Improving the diagnostic criteria and procedures for chronic fatigue
syndrome

Biological Psychology, Volume 68, Issue 2, Feb 2005, Pp 87-106

Caroline King [a] and Leonard A. Jason [b]

[a] Spinal Cord Injury Service (128), Hines VA Hospital, P.O. Box 5000,
Hines, IL 60141-5128, USA
[b] Department of Psychology, Northwestern Medical School, DePaul
University, 2219 N. Kenmore Ave., Chicago, IL 60614, USA

Received 9 September 2003;  accepted 19 March 2004.  Available online 4
June 2004.


Abstract
Since the publication of the case definition for chronic fatigue syndrome
(CFS) in 1988 the diagnostic criteria have been revised twice in the U.S.
None of the case definitions were derived empirically. As a result, there
is concern regarding the sensitivity, specificity, and reliability of the
criteria. The goal of the present study was to identify methods for
improving the diagnostic criteria for CFS. Three groups of 15
participants each were recruited: participants with (1) CFS, (2) major
depressive disorder (MDD), and (3) healthy controls. Using statistical
procedures, three methods for improving the diagnostic criteria were
explored: identification of new diagnostic symptoms, the use of severity
ratings for symptomatology, and the identification of standardized
measures that differentiate cases of CFS from other conditions. Results
of the present study suggest that these three methods hold promise for
improving the sensitivity, specificity, and reliability of the diagnostic
criteria for CFS.

Author Keywords: Chronic fatigue syndrome; Assessment; Diagnostic
procedures; Symptomatology; Diagnostic criteria


One of the main goals of classifying any disease or illness is to group
together patients who have an illness that may have many manifestations,
but a common underlying pathophysiological pathway (Hartz et al., 1998).
The benefit of classifying patients into diagnostic categories is that it
facilitates communication among clinicians/researchers, selection of
treatment methods, and prediction of response to treatment. Past
experience has shown that even in cases where the underlying
pathophysiological pathway has not been identified, research on the
etiology and treatment of the illness has been facilitated by simply
classifying these illnesses as syndromes of signs and symptoms (e.g.,
systemic lupus erythematosus or tuburcleousis). This has been the case
with chronic fatigue syndrome. Although the etiology of this illness
remains unknown, researchers have been able to examine and better
understand the nature of this illness primarily through the use of
clinical classification approaches (i.e., classification criteria
developed through clinical experience and observation).

In 1988, a team of experienced clinicians led by the Centers for Disease
Control and Prevention developed the first set of clinically derived
diagnostic criteria for chronic fatigue syndrome. These criteria,
developed through the consensus of an expert committee, provided health
care professionals with the first set of systematic criteria to follow
when assessing patients presenting with unexplained chronic fatigue.
Shortly following the publication of the Holmes et al. (1988) case
definition, researchers and clinicians in the United States became
dissatisfied with this set of diagnostic criteria ( Jason et al., 1997).
Since then, the case definition has been revised twice: once in 1992 by a
group who attended the 1991 National Institute of Allergy and Infectious
Disease/National Institute of Mental Health workshop on CFS, and a second
time in 1994 by the NIH/CDC CFS study group ( Fukuda et al., 1994). The
criteria published by the NIH/CDC CFS study group ( Fukuda et al., 1994)
is the current U.S. case definition for CFS.

It is important to note that neither the original U.S. case definition
nor the revised U.S. case definitions for CFS were derived empirically
(Jason et al., 1997). Over the past 4–5 years researchers have become
interested in attempting to validated the current U.S. case definition
through empirical and statistical approaches. Overall, the results of
these studies have suggested that there is moderate to strong empirical
support for the current CFS case definition ( Hartz et al., 1998 and
Jason and Taylor, 2002; Jason et al., 2002a and Jason et al., 2002b;
Komaroff et al., 1996 and Nisenbaum et al., 1998). There is some concern,
however, regarding the sensitivity (i.e., ability to identify those who
have the disease), specificity (i.e., ability to correctly identify those
who do not have the disease), and diagnostic reliability of the Fukuda et
al. (1994) criteria.

Some CFS researchers are concerned that the specificity of the current
U.S. case definition is poor (Jason et al., 1997). Even Fukuda, one of
the primary authors of the U.S. case definition, has stated that the
current CFS diagnostic criteria might not exclude people who have purely
psychosocial stress, or many psychiatric reasons for their fatigue
(Fukuda, personal communication, August 30, 1995). As a result,
individuals with purely psychiatric disorders and psychological
explanations for their fatigue might be included within the CFS rubric.
Although it is possible for some individuals with CFS to have psychiatric
problems before or after the onset of CFS, or even both, the inclusion of
individuals with purely psychiatric disorders may seriously complicate
the interpretation of epidemiological and treatment studies ( Jason et
al., 1997).

