACS Holter, Ontario, CA). Subjects and controls performed identical activities during that period. Recordings were analyzed at the Washington University School of Medicine HRV Laboratory using a GE-Marquette MARS 8000 Holter analyzer (GE Medical Systems, Milwaukee, WI). Recordings were considered acceptable if they had at least 12 hours of usable data (i.e., normal-to-normal interbeat intervals), with at least 6 hours of daytime and 3 hours of nighttime data present. Usable data were defined as 5-minute segments in which at least 80% of intervals were scored as normal-to-normal (N-N) intervals. Eight recordings were eliminated (2 GWI, 3 FM, and 3 controls), leaving 73 eligible recordings. Reasons for elimination included excessive signal-to-noise ratio associated with an inadequate hook up or missing data due to recorder failure. Beat-stream files with the time and classification of each QRS complex were transferred to a Sun computer (Sun Microsystems, Mountain View, CA) for HRV analysis. Indices of HRV were calculated for 24 hours, daytime (8:00 AM–8:00 PM), and nighttime (midnight–6:00 AM) periods.

HRV analyses.

Time domain indices of HRV are derived from statistical calculations performed on the set of N-N interbeat intervals. Frequency domain analysis partitions the variance in the heart rate (actually heart period or N-N intervals) signal into its underlying frequency components. The time and frequency domain indices examined in the current study are listed in Table 2. As can be seen in Table 2, HRV indices can be categorized according to the period over which they are assessed. Thus, longer-term HRV indices are those that primarily quantify HRV over periods >5 minutes. They are predominantly influenced by circadian rhythms and also by sustained periods of activity. Intermediate-term indices are those that quantify HRV over periods <5 minutes averaged over the entire recording period. These quantify a combination of sympathetic and parasympathetic influences on heart rate and may include thermoregulation and baroreceptor activity as well as the effect of daily activities (15, 16). Short-term HRV indices describe beat-by-beat changes in heart rate and reflect primarily parasympathetic influences (17). Ratio indices are derived from combinations of HRV indices and have been suggested to reflect autonomic balance (18). The methods used for frequency domain analysis have been previously described in detail (19). Frequency domain indices of HRV have a highly skewed distribution. As is customary, logarithmic transformation of the frequency domain indices of HRV was performed to provide an approximately normal distribution for the purpose of statistical analysis.

Table 2. Definitions for selected time and frequency domain heart rate variability indices
Long-term time domain 
 SDNNStandard deviation of normal-to-normal (N-N) intervals
Intermediate-term time domain 
 SDNNIDXAverage of the standard deviations of N-N intervals over 5 minutes
Long-term frequency domain 
 Ultra low frequency (ULF)Ultra low frequency power (1.15 × 10−5–0.003 Hz)
Intermediate-term frequency domain 
 Very low frequency (VLF)Very low frequency spectral power (0.0033–0.04 Hz)
 Low frequency (LF)Low frequency spectral power (0.04–0.15 Hz)
Short-term time domain 
 rMSSDRoot mean square of the differences between successive N-N intervals
 pNN50Percentage of N-N intervals >50 msec different from the prior interval
Short-term frequency domain 
 High frequency (HF)High frequency spectral power (0.15–0.4 Hz)
 Normalized LFLow frequency power divided by (total power − very low frequency power)
 LF/HF ratioLow to high frequency power ratio averaged for every 5 minutes

Pain tolerance.

A Chatillion dolorimeter was used to determine pain tolerance. The pressure (in kg/cm2) necessary to produce discomfort (pain threshold) and unbearable pain (pain tolerance) at 18 designated tender points and 4 control points was recorded. This testing produced 5 variables for each subject: tender point threshold and tolerance, control point threshold and tolerance, and tender point count.

Psychological measures.

The Medical Outcomes Study Short Form 36 health survey (SF-36) was used as a self-report measure of functional health status (20). Eight domains were assessed: limitations in physical activities due to health problems, limitations in social activities due to physical or emotional problems, limitations in usual role activities due to physical health problems, role limitations due to personal or emotional problems, general mental health (psychological distress and wellbeing), vitality (energy/fatigue), bodily pain, and general health perception. In addition, there is a question about change in health status during the prior year. Each domain is measured on a scale of 0–100.

