SEARCH

SEARCH BY CITATION

Keywords:

  • chronic fatigue syndrome;
  • Fukuda definition;
  • heterogeneity;
  • Holmes definition;
  • symptoms

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References

Abstract. De Becker P, MCGregor N, De Meirleir K (VUB, Vakgroep Interne Geneeskunde, Brussels, Belgium; and University of Newcastle, Callaghan, New South Wales, Australia). A definition-based analysis of symptoms in a large cohort of patients with chronic fatigue syndrome. J Intern Med 2001; 250: 234–240.

Objective. The Holmes and Fukuda criteria are widely used criteria all over the world, yet a specific European study regarding chronic fatigue syndrome (CFS) patient symptomatology has not been conducted so far. This study was performed to answer the need to assess the homogeneity of a large CFS population in relationship to the Fukuda or Holmes definitions and to assess the importance of a symptom severity scale.

Design. Multivariate analyses were performed to assess the symptom presentation within a fatigued population and the differences between the Fukuda and Holmes definitions compared with an excluded chronic fatigued group in a large cohort of fatigued patients.

Setting. An outpatient tertiary care setting fatigue clinic in Brussels.

Main outcome measures. Prevalence and severity of symptoms and signs in a CFS population and in a chronic fatigued population.

Subjects and methods. A total of 2073 consecutive patients with major complaints of prolonged fatigue participated in this study. Multivariate analyses were performed to assess the symptom presentation and severity and the differences between the Fukuda and Holmes definitions.

Results. Of the 2073 patients complaining of chronic fatigue (CF), 1578 CFS patients fulfilled the Fukuda criteria (100% of CFS group) and 951 (60.3% of the CFS group) fulfilled the Holmes criteria. Discriminant function analysis revealed that the Fukuda and Holmes definitions can be differentiated by symptom severity and prevalence. The Holmes definition was more strongly associated than the Fukuda definition with the symptoms that differentiated the CFS patients from the patients that did not comply with the CFS definitions. The inclusion of 10 additional symptoms was found to improve the sensitivity/specificity and accuracy for selection of CFS patients.

Conclusions. The CFS patients fulfilling the Holmes criteria have an increased symptom prevalence and severity of many symptoms. Patients fulfilling the Fukuda criteria were less severely affected patients which leads to an increase in clinical heterogeneity. Addition of certain symptoms and removal of others would strengthen the ability to select CFS patients.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References

The chronic fatigue syndrome (CFS) is a clinically defined syndrome that is characterized by chronic fatigue (CF) and a constellation of other symptoms and physical findings [1–3]. Of the five definitions of CFS, three were proposed in the USA [1, 4, 5], one in the UK [6] and one in Australia [7].

Komaroff and Geiger [8], compared patients who met the CDC (Centres for Disease Control and Prevention) case definition with those who did not and found that the CDC case definition according to Holmes did not identify a subgroup of patients more likely to have objective evidence of disease. Similarly, Katon et al. [9] also found that patients with CFS were indistinguishable from those with CF not meeting the CDC criteria. This failure to identify CFS patients is also evident in the lack of reproducible laboratory findings.

This lack of ability to identify CFS patients may mean that the Holmes case definition does not define a homogeneous patient group or that comorbid confounding factors are present that may mask the underlying pathology. Confounding comorbid factors may include mycoplasma infections which have an increased prevalence in CFS patients [10], psychiatric disease such as depression [11], or even other syndromes such as fibromyalgia [12, 13]. Thus there is a need to assess the homogeneity of a large CFS study population.

In a recent paper it was suggested that the Holmes and Fukuda case definitions identify distinct patient groups. Subtle changes in the wording, interpretation and application of the diagnostic criteria used to identify CFS patients can influence prevalence rates and change the characteristics of the samples being identified [14]. These investigators had earlier suggested that the Fukuda definition may need to be revised to clarify the significance of symptom severity in diagnostic and assessment procedures as the current approaches rely on the prevalence and not the severity of the various associated symptoms used to define CFS [15]. Clearly there is also a need to assess the homogeneity of the two different definitions in a large CFS study population and to compare these different definitions with an excluded chronic fatigued group.