One approach to improving the specificity as well as the sensitivity of
the diagnostic criteria for CFS is through the development of empirically
derived symptom criteria. Researchers attempting to empirically validate
the current U.S. case definition have already made some initial
suggestions regarding specific symptoms that should be added or removed
to improve the overall sensitivity and specificity of the criteria. In a
study conducted by Komaroff et al. (1996), patients meeting the major
criteria of both the original CFS case definition ( Holmes et al., 1988)
and the most recently revised CSF case definition ( Fukuda et al., 1994)
were compared to healthy controls and two clinical populations with
fatigue: patients with multiple sclerosis (MS), and patients with
depression. Komaroff et al. (1996) examined the occurrence of the minor
symptom criteria as well as the occurrence of several other medical
symptoms not included in the case definition between the four study
groups. Results of the study revealed that most of the minor criteria
symptoms of the U.S. case definition were found to discriminate patients
with CFS from patients with MS, depression, and healthy controls.
However, the following three diagnostic symptoms were identified as being
poor discriminators among the four study groups: muscle weakness,
arthralgias, and sleep disturbances. Komaroff et al. (1996) therefore
recommended that these items be omitted from future revisions of the CSF
case definition. The authors also found that two additional symptoms
currently not included in the case definition had good discriminatory
power among the four study groups: anorexia and nausea. Komaroff et al.
(1996) suggested adding these two symptoms to the case definition to
improve the specificity and sensitivity of the diagnostic criteria.

In a separate study, Hartz et al. (1998) examined persons with CFS and
compared them to persons with idiopathic fatigue and persons with no
symptoms of fatigue. Similar to Komaroff et al. (1996), Hartz and
colleagues found support for the symptoms included in the current case
definition. However, Hartz et al. (1998) also found additional symptoms
that separated participants with CFS from participants with idiopathic
fatigue and controls: frequent fever and chills, muscle weakness, and
sensitivity to alcohol. Based on their findings, the authors recommended
including these additional symptoms in the current U.S. case definition.

Lastly, a study conducted by Jason et al., 2002b and Jason et al., 2002b
found additional support for the current case definition. In comparison
to controls, individuals with CFS reported significantly higher
frequencies of all eight Fukuda et al. (1994) definitional symptoms.
However the authors found several other symptoms that occurred with
higher frequency and uniquely differentiated the CFS group from controls
that are not included within the Fukuda et al. (1994) criteria. These
symptoms included: shortness of breath, chest pain, dizziness after
standing, skin sensations, general dizziness, dizzy moving the head, and
alcohol intolerance.

A second approach to improving the sensitivity and specificity of the
criteria is the use symptom severity ratings. At present, the symptom
criteria for the case definition are scored as either being present or
absent with no consideration given to the severity of the symptoms.
Research suggests that this scoring system is problematic because many of
the symptoms included in the diagnostic criteria for CFS are commonly
experienced by people at one time or another (Denche et al., 1996). For
example, symptoms of fatigue, sore throat, headache, muscle pain, and
post-exertional malaise are frequently experienced by people who have a
cold or the flu. Denche et al. (1996) demonstrated the potential for
control participants to be misdiagnosed with CFS when using Fukuda et al.
(1994) symptom criteria alone. In a sample of healthy adults, 15% met
Fukuda et al. (1994) symptom criteria for CFS (i.e., complained of four
or more of the eight specified symptoms) ( Denche et al., 1996).

The symptoms included in the CFS case definition are also common to many
other fatiguing medical illnesses and psychiatric conditions. For
example, symptoms of fatigue, headache, unrefreshing sleep, muscle pain,
and impaired memory and concentration frequently experienced by people
with multiple sclerosis and major depression. Again, because of the
degree of symptom overlap, it is very possible for a person with either
MS or major depression to fulfill the symptomatic criteria of the current
CFS case definition when only symptom occurrence is measured. Although
many of the symptoms of CFS are common to many conditions, it is possible
that the severity at which these symptoms are experienced by individuals
with CFS is not. Use of symptom severity ratings with cutoff scores may
therefore help to differentiate CFS from other illnesses with similar
symptoms.

There is preliminary evidence that the use of severity ratings may
provide a way to improve the specificity of the current case definition.
Jason et al. (2000) found that symptom severity ratings were useful for
distinguishing individuals with CFS from individuals with a fatiguing
psychiatric illness (i.e., melancholic depression). Jason et al. (2000)
found when comparing the symptomatic criteria of participants with CFS to
participants with melancholic depression, only one significant difference
emerged between the groups. In other words, the occurrence of the Fukuda
et al. (1994) symptomatic criteria was very similar between the two
groups. However, when a symptom severity ratings were used, and a rating
of 40 or higher was used as a scoring rule for determining whether the
symptom fulfilled the diagnostic criteria, four significant differences
emerged between the two groups. These results demonstrate the utility of
severity ratings as a means for improving the specificity of the
diagnostic criteria.