The Center for Epidemiologic Studies Depression Scale (CES-D) (21) is a 20-item self-report questionnaire that assesses symptoms of depression in nonpsychiatric adults. This instrument possesses strong psychometric properties and has demonstrated strong associations with other measures of depressive symptoms.

The State Trait Personality Inventory (STPI) measures the presence and severity of current symptoms of anger and anxiety (22).


HRV indices were tested for normality and natural-log transformed as necessary to permit parametric statistical analyses. Analysis of variance (ANOVA) with least significant difference post-hoc testing compared self-report measures, pain tolerance, and HRV by study group (FM, GWI, controls) and by study group within sex. Self-report measures, pain tolerance, and HRV between sexes for each group were compared with a t-test. The significance level for all comparisons and correlations was set at 0.05. P values reported are sometimes nominal and readers should interpret them as such. The Statistical Package for the Social Sciences (SPSS Version 10.1, Chicago, IL) was used for data analysis.


Table 3 lists some of the psychosocial and pain measures obtained in this study. The number in each cohort includes only subjects with eligible HRV data. For each variable, however, the number of subjects may have differed slightly due to missing data points for that test. ANOVAs by group were highly significant for the following: all measures of the SF-36 (only physical function subscale shown), STPI trait anxiety and anger, depression, and pain. For most measures, differences between the CMI groups and controls were highly significant, but differences between GWI and FM were not. Exceptions were SF-36 mental health (P = 0.014), STPI trait anger (P = 0.030), and CES-D score (P = 0.028), all of which were significantly worse in the GWI group. Neither total exercise time, i.e., total times during the submaximal bicycle test performed on day 1, nor peak watts generated during the test (also shown in Table 3) were different between groups. This suggests that despite differences in symptoms, patient and control groups were reasonably comparable in fitness.

Table 3. Comparison of self-report scores and dolorimeter data for GWI, FM, and control groups*
M (n = 5)F (n = 5)PM (n = 7)F (n = 18)PM (n = 17)F (n = 17)P
  • *

    P values are for t-tests by sex within group. GWI = Gulf War illness; FM = fibromyalgia; SF-36 = Short Form 36 health survey; STPI = State Trait Personality Inventory; CES-D = Center for Epidemiological Studies Depression Scale.

  • Denotes significant difference (P < 0.05) for between-group comparisons of male GWI versus male control.

  • Denotes significant difference (P < 0.05) for between-group comparisons of female GWI versus female control.

  • §

    Denotes significant difference (P < 0.05) for between-group comparisons of male FM versus male control.

  • Denotes significant difference (P < 0.05) for between-group comparisons of female FM versus female control.

Pain rating3.1 ± 1.27.7 ± 1.80.0024.0 ± 2.5§5.4 ± 2.80.2950.9 ± 1.4§0.2 ± 0.50.102
SF-36 physical component score41 ± 1030 ± 70.11136 ± 10§34 ± 100.62555 ± 6§57 ± 60.427
STPI trait anger22 ± 919 ± 70.57316 ± 516 ± 50.79516 ± 515 ± 50.810
STPI trait anxiety22 ± 822 ± 90.94117 ± 420 ± 60.29915 ± 516 ± 40.520
CES-D17 ± 1229 ± 150.23213 ± 6§17 ± 110.3925 ± 5§6 ± 60.661
No. tender points6.6 ± 6.312.6 ± 5.30.14112.4 ± 5.1§14.3 ± 3.60.3193.6 ± 4.6§2.7 ± 3.90.552
Exercise; peak Watts120.9 ± 29.2106.3 ± 12.50.380125.0 ± 27.494.1 ± 28.70.032142.2 ± 43.5114.7 ± 280.038
Exercise; total test time (seconds)724.3 ± 187.8673.8 ± 108.80.643634.0 ± 180.6627.8 ± 157.10.938746.4 ± 231.2591.7 ± 1140.028

Comparisons for psychosocial and pain measures by sex within group are also shown in Table 3. Although mean values for the SF-36 tended to be higher among men, sex differences were not significant within groups. However, a general pattern was seen of worse values being found in the female GWI patients for pain and for the SF-36. Depression scores were also highest among the female GWI subjects.