This paper has elected to use multivariate analysis to assess the relationship between the definitions and the symptom presentation. To do this, scalar responses to the various symptoms assessed were required. Unlike univariate analysis, which assesses the relationship between two variables and hence indicates the potential predictors of the differences between groups, multivariate analysis assesses the contribution of a combination of potential predictors of the difference between the groups. Whilst the Fukuda and Holmes definitions are very similar, the Fukuda definition is a less stringent definition and is likely to include a greater and more heterogeneous group of patients with profound fatigue [14, 15]. Discriminant function analysis as used in this paper will assess the differences between the groups using all the variables, resulting in selection of a series of predictive symptoms that best describe each group. This should allow the determination of the predictive ability of the defined symptoms as well as the assessment of symptoms not currently included in the definitions. From these analyses a superior definition may be constructed. The disadvantages of this method include the ability of one to interpret the results of the analysis and problems of selection of patients based upon a definition which is from a heterogeneous group. The heterogeneity of the defined group may mask important discriminant symptoms for a subgroup of CFS patients.

This paper uses multivariate analyses to assess the symptom presentation within a fatigued population. The differences between the Fukuda and Holmes definitions and the ability to differentiate them from the excluded CF patients was assessed in a large cohort of patients.

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References

Study setting and population

The study was conducted in Brussels, at a university-based outpatient clinic, and approved by the University hospital ethics committee. We enrolled 2073 consecutive patients seeking care for prolonged fatigue as major complaint in this study. All patients were Caucasian. All patients were referred to our clinic by either their GP or other specialists. The selection and characterization of the subjects involved several steps.

All subjects underwent an extensive medical evaluation, consisting of a standard physical examination and medical history, an exercise capacity test, a symptom checklist and routine laboratory tests. The laboratory tests included a complete blood cell count, determination of the erythrocyte sedimentation rate, a serum electrolyte panel, measures of renal, hepatic and thyroid function, and rheumatological and virological screenings. When judged necessary, a structured psychiatric interview was performed. In a number of cases further neurological, gynaecological, endocrinological, cardiac, psychiatric and/or gastro-intestinal evaluation was performed. When positive results were found in any of the evaluations that met the exclusion criteria according to Holmes [1] or to Fukuda [5], the patients were not diagnosed as CFS patients, this group of patients was termed the CF group.

The medical records were reviewed to determine if patients suffered from organic or psychiatric illnesses that could explain their symptoms. All patients completed a questionnaire which included demographic information, dates of onset and current health status. Afterwards the subjects were examined by one physician, who interviewed the patients with respect to their signs and symptoms.

The participants were asked to complete the Goldstein Symptoms Checklist [16], on which they rated levels of fatigue, CFS definitional symptoms (i.e. postexertional malaise lasting more than 24 h; sore throat; tender neck or axillary lymph nodes; muscle pain; multiple joint pain without swelling or redness; headaches of a new type, pattern or severity; unrefreshing sleep; and impairments in memory or concentration) along with other somatic and cognitive symptoms on a scale from 0 (absent) to 3 (severe). Definitional symptoms were taken from the Holmes [1] and Fukuda [5] case definition cited previously.

When all differential diagnoses were excluded, and the patients fulfilled the CFS Case definition either to Holmes [1] and/or to Fukuda [5] all the data of the CFS patients as well as the data of the CF group were administered into Excel 98.0 on a Power Macintosh (Power Macintosh PC G3 (233 MHz)). The data were coded and transferred to the University of Newcastle, Callaghan, Australia where the statistical analysis was carried out.

Statistical analysis

All data were evaluated for linearity and normality analyses. Subject characteristics were assessed using chi-square probability and Student’s t-tests. Univariant group differences were assessed on untransformed data using the nonparametric Mann–Whitney U-test. Symptom associations were determined by standard discriminant function analysis. The patient classification capacity of the discriminant function module was used to assess the patient compliance within each model. This allowed an evaluation of the predictive capacity of the different symptom groupings in determining a potential diagnosis of CFS. These data were processed using Access97TM (Microsoft, Redmond, WA, USA), Excel97TM (Microsoft) and StatisticaTM (Ver. 5.1, Statsoft, Tulsa, OK, USA).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References

All 2073 patient data were analysed. The analysis group also included patients diagnosed to have fibromyalgia, idiopathic CF, sleep disturbance, depression and psychiatric disorders (termed the CF group), and the CFS patients who complied with the Fukuda and/or the Holmes criteria. Of the 2073 patients complaining of CF, 1578 CFS patients fulfilled the Fukuda criteria (100% of CFS group) and 951 (60.3% of the CFS group) fulfilled the Holmes criteria. In the CF group the primary diagnoses of the fatigue-associated conditions was: fibromyalgia (94–18.9%), idiopathic CF (65–13.1%), depression (52–10.5%), sleep disturbance related fatigue (46–9.3%) and obstructive sleep apnoea syndrome (OSAS) (23–4.6%).