Lastly, the absence of objective assessment approaches for the diagnostic
criteria has generated concern regarding the reliability of the U.S. case
definition. Researchers have noted that the case definition for CFS has
been "frequently modified in practice because some of the criteria are
difficult to interpret or to comply with" (Fukuda et al., 1994; p. 954).
Because no laboratory tests or objective indicators for CFS exist, case
identification depends primarily on information obtained through clinical
interviews. Although the clinical interview is often an integral
component to any assessment process, if the interview does not follow a
standardized format, the results of the interview can be quite variable
across examiners and different diagnostic conclusions may be reached (
Matarazzo, 1983). One approach to improving the diagnostic reliability of
the current U.S. CFS case definition is the identification of
standardized measures with scoring guidelines to be used in conjunction
with the clinical interview. The addition of standardized measures with
scoring guidelines would likely improve the reliability of diagnostic
decisions by providing clinicians with objective standards to follow when
assessing the various features of this syndrome.

In an effort to identify an objective method for discriminating CFS from
major depression, Johnson et al. (1995) administered the Beck depression
inventory (BDI) to people with CFS and people with major depression.
Items from the BDI were categorized into one of four symptom categories
(mood, self-reproach, somatic, and vegetative) and compared among the two
groups. Significant differences were found in the qualitative nature of
the symptoms endorsed on the BDI by people with CFS and people with major
depression. The BDI scores of people with CFS were comprised mainly of
items concerning physical complaints and somatic symptoms of fatigue.
Symptoms of disturbed mood and self-reproach, two cardinal signs of
depression, were not reported as frequently by the participants with CFS
as by the participants with depression ( Johnson et al., 1995). These
findings demonstrate that while depressive symptoms are common in samples
of people with CFS and depression, the types of items involved are
qualitatively different. The results of this study suggest that
incorporating a standardized measure, such as the BDI, into the
diagnostic procedure for CFS may help clinicians distinguish cases of CFS
from case of major depression.

The use of other standardized measures of functioning would likely
improve the diagnostic accuracy and reliability of CFS. For example, the
SF-36 Health Survey (Ware and Sherbourne, 1992) is a standardized
questionnaire that has been widely used to assess functioning in patients
who have a variety of medical conditions. For many conditions, distinct
SF-36 profiles have been identified (e.g., cardiovascular disease, major
depression). Buchwald et al. (1996) have demonstrated that the SF-36 may
also be a useful and reliable instrument for assessing functional status
in patients with CFS as well as distinguishing such patients from
patients with other fatiguing conditions. Additional studies replicating
the results of the Buchwald et al. (1996) study are needed to establish a
unique SF-36 profile for CFS patients.

The present study was conducted to explore three methods for improving
the diagnostic criteria and procedures for CFS. First, symptomatology was
compared among patients with CFS, major depressive disorder (MDD), and
healthy controls to explore which symptoms currently included in the case
definition differentiate cases of CFS from MDD and controls, and to
identify new symptoms (i.e., not currently included in the case
definition) that differentiate CFS from these conditions. Second, the use
of symptom severity ratings was examined as an additional means for
distinguishing CFS from MDD and controls. Third, the use of two
standardized measures, the BDI and SF-36 was evaluated to determine
whether these instruments were useful for identifying cases of CFS and
differentiating CFS from MDD and controls.

1. Method
1.1. Research participants
A total of 45 individuals (15 with CFS, 15 with major depression, and 15
healthy controls) were recruited from the Greater Chicago area for the
present study. Fifteen participants with CFS were solicited to
participate in the present study. Participants were drawn from two
sources, a local CFS support group in Chicago and previous research
studies conducted at DePaul University. Participants were required to
have been diagnosed with CFS, using the Fukuda et al. (1994) diagnostic
criteria, by a Board-certified physician and were required to have a
current (active) case of CFS. All participants had been seen by their
physician in the past year. Individuals who reported having
uncontrolled/untreated medical illnesses (e.g., untreated anemia) were
excluded. All participants were screened with the SCID-IV (to be
described later) to ensure that they did not have any exclusionary
psychiatric illnesses as stipulated by the Fukuda et al. (1994) case
definition.

Fifteen participants with a diagnosis of major depression were solicited
from a local chapter of the National Depressive and Manic Depressive
support group in Chicago. Participants were required to have been
diagnosed with major depression by a licensed psychologist or
psychiatrist. All participants were screened with the SCID-IV to ensure
that they met criteria for a current (active) case of major depression
and did not have any other current psychiatric illnesses. Individuals who
had other current psychiatric conditions in addition to major depression
were excluded. Individuals who reported having uncontrolled/untreated
medical illnesses (e.g., anemia, diabetes) were also excluded.