Comparison of HRV between FM, GWI, and control groups.

Table 4 compares 24-hour time and frequency domain HRV for all of the GWI and FM patients and healthy controls with recordings eligible for 24-hour HRV analysis. In general, controls had the highest HRV, FM patients had intermediate HRV, and GWI patients had the lowest mean HRV, but ANOVAs were significant only for the reduction in short-term vagally modulated HRV among the GWI subjects.

Table 4. Comparison of 24-hour HRV (mean ± SD) between all FM, all GWI, and all controls*
 All FM (n = 27)All GWI (n = 12)All controls (n = 36)P
  • *

    P values are for analysis of variance. HRV = heart rate variability; FM = fibromyalgia; GWI = Gulf War illness; HR = heart rate; bpm = beats per minute; SDNN = standard deviation of normal-to-normal intervals; Ln = natural logarithm; ULF = ultra low frequency; SDNNDIX = average of the standard deviations of normal-to-normal intervals over 5 minutes; VLF = very low frequency; LF = low frequency; HF = high frequency; pNN50 = percentage of normal-to-normal intervals>50 msec different from the prior interval; rMSSD = root mean square of the differences between successive normal-to-normal intervals.

Longer-term HRV    
 HR (bpm)74 ± 976 ± 970 ± 90.065
 SDNN (msec)125 ± 29117 ± 28140 ± 350.054
 Ln ULF power9.24 ± 0.449.19 ± 0.539.41 ± 0.600.109
Intermediate-term HRV    
 SDNNDIX (msec)63 ± 1753 ± 1768 ± 210.056
 Ln VLF power7.53 ± 0.527.19 ± 0.757.65 ± 0.540.054
 Ln LF power6.93 ± 0.666.44 ± 0.897.00 ± 0.620.050
Ratio-based HRV    
 Normalized LF power (%)55 ± 556 ± 854 ± 80.702
 LF/HF ratio3.9 ± 1.55.1 ± 2.44.0 ± 2.40.211
Short-term HRV    
 pNN50 (%)12.6 ± 9.76.7 ± 5.817.5 ± 14.50.023
 rMSSD (msec)37 ± 1427 ± 1043 ± 220.036
 Ln HF power5.86 ± 0.945.22 ± 0.976.05 ± 1.000.044

Comparison of HRV by sex within groups.

Tables 5, 6, and 7 compare 24-hour, daytime, and nighttime HRV between men and women in each of the groups. To ensure comparability of the analyses, only recordings eligible for HRV analyses during every time period were included. This series of tables shows that the reason that many of the differences in Table 4 were not significant when both men and women were compared together is that the differences in HRV were primarily seen in women.

Table 5. Comparison of 24-hour HRV indices by sex for GWI, FM, and control groups*
M (n = 6)F (n = 5)PM (n = 7)F (n = 19)PM (n = 18)F (n = 18)P
  • *

    For abbreviation definitions, see Table 4.

  • GWI and FM significantly different from controls.

  • Significant difference between GWI and FM.