Table 1 shows the variations in symptom severity and prevalence between the Fukuda and Holmes criteria groups. This shows that the patients who were included under the CFS definition using the Fukuda criteria had less severe symptoms and an altered symptom prevalence distribution to those patients classified under the Holmes criteria. Thus, the Fukuda CFS definition has allowed an increase in the number of subjects classified as CFS who have a significant difference in symptom severity and distribution.

Table 1.   Symptom prevalence and severity differences between the Fukuda and Holmes groups Thumbnail image of

CFS definitions compared with CF

The different definition groups (Holmes, Fukuda, CF) were compared using standard discriminant function analysis with the Holmes criteria symptom profile and the Fukuda symptom profile.

The Holmes criteria group had a strong regression model (Wilks’λ=0.49; F(13,1434)=116.87; P < 0.0000) compared with the CF group. All symptoms were different between the groups apart from arthralgia and low-grade fever. Ninety-two percent of the Holmes criteria CFS patients were designated to be in the Holmes group using this symptom profile. However, 25% of the CF group were classified as complying with the Holmes criteria symptom profile. The CF patients who complied with the Holmes group symptom profile did not have an increased prevalence of any of the exclusionary diagnoses (fibromyalgia, idiopathic CF, depression, sleep disturbance related fatigue or OSAS).

The Fukuda criteria group also had a strong regression model (Wilks’λ=0.63; F(14,2060)= 84.64; P < 0.0000) compared with the CF group. All symptoms were different between the groups apart from arthralgia, headache and photophobia. Ninety-three percent of the Fukuda criteria CFS patients were designated to be in the Fukuda group using this symptom profile. However, 42.9% of the CF group were classified as complying with the Fukuda criteria. The CF patients who complied with the Fukuda group symptom profile were more likely to be diagnosed with fibromyalgia (odds ratio: 3.1; 95% CL 1.7–5.5; P < 0.00005) and less likely to be diagnosed with idiopathic chronic fatigue (odds ratio: 2.1; 95% CL 1.2–3.5; P < 0.008).

The Holmes criteria group had a strong regression model (Wilks’λ=0.80; F(14,1563)=27.37; P < 0.0000) compared with the patients added under the Fukuda criteria. All symptoms were different between the groups apart from arthralgia, fatigue, nonrestorative sleep, myalgia and photophobia. Eighty-one percent of the Holmes criteria CFS patients were designated to be in the Holmes group using this symptom profile. However, 58.5% of the additional Fukuda patients were excluded from the Holmes criteria group as a result of the differences in symptom profiles.

Group comparisons using an enlarged symptom definition

The different definition groups (Holmes, Fukuda, CF) were compared using standard discriminant function analysis with the Holmes criteria symptom profile with the addition of attention deficit, paralysis, new sensitivities to food/drugs, difficulties with words, urinary frequency, cold extremities, photophobia, muscle fasciculations, lightheadedness, exertional dyspnea and gastrointestinal disturbance. These symptoms were chosen as they represent the 10 symptoms with the greatest prevalence differences between the Holmes and Fukuda criteria as shown in Table 1.

Using this increased symptom set the Holmes criteria group had a strong regression model (Wilks’λ=0.45; F(23,1424)=74.47; P < 0.0000) compared with the CF group. In addition to the lack of significance of arthralgia and low-grade fever, five of the added symptoms were not significant in the multivariate model. Ninety-three percent of the Holmes criteria CFS patients were designated to be in the Holmes group using this enlarged symptom profile. However, using the enlarged definition symptom profile only 20.7% of the CF group were classified as complying with the Holmes criteria. This improved the specificity of the definition by 4.3%. The CF patients who complied with the Holmes group symptom profile did not have an increased prevalence of any of the exclusionary diagnoses (fibromyalgia, idiopathic CF, depression, sleep disturbance related fatigue or OSAS).

The Fukuda criteria group also had a strong regression model (Wilks’λ=0.62; F(23,2051)= 54.42; P < 0.0000) compared with the CF group. In addition to arthralgia, headache and photophobia, six of the extra symptoms were not different in the multivariate model. Ninety-three percent of the Fukuda criteria CFS patients were designated to be in the Fukuda group using this symptom profile. However, the same number of the CF group (42.9%) were classified as complying with the Fukuda criteria. Addition of the extra symptoms did not improve the specificity of the definition for the Fukuda CFS patients. The CF patients who complied with the Fukuda group symptom profile were once again more likely to be diagnosed with fibromyalgia (odds ratio: 2.5; 95% CL 1.4–4.4; P < 0.001) and less likely to be diagnosed with idiopathic chronic fatigue (odds ratio: 2.0; 95% CL 1.2–3.5; P < 0.01).