Finally, fifteen healthy control participants were solicited from the
Greater Chicago area. Individuals who did not have any medical illnesses
or who did not have any uncontrolled/untreated illnesses were allowed to
participate. Individuals with uncontrolled/untreated medical illnesses
(e.g., anemia, diabetes) were excluded. All participants were screened
with the SCID-IV to ensure that they did not have any current psychiatric
illnesses. Individuals with current psychiatric conditions were excluded.


1.2. Procedure
All 45 participants were initially screened by a trained interviewer to
determine if they met the inclusion and exclusion criteria for the group
condition they were being considered for (i.e., CFS, depression, healthy
control). As part of this screening process, all participants were
administered the SCID-IV to assess for psychiatric conditions.

Participants who met criteria for participation were asked to complete a
battery of questionnaires that measured demographics, social, emotional,
and physical functioning, activity level, depression, and a comprehensive
list of physical, cognitive, and emotional symptoms.

1.3. Measures
1.3.1. Demographic variables
Basic demographic data was gathered, including age, ethnicity, marital
status, gender, occupation, work status, and educational level.

1.3.2. Socioeconomic status (SES)
This was measured by using occupation and highest educational level to
compute the revised Hollingshead scale (Wasser, in press) of
socioeconomic status. The scores on the Hollingshead scale ranged from 5
to 54, with lower scores indicating lower socioeconomic status, and
higher scores indicating higher socioeconomic status.

1.3.3. Beck depression inventory (BDI)
This is a self-rating scale which evaluates 21 symptoms related to
depression on a scale of 0 (absent) to 3 (most severe). Internal
consistency for the BDI ranges from 0.73 to 0.92 with a mean of 0.86.
(Beck et al., 1988). The BDI demonstrates high internal consistency, with
alpha coefficients of 0.86 and 0.81 for psychiatric and non-psychiatric
populations, respectively ( Beck et al., 1988). The BDI has a split-half
reliability coefficient of 0.93 ( Beck et al., 1988).

Research has shown that the individual questions on the BDI can be
divided into four categories: mood, self-reproach, somatic, and
vegetative characteristics of depression (Huber et al., 1990 and Huber et
al., 1993). The mood category included six items: #1, #2, #4, #10, #11,
and #12. The self-reproach category included seven items: #3, #5, #6, #7,
#8, #9, and #13. The somatic category consisted of the following three
items: #14, #15, and #20; and the vegetative category consisted of the
following four items: #16, #18, #19, and #21.

1.3.4. Medical outcomes study SF-36 (SF-36)
The SF-36 is 36-item instrument that is comprised of multi-item scales
that assess physical functioning, role limitations, social functioning,
bodily pain, general mental health, vitality, and general health
perceptions. Higher scores indicate better health, lower disability, or
less impact of health on functioning. Reliability and validity studies
have demonstrated that the 36-item version of the SF-36 has high
reliability and validity in a wide variety of patient populations
(Stewart et al., 1989). The SF-36 has also shown adequate psychometric
properties as a measure of functional status in a population of CFS
participants and is capable of distinguishing CFS from other fatiguing
illnesses ( Buchwald et al., 1996).

Principal component analytic studies of the MOS SF-36 have revealed that
the eight scales of MOS SF-36 comprise two distinct factors, a physical
component and a mental component. These two components have been found to
be valid and reliable summary measures of the MOS SF-36 and account for
80–85% of the reliable variance in the eight SF-36 scales.

The physical component scale (PCS) is a summary measure of the following
SF-36 scales: physical functioning, role-physical, bodily pain, and
general health. The physical component correlates most highly with the
physical functioning, role-physical, and bodily pain scales, which
contribute most to scoring of the PCS measure of that component. The
mental health component scale (MCS) is a summary measure of the following
SF-36 scales: vitality, social functioning, role-emotional, and mental
health. Three scales (social functioning, role-emotional, and mental
health) correlate most highly with the mental component and contribute
most to scoring of the MCS measure of that component. Three of the SF-36
scales have notable correlations with both components. The vitality scale
correlates highly with both components. The general health scale
correlates with both components, but has a stronger correlation with the
physical component. The social functioning scale also correlates with
both components, however it is more highly correlated with the mental
component.

1.3.5. Physical, cognitive, and emotional symptom checklist
Participants were asked to indicate whether or not they had a number of
somatic, cognitive, and emotional symptoms commonly experienced by people
with CFS. Symptoms on this list were taken from a variety of sources,
including a measure developed by Komaroff et al. (1996), the current U.S.
CFS case definition ( Fukuda et al., 1994), and the results of studies by
Hartz et al. (1998) and Komaroff et al. (1996) that suggested the
inclusion of new symptoms in the case definition.

For each symptom, participants were asked to indicate if the symptom had
been present for 6 months or longer, if the symptom began before the
onset of their fatigue or health problems, and how often the symptom is
experienced. Participants were also asked to rate the intensity of each
symptom they endorsed on a scale of 0–100, where 0 = no problem and 100 =
the worst problem possible.