Longer-term HRV         
 HR (bpm)73 ± 1078 ± 60.36771 ± 876 ± 80.10672 ± 1068 ± 70.142
 SDNN (msec)138 ± 897 ± 280.029149 ± 28116 ± 250.008142 ± 38138 ± 330.691
 Ln ULF9.56 ± 0.148.82 ± 0.570.0429.59 ± 0.429.12 ± 0.400.0159.56 ± 0.529.37 ± 0.490.280
Intermediate-term HRV         
 SDNNDIX (msec)64 ± 1341 ± 120.01479 ± 1357 ± 160.00467 ± 2269 ± 200.791
 Ln VLF7.66 ± 0.466.74 ± 0.760.0347.91 ± 0.447.39 ± 0.490.0237.64 ± 0.597.66 ± 0.490.917
 Ln LF7.04 ± 0.435.77 ± 0.920.0147.51 ± 0.346.69 ± 0.630.0047.04 ± 0.636.96 ± 0.620.718
Ratio-based HRV         
 Normalized LF (%)61 ± 349 ± 70.00456 ± 755 ± 50.48058 ± 751 ± 80.011
 LF/HF ratio5.5 ± 1.93.6 ± 1.10.0873.8 ± 2.04.0 ± 1.50.8284.8 ± 2.83.2 ± 1.50.041
Short-term HRV         
 pNN50 (%)9.8 ± 6.53.8 ± 3.00.08018.9 ± 6.110.1 ± 10.00.03815.5 ± 14.719.5 ± 14.50.422
 rMSSD (msec)33 ± 1122 ± 70.06148 ± 1732 ± 140.01641 ± 2245 ± 220.601
 Ln HF5.74 ± 0.674.86 ± 1.020.1196.61 ± 0.605.56 ± 0.920.0105.9 ± 1.16.2 ± 0.90.250
Table 6. Comparison of daytime HRV indices (mean ± SD) by sex for GWI, FM, and control groups*
M (n = 6)F (n = 5)PM (n = 7)F (n = 19)PM (n = 19)F (n = 18)P
  • *

    For abbreviation definitions, see Table 4.

  • GWI and FM significantly different from controls.

  • Significant difference between GWI and FM.

Longer-term HRV         
 HR (bpm)79 ± 1182 ± 60.70575 ± 1081 ± 80.12778 ± 1172 ± 70.068
 SDNN (msec)109 ± 2489 ± 220.185134 ± 29105 ± 240.019117 ± 41133 ± 320.198
 Ln ULF9.33 ± 0.408.86 ± 0.520.1909.34 ± 0.418.84 ± 0.440.0158.87 ± 0.639.30 ± 0.480.079
Intermediate-term HRV         
 SDNNDIX (msec)57 ± 1442 ± 130.08074 ± 1457 ± 160.01963 ± 2269 ± 180.338
 Ln VLF7.34 ± 0.556.67 ± 0.780.1307.79 ± 0.487.28 ± 0.570.0437.42 ± 0.677.63 ± 0.510.299
 Ln LF6.82 ± 0.465.85 ± 0.930.0507.30 ± 0.446.67 ± 0.630.0236.43 ± 0.686.99 ± 0.580.749
Ratio-based HRV         
 Normalized LF (%)63 ± 552 ± 80.01957 ± 556 ± 60.63260 ± 652 ± 80.001
 LF/HF ratio6.6 ± 2.24.6 ± 1.90.1684.3 ± 1.94.5 ± 1.60.8585.7 ± 3.53.7 ± 1.70.039
Short-term HRV         
 pNN50 (%)5.7 ± 4.72.7 ± 1.80.21514.0 ± 5.19.2 ± 8.80.18512.3 ± 13.217.2 ± 13.30.272
 rMSSD (msec)25 ± 820 ± 50.22438 ± 830 ± 120.10934 ± 1941 ± 190.268
 Ln HF5.17 ± 0.704.64 ± 0.910.3076.20 ± 0.485.43 ± 0.910.0455.52 ± 1.126.04 ± 0.930.139
Table 7. Comparison of nighttime HRV indices (mean ± SD) by sex for GWI, FM, and control groups*
M (n = 6)F (n = 5)PM (n = 7)F (n = 19)PM (n = 18)F (n = 18)P
  • *

    HRV = heart rate variability; FM = fibromyalgia; GWI = Gulf War illness; HR = heart rate; bpm = beats per minute; SDNN = standard deviation of normal-to-normal intervals; Ln = natural logarithm; ULF = ultra low frequency; SDNNDIX = average of the standard deviations of normal-to-normal intervals over 5 minutes; VLF = very low frequency; LF = low frequency; HF = high frequency; pNN50 = percentage of normal-to-normal intervals >50 msec different from the prior interval; rMSSD = root mean square of the differences between successive normal-to-normal intervals.

  • GWI and FM significantly different from controls.

  • Significant difference between GWI and FM.