The Holmes criteria group had a strong regression model (Wilks’λ=0.80; F(14,1563)=27.37; P < 0.0000) compared with the added Fukuda patients. Arthralgia, fatigue, postexertional fatigue, nonrestorative sleep, myalgia, memory disturbance and photophobia as well as five of the 10 additional symptoms were no different. Eighty-one percent of the Holmes criteria CFS patients were designated to be in the Holmes group using this symptom profile. However, 59.3% of the extra Fukuda patients were excluded from the Holmes criteria group. Thus, the addition of the 10 symptoms (using severity scores) when comparing the Holmes and the additional Fukuda patients did not result in an improvement in the specificity by adding the additional symptoms (58.5% compared with 59.3%). Approximately 60% of the patients included in the CFS definition by the use of the Fukuda criteria would be excluded under the enlarged Holmes criteria.

Thus, the addition of 10 extra symptoms to the Holmes criteria results in a small increase in definition sensitivity and a much larger increase in specificity and improves the accuracy of the definitions.

The prevalence of patients with an increase in symptom severity of 2 or greater on the 0–3 symptom severity scales was calculated. Table 2 shows these symptoms in order of their ability to differentiate between the Holmes criteria patients and the patients included in the Fukuda criteria. This clearly shows that symptoms other than those used in the definition are important for differentiating between the Holmes criteria patients and the additional Fukuda and CF patients.

Table 2.   Odds ratio analysis of the prevalence of moderate to severe symptoms (≥2 on the 0–3 scale) in the Holmes (H), Fukuda included (F) and chronic fatigued (CF, non-CFS) patients. The symptoms have been listed in order of their ability to predict the difference between the Holmes (H) patients and the CFS patients included in the CFS definition using the Fukuda criteria (F) Thumbnail image of

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References

This study showed that analysis of individual symptom severity and prevalence revealed that the Holmes criteria patients had increased symptom prevalence and severity of many of the symptoms that determine the difference between CFS patients and CF subjects compared with the Fukuda defined group. Whilst the symptom prevalence and severity under the Holmes definition could be implied to be purely the result of the more stringent Holmes definition criteria they also indicate that the Holmes defined patients are the most severely affected CFS patients. The patients added to the original CFS group by the use of the Fukuda criteria can be differentiated from the Holmes criteria patients using simple statistical analyses. Approximately 60% of the added patients would be excluded with addition of 10 extra symptoms to the Holmes criteria. Importantly, data from this study may also suggest that patients with fibromyalgia may be more likely to be classified as having CFS if the Fukuda criteria are used irrespective of whether the original or extended definitions are used. Thus, the addition of patients to the CFS definition by the Fukuda criteria has resulted in the selection of less severely affected patients. This has also resulted in the introduction of an increase in patient symptom heterogeneity.

Komaroff et al. [17] suggested that eliminating three symptoms (muscle weakness, arthralgia and sleep disturbance) and adding two others (anorexia and nausea) would appear to strengthen the CDC case definition of CFS. Our observations would also suggest that addition of certain symptoms and removal of others would strengthen the ability to select CFS patients. The CFS symptoms that gave the best group differentiation were the Holmes criteria defining symptoms of fatigue, swollen/tender lymph nodes, sore throat, muscle weakness, recurrent flu-like symptoms, postexertional fatigue, myalgia, memory disturbance and nonrestorative sleep. The inclusion of 10 additional symptoms: hot flushes (in place of low-grade fever), paralysis, new sensitivities to food/drugs, cold extremities, gastrointestinal symptoms, difficulties with words, exertional dyspnea, attention deficit, urinary frequency, muscle fasciculations and light headedness increased the sensitivity of the Holmes definition by 0.5% and the specificity of the definition by 6%. As with the findings of Komaroff [17] the analysis used in this study also did not find arthralgia to be a significant predictive variable.