Included in this list were fatigue and the eight diagnostic symptoms
specified by Fukuda et al. (1994). Again, participants were asked to
report if each symptom had been present for 6 months or longer, began
before the onset of their fatigue or health problems, how often it is
experienced, and rate the intensity of each symptom on the same scale of
0–100.

1.3.6. The structured clinical interview for the DSM-IV (SCID)
The SCID is a valid and reliable semi-structured interview guide that
closely resembles a traditional psychiatric interview (First et al.,
1996). The SCID is designed to identify current, past, and lifetime
(chronic or reoccurring, current, and past) diagnoses for a majority of
DSM-IV, Axis I psychiatric disorders. The SCID is commonly administered
during a single session lasting from 45 min to 1 h. Diagnostic decisions
generated by the SCID are based on all possible sources of historical,
symptomatic, and behavioral information. The SCID begins with a
semi-structured interview portion designed to yield a tentative
diagnosis. The tentative diagnosis is then systematically assessed during
the structured portion of the interview through the use of embedded
questions that conform to the exact, Axis I criteria set forth by the
DSM-IV.

2. Results
Demographic characteristics of the three groups of participants are shown
in Table 1. Sociodemographic data were compared across the three groups
using Pearson’s Chi-square for dichotomous and multinomial data and
analysis of variance for age and SES. Exact significance levels were used
for the Chi-square analyses because of the occurrence of low cell
frequencies in many of the analyses. Exact significance is the
significance level based on the exact distribution of a test statistic.
Exact significance levels are preferable when the data set is small,
sparse, contains many ties, or is poorly distributed (Norusis, 2000).
There were no significant differences between groups with respect to
gender, race, age, SES, education, marital status, occupation, work
status, and additional roles.



Table 1. Demographic variables across groups




In regard to psychiatric co-morbidity, in the CFS group 3 (20%)
participants met DSM-IV diagnostic criteria for dysthymia. No other
current diagnoses were detected in the CFS group. In the MDD group all 15
(100%) participants met DSM-IV diagnostic criteria for MDD. None of the
participants in the MDD group met criteria for MDD with catatonic,
melancholic, psychotic, or atypical features. Participants in the MDD
group did not meet criteria for any other Axis I disorders. None of the
participants in the control group met criteria for any Axis I disorder.

2.1. Comparison of symptomatology between groups and use of severity
ratings
Two series of analyses were conducted to identify symptoms with the
potential to discriminate CFS from MDD and healthy controls. In the first
series of analyses, frequencies of the 66 symptoms from the physical,
cognitive, and emotional symptom checklist were compared among the three
study groups using pairwise Fisher’s exact test. These comparisons are
also shown in Table 2. Only differences at the P<0.01 level were reported
to control for the high number of comparisons. The CFS group differed
most from the control group in that 37 of the 66 symptoms were found to
occur significantly more frequently in the CFS group. In comparing the
CFS group to the MDD group, 16 of the 66 symptoms were occurred
significantly more frequently in the CFS group. Lastly, in comparing the
MDD group to the control group, five symptoms occurred significantly more
frequently in the MDD group.



Table 2. Percent of participants reporting symptom as present for 6
months or longer and pairwise Fisher’s exact test comparisons across the
three groups


If small letter occurs for two groups, they are significantly different.


When looking specifically at the occurrence of fatigue and the eight
symptoms of the current U.S. case definition, only one symptom,
post-exertional malaise, was found to occur significantly more frequently
in the CFS group in comparison to the MDD group. When comparing the CFS
and control group, fatigue and all eight of the symptoms were found to
occur significantly more often in the CFS group. Comparisons between the
MDD and control group revealed that fatigue, unrefreshing sleep, impaired
memory and concentration, headaches, and muscle pain all occurred
significantly more in the MDD group.

None of the 66 symptoms from the physical, cognitive, and emotional
symptom checklist (shown in Table 2) were found to separate all three
groups when looking at symptom occurrence. However, three symptoms in the
fatigue/weakness group (post-exertional malaise, muscle weakness, feeling
unsteady on feet), five symptoms in the neuropsychological category (need
to focus on one thing at a time, confusion/disorientation, difficulty
finding the right word, frequently losing train of thought, slowness of
thought), three symptoms in the infectious category (hot and cold spells,
feeling like having a temperature, feeling chilled/shivery), one in the
cardiopulmonary (chest pain), and one in the gastrointestinal category
(upset stomach), were found to occur significantly more in the CFS group
than either the MDD or control group.

2.2. Severity ratings
A second series of analyses was conducted to further explore symptoms
that differentiate the three study groups. For these analyses, symptom
severity ratings were compared across the three groups for 66 symptoms
using one-way ANOVAs. Tukey Honest significant difference (HSD) post hoc
analyses were performed on all significant ANOVAs at the P<0.01 level.
The results of these analyses are shown in Table 3. Once again, the CFS
group differed most from the control group in that 31 of the 66 symptoms
were reported as being significantly more severe in the CFS group. In
comparison to the MDD group, 17 symptoms were reported as being
significantly more severe in the CFS group. Only three symptoms were
reported as being significantly more severe in the MDD group in
comparison to the control group.



