The diagnosis of chronic fatigue syndrome (CFS) for research and clinical purposes is based on a clinical definition: persistent or relapsing fatigue of at least 6-months' duration, that is not alleviated by rest and that causes a substantial reduction in activities; a fatigue that cannot be explained by medical or psychiatric conditions and must be accompanied by at least four of the eight case defining symptoms (unusual post exertional malaise, impaired memory or concentration, unrefreshing sleep, headaches, muscle pain, joint pain, sore throat and tender cervical nodes) [1].

Since CFS definitions were originally formulated for research purposes, the exclusion of active medical and psychiatric illnesses or diseases associated with fatigue was considered necessary for two reasons [1-5]. First, some of those suffering such symptoms have an explanation for their fatigue; and, secondly, mislabelling a condition that would respond to specific therapy (i.e., anaemia or hypothyroidism) as CFS would not be appropriate. In addition, those who had psychotic psychiatric illnesses might not be able to realistically report their symptoms and levels of function. Thus, certain diseases such as hypothyroidism and diabetes mellitus are considered exclusionary until adequately treated [4]. Other diseases with core symptoms that are also seen in CFS - for example, a major depressive disorder with melancholic features (MDDM) or chronic hepatitis C - were considered permanent exclusions because of their persistent or recurrent course.

In addition to its use in research, the definition is widely used for clinical purposes. The clinical application has been hampered because it gives practitioners little guidance on how to determine when a patient is sufficiently impaired or symptomatic to warrant the diagnosis and when to include patients with adequately treated 'exclusionary' diseases who may continue to have symptoms and consequences compatible with CFS. The first issue was partly addressed by the application of empiric methods to the diagnostic process [6]. However, the problem of how to handle exclusionary diseases remained unanswered. For example, patients with hypothyroidism or systemic lupus erythematosus that is 'under control' or inactive, based on physical examinations and laboratory evaluations, may still fulfill the criteria for CFS. Clinically, they are frequently managed as CFS patients, raising the question of whether those who would meet the research criteria for CFS, except for the single fact that they have an exclusionary condition, are similar to those who fully meet the CFS case definition criteria of fatigue severity, impairment and specific symptoms and, by inference, of the underlying pathophysiology of fatigue. It is also recognized that, while the illnesses are considered exclusionary because of their frequent association with fatigue, many with those diagnosed do not have fatigue.

The purpose of this analysis is to describe the prevalence of exclusionary diagnoses identified in a population-based surveillance study of fatigue illnesses and to compare those with and without such conditions in terms of: (1) demographics; (2) specific exclusionary categories; (3) the presence of CFS case-definition criteria; and (4) levels and characteristics of fatigue and impairment. The frequency with which potentially treatable conditions were identified in subjects otherwise meeting research criteria for CFS emphasizes the need for clinicians treating patients presenting with unexplained fatigue to consider these illnesses in the differential diagnosis.

This study was approved by the US Centers for Disease Control and Prevention (CDC) Institutional Review Board, as required by Department of Health and Human Services regulations. All participants were volunteers who gave their informed consent. The study was conducted in English. Non-English speaking respondents were excluded.

The study illustrates the complexity that exclusionary conditions introduce into both research and clinical considerations of subjects potentially classified as CFS. The exclusionary diagnoses are common; 28% of subjects in a population-based sample had exclusions identified during a personal history telephone interview. Exclusionary conditions/diseases were subsequently recognized in an additional 36% of subjects undergoing a full clinical examination. It appears that during the telephone interview subjects did not know of or did not remember all of their exclusionary diagnoses. This discrepancy could be due to: (1) greater accuracy and specificity of information obtained by clinical examination compared to telephone interview; (2) variable access to healthcare which, in turn, means variable access to diagnostic means [16]; (3) asymptomatic status contributing to unrecognized conditions; or (4) a lack of perception of bodily feelings as being abnormal. The types of diseases identified in the clinic are recognized to be important in the differential diagnosis of CFS and emphasize the need for clinicians to evaluate patients with appropriate history taking, physical examination and laboratory tests. The finding that Black people and residents of rural and urban areas were more likely to have exclusionary conditions diagnosed at a clinic suggests that access to healthcare may be a contributing factor, as limited access to quality healthcare has been documented among these subgroups [16]

The finding that increased BMI was independently associated with having an exclusionary condition among CFS patients, suggests that the condition may be 'silent' as in some correlates of elevated BMI, including metabolic conditions, sleep disorders and hypothyroidism. This interpretation is supported by the finding of exclusionary conditions in 26 people who, in the telephone interview, had considered themselves 'well'.

The diagnoses considered exclusionary for CFS were originally selected because they frequently result in symptoms similar, if not identical, to those characteristic of CFS subjects. As might be expected, differences between subjects with and without exclusionary diagnoses were not universally present. For example, general fatigue and reduced activity scores of the MFI and physical function, role physical, social function and role emotional scores of the SF-36 were equivalent. Likewise, CFS and non-CFS symptom scores were not significantly different. Thus, in terms of level of overall 'sickness' [17] and impairment, included and excluded subjects were comparable. It is not surprising that subjects with both medical and psychiatric diagnoses have more fatigue and impairment, as well as higher symptom scores, than those with only medical or psychiatric conditions. However, greater impairment in selected parameters in those subjects with only psychiatric exclusions versus those with only medical exclusions reflects a true difference or bias introduced by the self-report procedure.

