The results of this study lead to two main discussion points – whether CFS and FM are the same or different illnesses and whether patients with these clinical diagnoses may actually have SDB as the cause of their symptoms.

Concerning the first point, a review of the literature suggested that FM and CFS might be different disease processes in that SDB was reported to be common in FM while several studies, including our own [9-11], indicate SDB is not common in CFS. One study reported finding SDB to occur in men but not women with FM [13], and two others reported finding SDB at high rates in women with FM [14,15]. We did not find an increased rate of FM in our earlier study of sleep characteristics in patients with CFS alone or CFS plus FM [11]. Two small studies used the approach we had taken here of evaluating patients with PSG-proven SDB for FM. One study found no increase in FM in SDB [12] while the second did [16]. We expect these discrepancies may be due to methodological differences related to our excluding patients with multiple sleep diagnoses and/or issues related to sample size. In contrast to those reports reporting an increased rate of FM in SDB, we found an actual dissociation between the diagnosis of FM and SDB. None of our SDB patients had FM alone and only 3% – the same rate as seen in the general population for FM [28] – had FM with CFS.

Because CFS and FM co-exist for a substantial number of patients, some researchers have suggested both ailments fall within the category of somatic amplification, having functional or psychogenic rather than organic bases [3,4]. The data reported here provide two pieces of evidence against that interpretation. First, if CFS, FM and IBS were all variants of a similar psychological process, Beck scores should be the same for patients bearing the diagnosis of one, two or three of these symptom-based syndromes. This was not the case: Beck scores increased monotonically with frequency of symptom-based syndromes – a finding we have reported earlier [29].

Second, if CFS and FM were essentially the same illness, then rates of both illnesses should co-vary and rates of each should not differ between different primary sleep disorders. This was not the case. Rates of CFS and FM did covary in insomnia, but rates of CFS and FM diverged in SDB with these patients having high rates of CFS but rates of FM at the population norm. Finding this difference suggests that CFS and FM have different pathophysiological underpinnings – at least in the face of SDB. Therefore these data argue against the “unitary” hypothesis that CFS and FM are variants of a functional somatic syndrome due to a psychologically based amplification of symptoms that every person experiences. That conclusion must be tempered by the possibility that such a mechanism could produce different symptoms in different patients.

Despite the small sample size, finding that narcolepsy patients had predominantly CFS and not FM further supports the idea that CFS and FM are not necessarily the same. The fact that differences do exist between CFS and FM supports the continued use of individual case definitions for each disorder and argues against modifying diagnostic criteria for FM [30] to bring the two disorders closer together.

The second discussion point has to do with the issue of ruling out underlying SDB as a possible, but unappreciated, cause of apparent chronic fatigue syndrome. Current guidelines do not recommend a formal sleep evaluation including PSG as part of the work up to rule out medical causes of fatigue before making the diagnosis of CFS. However, we found that 13% of patients with PSG-proven SDB fulfilled case criteria for CFS. In view of the fact that the community prevalence of CFS is 0.42% [31], the rate of 13% is greatly increased and of clinical importance. This finding leads to two research questions. First, is it possible to predict the existence of SDB in patients receiving the diagnosis of CFS? Being able to make this prediction would lead to PSG to corroborate the existence of this sleep disorder and then to try to treat it.

Excessive daytime sleepiness as assessed by the Epworth Sleepiness Scale has turned out to be a disappointing predictor of SDB in that the majority of SDB patients have Epworth scores in the normal range [32]. However, the predictability of elevated Epworth scores in identifying SDB is greatly improved for those SDB patients who also fulfill criteria for CFS. More than half of the 16 patients in this group had Epworth scores >10 with the mean, median and 75th percentile being 13.3, 16 and 17.3, respectively. In considering these data, it would be important to know Epworth data for CFS patients without SDB. While our study did not provide these data, two studies do provide such data for samples of 24 and 415 CFS patients, respectively (mean = 9.4; SD = 4.5 [33]; mean = 10.5; SD = 5.5 [34]). Since available data indicate SDB to be at population rates for CFS samples, we accept the result of these studies as being typical of CFS patients without SDB. Epworth scores were significantly lower by t-tests (p < 0.05) than in the sample with SDB reported here. Thus, patients with SDB who also fulfill criteria for CFS alone or CFS with FM have more daytime sleepiness than CFS patients without SDB.

The second research question is whether treating the SDB in CFS-SDB patients with CPAP would produce symptomatic improvement. While the original case definition for CFS did not require PSG in the work up of patients with severe and chronic fatigue, the 2003 discussion of shortcomings in that case definition [35] emphasized that “assessment of sleep must detect treatable primary sleep disorders and evaluate sleep-related symptoms that may be part of CFS”. Our finding that 13% of patients with SDB also fulfilled criteria for CFS supports that conclusion. The data presented here suggest an algorithm to help the physician determine which CFS patient should undergo PSG based on his/her Epworth Sleepiness Score. We recommend Epworth scoring on all patients being evaluated, and then, to reduce the risk of a negative sleep study, PSG should be considered for the patient who fulfills criteria for CFS and who has an Epworth score of 16 or above – i.e., at and above the 50th percentile for this group; based on Epworth data from 415 CFS patients [34], approximately 13% of all patients would fall into this category. Using these criteria to perform PSG in patients with the diagnosis of CFS will allow the clinician to make the diagnosis of SDB and then determine whether treating it may greatly reduce the complaint of severe fatigue. This important study remains to be done.

The use of antidepressants as a second possible treatment for this group of patients emerges from our finding an association of Epworth scores in patients with SDB and CFS with high BDI scores, a measure of depressed mood. The link between depression and excessive daytime sleepiness has been shown in general population samples as well as in those comprised of SDB patients [36-38]. This treatment may be most effective in the subgroup of patients with both high Epworth and high BDI scores.

Strengths and Weaknesses: A major strength of this study was the fact that we had a large sample size of 122 patients with SDB. Having so many SDB patients provided another strength in that we had an adequate sample to examine whether patients with a milder form of SDB might differ from those with more severe SDB in rates of CFS and/or FM. They did not. Being able to look across the breadth of SDB helped rectify another weakness – that FM is predominantly a disease of women and thus might not be seen in adequate numbers in an SDB population; however, the data were essentially the same for the two SDB groups even though the gender prevalence in the two groups did differ. One potential weakness had to do with our sample – people with sleep complaints referred for evaluation in a sleep lab. While the complaint of unrefreshing sleep is among the most common symptoms of CFS [39], it is not endorsed by all patients and so the patients studied here may have been biased toward those with identifiable sleep disorders; conversely, patients with CFS may have unrefreshing sleep stemming from their other symptoms/problems such pain, sore throat, headaches. Sample size was also a weakness – at least for the small groups with IH and narcolepsy. Nonetheless, finding high rates of CFS in narcolepsy was not unexpected as these patients are notable for having marked daytime sleepiness, but the result does suggest the possibility of misdiagnosis for some patients thought to have CFS – further support for doing PSGs in selected cases of excessively sleepy patients with severe fatigue. A next step for researchers interested in the relation between sleep and symptom-based illness will be to expand the sample size of patients with these two sleep diagnoses.