Table 3. ANOVA: one-way analysis of variance for symptom severity ratings
across the three groups, means and Tukey HSD post hoc comparisons


If small letter occurs for two groups, they are significantly different.




In examining the symptom severity ratings for fatigue and the eight
symptoms of the U.S. case definition, six symptoms were reported as being
significantly more severe in the CFS group than the MDD group: fatigue,
post-exertional malaise, unrefreshing sleep, sore throat, tender/sore
lymph nodes, muscle pain, and pain in multiple joints without swelling or
redness. In comparing the CFS and control group, fatigue and all eight
symptoms were found to be significantly more severe in the CFS group.
Comparisons between the MDD and control group revealed that two symptoms
were reported as significantly more severe in the MDD group: unrefreshing
sleep and impaired memory and concentration.

Only one symptom, unrefreshing sleep, was found to be significantly
different across all three groups. Comparing the CFS group against the
MDD group and control group revealed significant differences for the
following items: four symptoms in the fatigue/weakness group (fatigue,
post-exertional malaise, muscle weakness, need to nap each day), three
symptoms in the neuropsychological category (frequently losing train of
thought, difficulty finding the right word, confusion/disorientation),
four symptoms in the infectious category (sore throat, tender lymph
nodes, hot and cold spells, feeling chilled/shivery), three in the
rheumatological category (muscle pain, pain in multiple joints without
swelling, night sweats), one in the cardiopulmonary (shortness of
breath), and one symptom in the neurological category (blurred vision).

2.3. Comparison of standardized measures between groups
Overall composite and subscale scores on the BDI and SF-36 were compared
between groups to test these instruments ability to distinguish cases of
CFS, MDD, and healthy controls.

2.4. BDI analyses
ANOVA was used to examine differences in the total BDI score across the
three groups. A significant main effect for group membership on total BDI
score was found (F(2, 42)=17.39, P=0.01). The MDD group had the highest
mean total BDI score (M=17.67), indicative of moderate to severe
depression, while the control group had the lowest score (M=4.20) and
were not clinically depressed. The CFS group had an intermediate score
(M=14.53) indicating mild depression according to the BDI classification
system. Post hoc analysis using Tukey HSD test at the 0.05 level revealed
that the control group’s total BDI score was significantly lower from the
CFS and MDD group (P<0.00). Significant differences were not found
between the total BDI scores for the CFS and MDD groups.

A second analysis on the BDI was conducted to examine possible
differences in the type of depressive symptoms (mood, self-reproach,
somatic, and vegetative) endorsed by participants in the CFS, MDD, and
control groups. Specifically, multivariate analysis of variance (MANOVA)
was used to examine whether of group membership influenced participants’
scores on the four symptom categories (mood, self-reproach, somatic, and
vegetative). Group membership was found to have a significant effect on
all outcome variables, as measured by Wilks’ Lambda (F(2, 42)=6.11,
P=0.01). Significant differences were found between the three groups in
reporting mood, self-reproach, somatic, and vegetative symptoms of
depression. Tukey HSD post hoc analysis at the 0.05 level was used to
examine group differences on the four symptom categories. The CFS group
reported significantly less severe symptoms of self-reproach (M=3.33) in
comparison to the MDD group (M=6.33), yet comparable levels of mood (CFS
group M=3.67, MDD group M=4.73), somatic (CFS group M=4.33, MDD group
M=3.93), and vegetative symptomatology (CFS group M=3.20, MDD group
M=2.67). In comparison to the control group, the CFS group exhibited
significantly higher levels of symptomatology for all four symptom
categories. The MDD group in comparison to the control group had
significantly higher levels of self-reproach (control group M=0.53), mood
(control group M=0.93), and somatic symptomatology (control group
M=1.47), but did not differ with respect to vegetative symptoms (control
group M=1.27).

2.5. SF-36 analyses
The effect of group membership on level of functioning was examined using
MANOVA with the eight subscales of the SF-36 as outcome variables. In
examining the main effect of the independent variable, group membership
was found to have a significant effect on the outcome variables, as
measured using Wilks’ Lambda (F(2, 42)=7.45, P=0.01). Between-subject
tests found that group membership predicted all eight outcome variables.
Tukey HSD post hoc tests at the 0.05 level were used to examine the
differences between the three groups (CFS, depressed, and controls) on
all eight dependent variables. Using Tukey HSD to compare differences
between the three groups for the physical functioning scale, significant
differences were found between the CFS group and the two other groups.
CFS participants were found to have significantly lower scores on the
physical functioning scale (M=44) in comparison to the MDD group
(M=70.33, P=0.01) and the control group (M=88.0, P=0.01). Significant
differences were not found between the MDD and control group on this
scale.