Although not emphasizing the differences between subjects who fulfill defined requirements for exclusionary diagnoses and those who do not, previous studies of chronic fatigue and CFS have shown high rates of psychiatric morbidity and functional morbidity and have documented these outcomes as important public health burdens [18,19]. Another study, which included 98 subjects with chronic fatigue and compared disability and psychosocial distress in those that met criteria for CFS and those who had medical or psychiatric exclusions, failed to meet the definition or were using medications specific to the study [20]. The study results showed that the CFS subjects could not be differentiated from those who were excluded based on the study variables that addressed symptoms of depression, general health, impairment, symptom perception and somatic and psychological stress.

In principle, the results of this study, and of previous observations, support the original decision to exclude subjects with fatigue and selected concurrent identifiable illnesses/diseases from the research diagnosis of CFS as the latter shares illness characteristics and consequences with the syndrome [1]. The excluded group may have exposures or disease components that would confound efforts to address the incidence, prevalence or pathophysiology of an otherwise unexplained condition such as CFS if it were unique [21,22]. We did not address the question of subject competency to complete the in the psychiatric exclusions. Thus, considering CFS as a diagnostic possibility and pursuing a differential diagnostic process should allow identification of patients who need careful evaluations as is recommended in the 1994 definition. In particular, obesity, anaemia, thyroid disease, diabetes and heart disease are common in fatigued as well as non-fatigued subjects (Table 3 and 4).

It is equally clear that subjects with exclusionary diagnoses are at least as comparably functionally impaired as the included subjects. The chronically unwell population, identified here with a fatiguing illness, is likely to have a medical disease and/or a psychiatric disease. However, these conditions may go undiagnosed. Since subjects with and without identifiable exclusionary disease processes share many symptoms, clinical management requires careful attention in order to correctly identify and treat the medical and psychiatric illnesses. It is clear that treatment of the underlying diseases will not resolve fatigue and symptoms in all instances and it is possible that illnesses/diseases with chronic fatigue share a common underlying pathophysiologic mechanism. In order to examine this possibility, subjects with exclusionary conditions could be included with CSF cases for comparison.

The primary limitation of this study is the inclusion of subjects who have been ill for an average of 6-7 years. The majority of the subjects had experienced a gradual onset of their illness. Thus the exclusion of subjects fulfilling CFS criteria and having medical and/or psychiatric illnesses may be observed more frequently in the population under study than in the younger individuals with an acute onset. Likewise, the levels of impairment may also be more applicable to this population. Replication of these observations will be possible in future follow-up studies.

As those with CFS suffer from personal, social, workplace [1] and observed financial losses [23], should not all individuals fulfilling CFS inclusion criteria, with or without exclusionary diagnoses, be considered in future public health planning? For instance, would both groups benefit from prevention and intervention efforts such as cognitive behavioral therapy and graded exercise therapy [24,25]? A similar question could be asked of those who are unwell but who do not reach the diagnostic threshold.

This analysis allows three different, but complementary, interpretations of the results that need to be addressed in future studies. First, it substantiates the need for the identification of underlying diagnoses that impact the choice of subjects for epidemiological and mechanistic studies of a specifically described illness construct. Secondly, it suggests that those who would otherwise fulfill the CFS definition, but who have an exclusionary illness or disease, are as impaired as those defined as having CFS. The third interpretation is that CFS, in the population studied here, may reflect underlying medical or psychiatric illnesses or diseases that are not necessarily 'under control'. The data therefore suggest that, in order to substantiate CFS as a specific entity or to identify shared pathophysiological factors, subjects fulfilling the criteria of fatiguing illnesses, but who have exclusionary and non-exclusionary diagnoses, need to be included as comparison case groups in both mechanistic and therapeutic studies. The identification of subjects for CFS studies requires a clinical evaluation; medical and psychiatric histories obtained by telephone interviews are not completely reliable.

BMI: body mass index; CATI: computer-assisted telephone interview; CDC: US Centers for Disease Control and Prevention; CFS: chronic fatigue syndrome; DSM: Diagnostic and statistical manual of mental disorders; DSM-IV: disorders described in the fourth edition of the DSM; ISF: insufficient fatigue; MDDM: major depressive disorder with melancholic features; MFI: Multidimensional Fatigue Inventory; MCS: mental component summary; NF: non-fatigued; OR: odds ratio; PCS: physical component summary; SCID: structured clinical interviews for DSM disorders; SF-36: short form 36; SI: symptom inventory.

The authors declare that they have no competing interests.

JFJ, ERU and WCR were involved in the conception and design. JMSL, EMM, RSB, UMN, ERU, JFJ and WCR were involved in the analysis and interpretation of the data. JFJ. JMSL, EMM, ERU, RSB and WCR drafted the article. JFJ, JMSL, EMM, RSB, UMN, ERU and WCR were responsible for critical revision of the article for intellectual content. JFJ, JMSL, EMM, RSB, UMN, ERU and WCR gave final approval of the article. JMSL, EMM and WCR provided statistical expertise. JMSL, JFJ and WCR were involved in the collection and assembly of the data.

The pre-publication history for this paper can be accessed here: