Chronic fatigue syndrome (CFS) is a medically unexplained condition occurring mostly in women and is characterized by persistent or relapsing fatigue that lasts at least 6 months and substantially interferes with normal activities. In addition to severe fatigue, one of the symptoms used for diagnosing CFS is unrefreshing sleep, and, in fact, this sleep-related problem is the most common complaint among patients with severe medically unexplained fatigue 1. An obvious possibility is that patients with this problem have an underlying sleep disorder or substantial amounts of interrupted sleep which may be responsible for the genesis of the illness. This idea was supported by a recent longitudinal study that indicated that 20% of a carefully delineated group of CFS patients were found to have sleep apnea or narcolepsy, exclusions for the diagnosis of CFS 2.

A number of reports of polysomnography in CFS patients were remarkable for finding high rates of these sleep disorders plus periodic leg movement (PLM) disorder 345, whereas several recent studies have found rates to be the same as in controls 67. One possible reason for this discrepancy is the existence of coexisting fibromyalgia (FM). None of these previous papers stratified their patient sample as to the existence of coexisting FM, and recent data indicate substantial amounts of sleep-disturbed breathing in patients with this disorder 89. FM is a medically unexplained syndrome characterized by four quadrant pain and multiple tender points and frequently occurs in conjunction with CFS 10. Hence, we evaluated polysomnograms (PSGs) of CFS patients with and without FM to determine whether patients in either group had elevated rates of sleep-disturbed breathing (obstructive sleep apnea or upper airway resistance syndrome) or PLM disorder.

Another important issue was whether CFS patients would show abnormalities in their sleep architecture even if clinical sleep disorders were not present. Two studies have been done on patients with 'pure' CFS (that is, in patients with neither sleep disorder nor evidence of current depression, another illness that can interfere with sleep) with differing results: one reported low sleep efficiency with increased periods of wakefulness in CFS 11 and the second found normal sleep architecture 12. We decided to extend these studies and to determine whether the patient's subjective response to sleep, another source of variability in a clinical sample, might correlate with or predict sleep disturbance. Therefore, we also determined whether feelings of unrefreshing sleep were associated with differences in sleep architecture from normal.

Evaluation of the PSG led us to exclude 10 subjects with clinically significant sleep abnormalities: 3 controls with RDIs of 26, 22.4, and 18/hour, 1 CFS patient with an RDI of 22.1/hour, 1 CFS/FM patient with an RDI of more than 40/hour, and 1 control, 3 CFS patients, and 1 CFS/FM patient with PLMs. We included 1 CFS patient and 3 healthy controls with RDIs of 10.4, 10.8, 10.1, and 9.5/hour, respectively, as these fall within the range seen in asymptomatic normal subjects in the study by Ayappa and colleagues 20. This left a total of 26 CFS patients, 12 with comorbid FM, and 26 healthy control subjects.

Table 1 depicts the key PSG measures of the healthy controls and CFS patients. Total sleep time was significantly longer for healthy controls than patients as were the total durations of stage 1, stage 2, and rapid eye movement (REM) sleep, whereas total duration of wakefulness did not differ between healthy controls and patients. As a result, patients had a significantly lower sleep efficiency (that is, the percentage of the total time asleep after falling asleep relative to the time spent in bed) than healthy controls. However, sleep latency, defined as the time from lights out to the first three consecutive epochs of sleep stage 1 or the first epoch of other stages of sleep (that is, sleep onset), and total duration of slow-wave sleep (SWS) (that is, the sum of stage 3 and 4 sleep) did not differ significantly between groups. Data were also evaluated based on whether the patient had CFS alone or CFS plus FM (Table 2). Patients with CFS plus FM had sleep structures similar to those of patients with CFS alone.

Table 1 also shows that sleepiness, fatigue, and pain before and after the PSG night were significantly higher in patients than healthy controls. Values for subjective a.m. sleepiness, fatigue, and feeling blue decreased compared with the evening in healthy controls, whereas none of these decreased for patients.