When using the Tukey HSD post hoc test to examine differences between the
three groups on the physical role functioning scale, significant
differences were found between all three groups. Participants with CFS
had the lowest score on this scale (M=13.33), followed by participants
with MDD who had an intermediate score (M=43.33, P=0.039), with control
participants having the highest score on this scale (M=90.0, P=0.01).

Post hoc analysis of the general health scale revealed significant
differences between all three groups. The CFS group had the lowest score
on this scale (M=31.60), followed by MDD participants who had an
intermediate score (M=49.93), with the control group having the highest
score on this scale (M=66.67).

Similarly, post hoc analysis of the vitality scale revealed significant
differences between all three groups. Specifically, the CFS group
(M=14.0) scored significantly lower on the vitality scale than the MDD
group (M=44.0, P=0.01). In turn, the MDD group scored significantly lower
on the vitality scale than the control group (M=66.67, P=0.01).

Tukey HSD post hoc analysis of the social functioning scale also revealed
significant differences between all three groups. The CFS group (M=35.0)
scored significantly lower on the social functioning scale than the MDD
group (M=60.83, P=0.01). In turn, the MDD group scored significantly
lower on the social functioning scale than the control group (M=90.83,
P=0.01).

Post hoc analysis of the emotional role functioning scale revealed
significant differences between the control group and the other two
groups. Control participants were found to have significantly higher
scores on this scale (M=91.11) in comparison to CFS participants
(M=48.89, P=0.01) and MDD participants (M=37.78, P=0.01). The CFS and MDD
group did not differ significantly from one another on this scale.

Significant differences were also found between the control group and the
other two groups on the bodily pain scale. Using the Tukey HSD test,
control participants were found to have significantly higher scores on
this scale (M=76.27) in comparison to CFS participants (M=38.53, P=0.01)
and MDD participants (M=55.60, P=0.041). The CFS and MDD group did not
differ significantly from one another on this scale.

Using the Tukey HSD test, significant differences were found between the
control group and the other two groups on the mental health scale. The
control group scored significantly higher (M=79.73) on the mental health
scale than both the CFS group (M=39.27, P=0.04) and the MDD group
(M=39.13, P=0.17). Significant differences were not found between the CFS
and the MDD group on the mental health scale.

A second MANOVA was run on to evaluate the influence of group membership
the two summary scale scores of the SF-36: the physical component scale
and the mental component scale. The results of this analysis revealed
that group membership did have a significant main effect on the outcome
variables, as measured using Wilks’ Lambda statistic (F(2, 42)=13.69,
P=0.01) Between-subject tests found that group membership predicted both
the physical component and mental component scale scores.

Tukey HSD post hoc analyses were conducted to compare differences between
the three groups on the physical component and mental component scales.
Examination of group differences on the physical component scale revealed
that the CFS group (M=28.52) scored significantly lower on this scale in
comparison to the MDD group (M=42.32, P=0.01) and the control group
(M=51.21, P=0.01). The MDD group and control group did not differ
significantly from one another.

Tukey HSD post hoc analysis of group differences on the standardized
mental component scale revealed that the control group scored
significantly higher on this scale (53.37) than the CFS group (M=39.27)
and the MDD group (M=39.13). Significant differences were not found
between the CFS and MDD group.

3. Discussion
The overarching goal of the present study was to identify methods for
improving the diagnostic reliability of CFS. Three approaches for
improving the diagnostic reliability of CFS were explored. First, the
identification of new symptoms that differentiate CFS from other
conditions (i.e., MDD, healthy controls) was explored. Interestingly,
results of the analyses examining differences between the groups for the
symptom occurrence and severity data did not provide support for previous
research in this area by Komaroff et al. (1996). Komaroff et al. (1996)
suggested that symptoms of muscle weakness, arthralgias, and sleep
disturbances be eliminated from the diagnostic criteria. In the present
study, reports of muscle weakness, arthralgias, and sleep disturbances
were significantly different amongst the three study groups.

Muscle weakness was reported by 93% of the CFS group and was found to
occur significantly more in the CFS than either the MDD or controls
groups. Muscle weakness was also significantly more severe in the CFS
group in comparison to the MDD or control groups. Muscle weakness was
therefore found to be a good discriminator among the three groups. Pain
in multiple joints without swelling or redness (arthalgias) was reported
by 73% of the CFS group and was found to occur significantly more in the
CFS group than the control group. Severity ratings for pain in multiple
joints without swelling or redness were also significantly higher in the
CFS group as compared to the MDD and controls groups. Thus, when looking
at severity data, this symptom also discriminated among the three groups.
Lastly, unrefreshing sleep was reported by 100% of the CFS group and
difficulty staying asleep was reported by 73% of the CFS group. Problems
with unrefreshing sleep occurred significantly more in the CFS group than
the control group and were rated as being significantly more severe in
the CFS group in comparison to the MDD and controls groups. Again, when
using severity data, unrefreshing sleep discriminated among the three
groups. Overall, these findings suggest that muscle weakness, pain in
multiple joints without swelling or redness, and unrefreshing sleep may
be key symptoms for distinguishing CFS from other conditions, especially
when symptom severity is taken into consideration.