Virtually all the symptoms reported by the CFS population were increased in severity and prevalence compared with the CF population. This was also the case when comparing the Holmes and Fukuda groups. This could argue for incorporating a severity index of symptoms in the CFS Case definitions. This was already mentioned by Jason et al. [14] who suggested that there is a need to assess levels of symptom severity rather than just symptom occurrence alone and that future definitions of CFS may need to include specific guidelines pertaining to the importance of symptom severity in the diagnostic procedure. The establishment of symptom severity score requirements for the case definition of CFS could be a very important tool for CFS patients categorization. In this study a severity score of ≥2 of the 3-point scale using the Holmes defining symptoms plus the additional symptoms showed that certain of the additional symptoms could be important when used to differentiate between the Holmes and Fukuda criteria patients. The use of the multivariate method of analysis has demonstrated the importance of considering the relationship between symptoms and not simply increases in prevalence of individual symptoms. The increases in severity and prevalence of groups of symptoms has allowed an evaluation of the definitions and allowed changes to be suggested that may increase the ability to predict a CFS patient group. The problems with the multivariate methods did not appear to reduce the ability to interpret the changes identified. Thus a new or extended definition is required to improve the patient selection for CFS and this should be assessed using large populations of well-defined disease comparison groups, such as multiple sclerosis (MS), Rheumatoid Arthritis and major depression, as was carried out in the study by Komaroff [17].

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. References
  • 1
    Holmes G, Kaplan J, Gantz J et al. Chronic fatigue syndrome: a working case definition. Ann Intern Med 1988; 108: 3879.
  • 2
    Komaroff AL & Buchwald D. Symptoms and signs of chronic fatigue syndrome. Rev Infect Dis 1991; 13 (Suppl. 1): S8S11.
  • 3
    Bates D, Buchwald D, Lee J et al. A comparison of case definitions of chronic fatigue syndrome. Clin Inf Dis 1994; 18 (Suppl. 1): S115.
  • 4
    Schluederberg A, Straus SE, Peterson P et al. Chronic fatigue syndrome research. Ann Intern Med 1992; 117: 32531.
  • 5
    Fukuda K, Straus S, Hickie I, et al.The chronic fatigue syndrome: a comprehensive approach to its definition and study. Ann Intern Med 1994; 121: 9539.
  • 6
    Sharpe MC, Archard LC, Banatvala JE et al. Guidelines for research. J R Soc Med 1991; 84: 11821.
  • 7
    Lloyd AR, Hickie I, Boughton CR, Spencer O, Wakefield D. The prevalence of chronic fatigue syndrome in an Australian population. Med J Aust 1990; 153: 5228.
  • 8
    Komaroff AL & Geiger A. Does the CDC working case definition of chronic fatigue syndrome identify a distinct group? (Abstract). Clin Res 1989; 37: 778A778A.
  • 9
    Katon W & Russo J. Chronic fatigue syndrome criteria: a critique of the requirement for multiple physical complaints. Arch Intern Med 1992; 152: 16049.
  • 10
    Nicolson GL, Nasralla M, Haier J, Nicolson NL. Diagnosis and treatment of mycoplasmal infections in fibromyalgia and chronic fatigue. Biomed Ther 1998; 16: 26671.
  • 11
    Hickie I, Lloyd A, Wakefield D, Parker G. The psychiatric status of patients with chronic fatigue syndrome. Br J Psychiatry 1990; 156: 53440.
  • 12
    Bombardier C & Buchwald D. Chronic fatigue, chronic fatigue syndrome and fibromyalgia. Med Care 1996; 34: 92430.
  • 13
    Goldenberg DL, Simms RW, Geiger A, Komaroff AL. High frequency of fibromyalgia in patients with chronic fatigue seen in a primary care practice. Arthritis Rheum 1990; 33: 3817.
  • 14
    Jason LA, King CP, Frankenberry EL et al. Chronic fatigue syndrome: assessing symptoms and activity level. J Clin Psychol 1999; 55: 41124.DOI: 10.1002/(sici)1097-4679(199904)55:4<411::aid-jclp6>3.0.co;2-n
  • 15
    Jason LA, Ropacki MT, Santoro NB et al. A screening scale for chronic fatigue syndrome: reliability and validity. J Chronic Fatigue Syndrome 1997; 3: 3959.
  • 16
    Goldstein J. How do I diagnose a patient with chronic fatigue syndrome? In: Hyde BM, ed. The Clinical and Scientific Basis of Myalgic Encephalomyelitis/Chronic Fatigue Syndrome. Ottawa, ON, Canada: The Nightingale Research Foundation, 1992; 247–52.
  • 17
    Komaroff AL, Fagioli LR, Geiger AM et al. An examination of the working case definition of chronic fatigue syndrome. Am J Med 1996; 100: 5664.