For patients, self-rated sleepiness, fatigue, and pain before sleep correlated positively with sleep efficiency (r = 0.39, 0.59, 0.57; P < 0.05) and duration of stage 4 sleep (r = 0.48, 0.42, 0.56; P < 0.05) and negatively with sleep latency (r = -0.40, -0.42, -0.40; P < 0.05); self-rated fatigue and pain correlated negatively with durations of wakefulness after sleep onset (r = -0.50, -0.43; P < 0.05) and wakefulness plus stage 1 sleep (r = -0.50, -0.57; P < 0.05). Self-rated fatigue correlated negatively with REM latency (r = -0.49; P < 0.05). Self-rated pain correlated positively with total sleep time (r = 0.49; P < 0.05) and durations of stage 3 sleep (r = 0.49; P < 0.05) and SWS (r = 0.59; P < 0.05). Moreover, change in self-rated sleepiness and fatigue over the night correlated positively with sleep latency (r = 0.39, 0.49; P < 0.05) and negatively with sleep efficiency (r = -0.41, -0.54; P < 0.05). Changes in self-rated fatigue over the night correlated negatively with total sleep time (r = -0.39; P < 0.05) and total duration of stage 3 sleep (r = -0.41; P < 0.05). No significant relations were found among any of these variables for the healthy control group. When we looked at correlations between self-rated variables reported after sleep and sleep stage variables, none was significant except that self-rated sleepiness after sleep correlated positively with duration of stage 2 sleep (r = 0.48; P < 0.05) for the healthy controls.

Patients in the a.m. sleepier group showed significantly longer sleep latency, poorer sleep efficiency, and shorter duration of median sleep run than healthy controls (Table 3). The survival curve of all sleep runs depicted in Figure 1 shows that patients in the a.m. sleepier group had a lower percentage of long runs of sleep than the other two groups and healthy controls (that is, less continuous sleep). For example, the proportions of runs lasting more than 10 minutes were 39.3%, 45.5%, and 49.0% for patients in the a.m. sleepier group, the a.m. less sleepy group, and healthy controls, respectively. The difference in temporal distribution of periods of wakefulness is evident from the representative data in Figure 1. Both the control subject and the a.m. less sleepy patient have periods of wakefulness that are spaced more evenly over time than is the case for the a.m. sleepier patient, whose periods of wakefulness appear bunched in time (Figure 1a). The frequencies of these bouts occurring after sleep onset did not differ among groups (25 ± 15, 21 ± 5, and 23 ± 7 for patients in the a.m. sleepier group, in the a.m. less sleepy group, and for healthy controls, respectively).

The existence of coexisting FM did not predict sleep quality for patients (n = 7 in the a.m. less sleepy group and 5 in the a.m. sleepier group). Patients in the a.m. less sleepy group had significantly higher CES-D scores than those in the a.m. sleepier group (Table 3).

Prior to going to sleep, the a.m. less sleepy patient group reported more sleepiness than both healthy controls and patients in the a.m. sleepier group (P < 0.05). On the morning after their night in the sleep lab, patients in the a.m. sleepier group had significantly more sleepiness, fatigue, and pain than both healthy controls and patients in the a.m. less sleepy group (P < 0.05). Whereas fatigue and pain did decrease for patients in the a.m. less sleepy group, neither of these symptoms changed for patients in the a.m. sleepier group (Table 3). Visual analog scores for feeling blue showed minor differences among groups with little change after the night in the sleep lab (Table 3). Sleepiness and fatigue in both evening and morning were correlated (P < 0.05) in both patients and controls. Significant correlations were found for pain and fatigue reported in the evening and morning for patients but only in the evening for controls.

CFS is diagnosed using clinical criteria and is therefore probably comprised of a heterogeneous patient pool. The data reported here indicate that some CFS patients have a problem with normal regenerative sleep, which may be responsible for the genesis of their symptoms. In this study, we reduced patient pool heterogeneity by studying women only during a fixed period of their menstrual cycle and after excluding patients with either major depressive disorder or PSG-defined sleep disorders. As a group, the patients studied showed evidence for sleep disruption in the form of significantly reduced total sleep time, reduced sleep efficiency, and shorter bouts of sleep than healthy controls. In comparison with controls, sleep in CFS had little effect on either self-reported sleepiness or fatigue. And, interestingly, for patients only, ratings of sleepiness and fatigue correlated well with total sleep duration and efficiency.

Stratifying patients as to the presence of comorbid FM did not further reduce the heterogeneity seen in the patients' sleep structure. However, asking them about their level of sleepiness before they went to sleep and immediately after awakening did. Dichotomizing the patients into a group that felt sleepier after a night's sleep than before sleep and a group that felt less sleepy after a night's sleep reduced the variability of the sleep records considerably. Those patients reporting less sleepiness after a night's sleep had sleep structures similar to those for healthy controls except for a shorter total sleep time and a commensurate reduction in stage 2 sleep; moreover, they reported their fatigue and pain to diminish following sleep. In contrast, patients in the a.m. sleepier group had the greatest abnormalities of sleep architecture, including poor sleep efficiency, longer sleep latency, and more disrupted sleep as manifested by a higher percentage of short-duration sleep runs, than either controls or patients in the a.m. less sleepy group (Figure 1).