Longer-term HRV         
 HR (bpm)62 ± 771 ± 60.06062 ± 969 ± 80.06762 ± 1061 ± 70.741
 SDNN (msec)102 ± 1871 ± 160.016111 ± 2586 ± 230.026106 ± 3093 ± 260.166
 Ln ULF8.14 ± 0.777.89 ± 0.340.5238.18 ± 0.808.12 ± 0.660.8548.38 ± 0.608.19 ± 0.770.415
Intermediate-term HRV         
 SDNNDIX (msec)73 ± 1640 ± 130.00586 ± 2257 ± 170.00174 ± 2667 ± 250.445
 Ln VLF8.01 ± 0.466.80 ± 0.690.0077.97 ± 0.587.48 ± 0.560.637.87 ± 0.567.66 ± 0.540.255
 Ln LF7.22 ± 0.465.63 ± 0.900.0047.74 ± 0.606.60 ± 0.770.0027.17 ± 0.636.88 ± 0.750.210
Ratio-based HRV         
 Normalized LF (%)58 ± 544 ± 70.00456 ± 1353 ± 90.47854 ± 1150 ± 120.251
 LF/HF ratio3.8 ± 1.72.2 ± 0.80.0923.0 ± 2.03.1 ± 1.50.7863.3 ± 2.32.2 ± 1.40.115
Short-term HRV         
 pNN50 (%)20.2 ± 15.16.0 ± 5.20.7730.0 ± 16.011.5 ± 12.80.00423.2 ± 20.423.6 ± 18.40.950
 rMSSD (msec)45 ± 1725 ± 90.04061 ± 1935 ± 170.00353 ± 3250 ± 310.824
 Ln HF6.23 ± 0.765.16 ± 1.160.0986.98 ± 0.845.68 ± 0.940.0046.36 ± 1.176.31 ± 0.940.910

Twenty-four-hour HRV (Table 5).

Among healthy controls, HRV was similar between women and men. The exceptions were the ratio of HRV indices, low frequency (LF) to high frequency (HF) ratio, and normalized LF power, which were significantly higher in male controls, suggesting a difference in the autonomic balance of healthy men compared with women.

In contrast, among both the FM and GWI subjects, strong and significant differences by sex were seen for longer- and intermediate-term HRV, with women having lower HRV. Women with FM or GWI also tended to have higher mean heart rates than men (rather than lower heart rates as seen in female controls). Short-term HRV was significantly reduced among female FM patients compared with male FM patients, but sex differences in short-term HRV did not attain statistical significance for the GWI group. Also, the gender differences in ratio HRV, seen in controls, were also seen in GWI but not in FM, where ratio HRV was similar in men and women.

Daytime compared with 24-hour HRV (Table 6).

Daytime HRV reflects the effect of activity, which was the same for all subjects, on cardiac autonomic function. Patterns were similar to those for 24-hour HRV and only differences will be described here. In female controls, longer- and intermediate-term daytime HRV indices were similar to 24-hour indices. In men, however, daytime HRV was lower than 24-hour indices. As a result, ultra low frequency power, which reflects heart rate responses to circadian rhythms and sustained activity, became significantly higher in women.

Nighttime HRV (Table 7).

Nighttime HRV reflects cardiac autonomic function under resting conditions. Among the healthy controls, all indices of HRV, even the ratio indices (which were markedly different when measured over 24 hours and during the daytime) were similar in women and men during the nighttime.

In contrast, significant reductions in longer-term and intermediate-term HRV for FM and GWI women compared with men continued to be seen at night. The average of the standard deviations of N-N intervals over 5 minutes (SDNNIDX), reflecting combined sympathetic and parasympathetic modulation of heart rate, was relatively unchanged between daytime and nighttime in all of the women, and increased at night in all of the men, with the highest value seen in men with FM.

Nighttime compared with daytime HRV (Table 8).

When the difference in HRV between nighttime and daytime was examined, significant sex differences were seen for the control group, with greater night:day differences in heart rate and smaller night:day differences in both standard deviation of N-N intervals (SDNN) and SDNNIDX for women. Among the GWI group, significant sex differences in the night:day difference in SDNNIDX, similar in magnitude and direction to that seen in controls, were also observed. No differences, however, were seen in night:day HRV changes between men and women in the FM group. In addition, night:day changes in normalized LF power and root mean square of the differences between successive N-N intervals (rMSSD) were different by sex in GWI. As Table 8 shows, however, confidence intervals were wide, suggesting considerable interindividual variability in these parameters.

Table 8. Comparison of night:day differences for selected HRV indices (mean ± SD) by sex for GWI, FM, and control groups*
M (n = 6)F (n = 5)PM (n = 7)F (n = 19)PM (n = 18)F (n = 18)P
  • *

    For abbreviation definitions, see Table 7.

Longer-term HRV         
 HR (bpm)−17 ± 7−11 ± 50.155−13 ± 6−12 ± 40.593−16 ± 5−11 ± 3.50.002
 SDNN (msec)−7 ± 14−18 ± 100.169−23 ± 38−19 ± 270.764−11 ± 30−40 ± 160.001
Intermediate-term HRV         
 SDNNDIX (msec)16 ± 15−2 ± 70.04312 ± 260 ± 140.13511 ± 16−2 ± 140.011
Ratio-based HRV         
 Normalized LF (%)−5 ± 7−8 ± 60.004−3 ± 9−3 ± 100.646−5 ± 8−2 ± 90.214
 LF/HF ratio−2.6 ± 1.5−2.3 ± 1.60.789−1.4 ± 1.1−1.3 ± 1.10.894−2.4 ± 2.11.4 ± 1.20.094
Short-term HRV         
 pNN50 (%)14.6 ± 13.03.3 ± 4.60.10016.6 ± 18.22.3 ± 7.40.08710.9 ± 11.56.4 ± 8.90.197
 rMSSD (msec)20 ± 145 ± 60.0475 ± 1135 ± 170.07319 ± 219 ± 160.132

Comparison of HRV by group within sexes.

A different perspective emerges when HRV patterns are compared between groups by sex. None of the ANOVAs comparing HRV or night:day HRV by CMI group was significant for the male subjects. As can be seen in Tables 5, 6, and 7, HRV in GWI men was similar to that in male controls, and appeared to be slightly higher in men with FM. However, perhaps due to the small sample size, these differences did not attain statistical significance.

Autonomic differences were much sharper among the women. ANOVAs by CMI group for women were significant for most HRV indices. The ANOVA was also significant for the night:day change in SDNN (only) among women. Significant post-hoc comparisons are indicated in Tables 5, 6, and 7. In general, both FM and GWI women had significantly increased heart rates and decreased longer-term, intermediate-term, and short-term HRV compared with female controls, and differences between FM and GWI women were not statistically significant. ANOVAs for differences in ratio indices between female CMI groups were not significant. However, values for very low frequency and LF power were significantly lower in GWI women than in FM women. Mean values for these indices in FM were intermediate between GWI and controls but were not significantly different from controls.


Results of our study extend prior findings of abnormal autonomic function among women with FM to women with GWI, in whom similar findings were noted. This study is the first to note similar changes in HRV in women with Gulf War Illnesses: individuals with symptoms similar to those of fibromyalgia and chronic fatigue syndrome, but who developed these symptoms temporally related to having served in the military during the 1991 Gulf War. Previous studies have pointed out the similarity in symptoms between FM, chronic fatigue syndrome, and GWI, but this is among the first to show that there may also be similar mechanisms operative in GWI (5–9). Further studies with larger sample sizes will be needed to confirm these findings.

However, we did not note these findings in either group (FM or GWI) of men, suggesting that specific autonomic effects of chronic multisymptom illness may be sex-dependent. Thus, although we did not note differences between most HRV measures in healthy men and women, we observed a consistent reduction in HRV in women compared with men with FM or GWI.

Moreover, the magnitude of the abnormalities seen in women with this spectrum of illness was not trivial. Mean HRV values in female FM patients (e.g., 24-hour SDNN = 116 msec) were decreased to values similar to those reported for predominantly male cardiac patients 1 year after myocardial infarction (mean SDNN = 112 msec), indicating marked abnormalities in cardiac autonomic modulation. Mean SDNN was even lower (97 msec) for the female GWI subjects. In the same postmyocardial infarction study, 24-hour SDNN (141 msec for age- and sex-matched healthy controls) was similar to SDNN for our control subjects (23). Although results did not reach statistical significance, our data also suggest the possibility that HRV could be higher than normal in men with FM.

These results are not inconsistent with previous work in this field. We know of no studies of 24-hour HRV among GWI patients, but 2 prior studies have compared 24-hour HRV in FM patients and controls (11, 12). Time domain results for Raj et al (11) were similar to ours, but frequency domain results, which were based on an averaged analysis of 2-minute rather than 5-minute periods were not directly comparable.

Only time domain HRV was reported by Martínez-Lavín et al (12). Results for a group of 28 female and 2 male FM patients were consistent with ours, except for the report of similar values for rMSSD for FM patients and controls. This is difficult to explain because rMSSD and percentage of N-N intervals >50 msec different from the prior interval (pNN50), which were reported to be significantly decreased in FM, each reflect beat-to-beat changes in HRV and are usually highly correlated.

Cohen et al performed the only HRV analysis in both women and men with FM, although this study was performed at rest rather than during daily activities (10). HRV was measured during a 20-minute supine period under controlled conditions and calculated (and presumably averaged) over 256-beat (roughly 5-minute) intervals. Results of higher heart rates among female FM patients compared with female controls are consistent with ours, but higher heart rates among male FM patients compared with controls are not. Moreover, supine resting heart rates in both female groups were higher than daytime (active) heart rates among our subjects, while supine resting heart rates among men were lower. The results of Cohen et al are also difficult to interpret because reported values for LF and HF power do not resemble ours nor do those reported by anyone else, and the procedure for determining normalized LF and HF power, which also resulted in values not seen anywhere else, are not explained (10). Thus, results from this study cannot be compared with those of the current investigation.

Martínez-Lavín et al (12) and Raj et al (11) also have reported that, among other HRV differences between female patients and controls, those with FM have significantly higher values for the ratio of low to high frequency power. In the current study, although mean values for the LF/HF ratio were higher in FM, this difference did not achieve statistical significance. It is possible that this different result derives from the measurement of HRV under controlled conditions in the GCRC, rather than during routine, and likely different, daily activities.

Limitations of this study must be noted. The number of subjects in each group was relatively small and was extremely small in some subgroups. Thus, results need to be validated on a larger population. Also, although subjects with known cardiac disease were excluded from the study, 12 subjects were >50 years old and the possibility of occult disease, which could affect cardiac autonomic function, cannot be ruled out. Finally, although the fact that subjects engaged in identical activities was a strength of the study, it must be noted that subjects slept at the GCRC. The first night of sleep in a clinical setting may not be representative of sleep for that person (the so-called first night effect). Although it is not clear that this would necessarily have differentially affected patients and controls, the first night effect could have influenced HRV.

How then can HRV results be interpreted? Heart rates were increased and HRV indices reflecting circadian rhythms, combined sympathetic and parasympathetic modulation, and purely parasympathetic modulation of heart rate were significantly reduced in women with CMI. When further evidence for increased sympathetic activation is available, e.g., increased catecholamines, these findings, in combination with increases in the LF/HF ratio, can reasonably be assumed to reflect increased sympathetic activity. In fact, unpublished results from our group are more consistent with this latter interpretation.

Although results of our comparison of HRV in female FM and GWI patients and healthy controls clearly reveal abnormalities in cardiac autonomic modulation among patients, the mechanisms that underlie this abnormality are still unknown. Decreased HRV can reflect abnormalities in centrally mediated autonomic outflow or in cardiac responsiveness to autonomic input. Decreased HRV could also reflect a normal autonomic response to abnormal physiologic conditions, e.g., hypovolemia. This not likely to be the case here, because systolic and diastolic blood pressures were the same in the CMI and control groups. It is not yet clear whether HRV is abnormal in a subset of patients, or whether HRV in CMI is decreased relative to normal values in that individual. It is also not known whether normal persons with abnormal HRV are at greater risk for CMI, or if abnormal HRV develops subsequent to the onset of the syndrome. Measurement of HRV in a prospective manner in individuals who move from asymptomatic to symptomatic will answer these questions.