The net effect of this sleep disruption may be the genesis of symptoms reported by this group of CFS patients. The effects of sleep disruption are well known to produce severe daytime fatigue, an example being patients with sleep apnea who have very disturbed sleep. In the case of CFS, neither arousals nor periods of wakefulness per se may be the problem so much as the pattern in which they occur. Patients in the a.m. sleepier group had a shift away from longer bouts of sleep to more frequent short-sleep bouts (that is, fragmented sleep, which may prevent them from falling back to sleep after awakening), resulting in their developing fatigue, unrefreshing sleep, cognitive problems, and achiness. These data appear to support the sleep continuity theory, which hypothesizes that good sleep quality requires longer periods of uninterrupted sleep 24. Reduced energy and cognitive problems are known to occur in healthy controls who have normal sleep time despite disrupted sleep produced experimentally 25. In addition, some studies in healthy volunteers have reported increases in musculoskeletal pain and/or decreases in pain threshold after a period of sleep disruption or deprivation 262728. We are currently testing the hypothesis that the process responsible for disturbing the sleep of this group of CFS patients is an imbalance of the cytokine sleep network (that is, sleep-producing and sleep-disrupting cytokines) in favor of sleep-disrupting cytokines.

One purpose of this study was to determine whether stratifying our patient sample into those with and without comorbid FM would explain discrepancies in the literature as to rates of sleep pathology. It did not. Regardless of the presence of FM, our findings were similar to earlier reports of rather low rates of sleep pathology in CFS 67. The low rates of sleep-disturbed breathing and PLMs we found in both patient groups are similar to those we found in our control group of sedentary women – rates that approached the values reported in the literature for unselected populations of healthy women 2930. However, in our hands, we found low rates of sleep-disturbed breathing for patients with CFS alone or CFS plus FM.

Importantly, the rates we found for sleep disturbed breathing include data along the entire spectrum of sleep disturbed breathing from overt sleep apnea to the upper airway resistance syndrome. The monitoring technique we used for airflow, a nasal cannula and examination of the flow signal for the characteristic shape of flow limitation, should have detected subtle forms of sleep disturbed breathing in our sleep studies, but only rare occurrences of patterns of air flow consistent with flow limitation were found here. Thus, although we did not use the more invasive technique of esophageal manometry to detect respiratory effort-related arousals, our results do not support an association between subtle forms of sleep disturbed breathing and CFS, even when co-morbid FM is present. This conclusion contrasts with an earlier report of EEG patterns "related to subtle, undiagnosed sleep-disordered breathing" in patients with chronic fatigue 31. The apparent difference between these studies may relate to diagnostic specificity for CFS. All of our patients fulfilled the 1994 case definition for CFS, which requires their having disabling fatigue for at least 6 months plus at least four of eight infectious, neuropsychiatric, or rheumatological symptoms 15; the subjects in the earlier study just had fatigue of long duration. Thus, our study does not eliminate the possibility that some patients with severe fatigue alone may have this problem as a result of subtle forms of sleep-disturbed breathing.

In summary, based on sleep patterns as assessed by polysomnography, patients with CFS alone and CFS plus FM have a similar rate of diagnosable sleep disorders; in fact, neither group has rates of sleep disorders higher than those found in healthy controls. Thus, sleep-disturbed breathing, narcolepsy, and leg movement disorders are an uncommon cause of medically unexplained fatigue or pain syndromes. Moreover, after excluding those patients from further analysis, CFS and FM patients have similar sleep structures. Our results also suggest that, even when the rate of arousals is within the normal range, sleep quality may be affected by a decrease in the length of episodes of uninterrupted sleep.

CFS patients had significant differences in polysomnograpic findings from healthy controls and felt sleepier and more fatigued than controls after a night's sleep. This difference was due neither to diagnosable sleep disorders nor to coexisting FM but primarily to a decrease in the length of periods of uninterrupted sleep in the patients with more sleepiness in the morning than on the night before. This sleep disruption may explain the overwhelming fatigue, report of unrefreshing sleep, and pain of patients in this subgroup.

CES-D = Centers for Epidemiological Study-Depression; CFS = chronic fatigue syndrome; ECG = electrocardiogram; EEG = electroencephalogram; EMG = electromyogram; EOG = electrooculogram; FM = fibromyalgia; PLM = periodic leg movement; PSG = polysomnogram; RDI = respiratory disturbance index; REM = rapid eye movement; SWS = slow-wave sleep.

The authors declare that they have no competing interests.

FT provided interpretation of the results, statistical analyses, and preparation of the manuscript. BHN provided the design of the study, recruitment of the patients, organization and realization of the experimental design, interpretation of the results, and preparation of the manuscript. NSC provided the design of the study, interpretation of the results, and preparation of the manuscript. JF and CG provided recruitment of the patients, acquisition of data, and preparation of the manuscript. DMR assisted in study design, interpretation of the results, and preparation of the manuscript. All authors read and approved the final manuscript.