Komaroff et al. (1996) also suggested the addition of nausea as one of
the diagnostic symptoms. Results of the present study do not support the
inclusion of nausea as this symptom was reported by only 20% of the CFS
group and differences between the three groups were not found in terms of
symptom occurrence and severity. Hartz et al. (1998) also made
recommendations regarding the inclusion of fever and chills, and
sensitivity to alcohol. Neither symptoms of fever and chills nor
sensitivity to alcohol were found to discriminate between the CFS, MDD,
and control groups in the present study. These results do not support the
inclusion of these latter two symptoms in the case definition.

The second approach taken in the present study to improve the specificity
and reliability of the diagnostic criteria for CFS was the use of symptom
severity ratings. In the present study symptom occurrence and symptom
severity were examined to explore differences in the ability of
occurrence data and severity ratings to discriminate among the three
groups, and to explore whether there are symptoms which differentiate the
three groups that are not currently included in the diagnostic criteria
for CFS. Symptom occurrence was examined first for fatigue and the eight
symptoms of the current U.S. case definition.

The analyses examining differences in symptom occurrence of fatigue and
the eight symptoms of the current U.S. case definition amongst
individuals with CFS, MDD, and healthy controls revealed several
interesting findings. First, both the CFS group and the MDD group
reported significantly more symptoms than the control group. Fatigue and
all eight of the symptoms were found to occur significantly more often in
the CFS group than control group and fatigue and four other symptoms
occurred significantly more in the MDD group than control group. Second,
only one symptom, post-exertional malaise, was found to occur
significantly more frequently in the CFS group in comparison to the MDD
group. Thus, in comparison to the control group and with one another, the
CFS and MDD groups were similar. Third, when looking at symptom
occurrence alone, some individuals in the MDD group met the in
inclusionary criteria for the current U.S. case definition for CFS (i.e.,
the presence of fatigue plus at least four of the eight symptoms
diagnostic symptoms). Taken together, these findings strongly suggest
that when using symptom occurrence alone, there is a strong potential for
individuals with MDD to be misclassified as having CFS.

The analyses examining differences in symptom severity ratings for
fatigue and the eight symptoms of the U.S. case definition amongst
individuals with CFS, MDD, and healthy controls demonstrated an improved
ability to distinguish the CFS group from the MDD group than when symptom
occurrence was used. Six symptoms were reported as being significantly
more severe in the CFS group than the MDD group. These results are quite
different from the occurrence data in which only one symptom
distinguished the CFS group from the MDD group. Comparisons between the
MDD and control group revealed that only two symptoms were reported as
significantly more severe in the MDD group as opposed to four symptoms
when the occurrence data was used. This finding is important and
represents and an improvement over the use of the occurrence data because
the MDD group no longer meets the inclusionary criteria for CFS. These
results suggest that the use of symptom severity ratings may improve the
diagnostic specificity of the current U.S. case definition for CFS.

In an effort to identify new symptoms that might help distinguish CFS
from other conditions, group differences were examined for symptom
occurrence and symptom severity using the remaining 57 symptoms from the
physical, cognitive, and emotional checklist. When looking at symptom
occurrence, the following 12 symptoms were found to occur more frequently
in the CFS group than either the MDD or control group: muscle weakness,
feeling unsteady on feet, need to focus on one thing at a time,
confusion/disorientation, difficulty finding the right word, frequently
losing train of thought, slowness of thought, hot and cold spells,
feeling like having a temperature, feeling chilled/shivery, chest pain,
and upset stomach. However, when looking at symptom severity, only 10
symptoms were found to occur more frequently in the CFS group than either
the MDD or control group: muscle weakness, need to nap each day,
frequently losing train of thought, difficulty finding the right word,
confusion/disorientation, hot and cold spells, feeling chilled/shivery,
night sweats, shortness of breath, and blurred vision. Because severity
ratings were demonstrated in the previous set of analyses to be superior
discriminators, it is suggested that emphasis be given to the results of
the analyses examining severity ratings for the additional 57 symptoms
rather than the analyses looking at occurrence data. Future research is
needed to test whether these symptoms would indeed improve the
specificity and sensitivity of the current U.S. diagnostic criteria for
CFS.

The third approach taken in the present study to improve the diagnostic
reliability of CFS was the identification of standardized measures to
assist clinicians in the identification of cases of CFS. Two standardized
measures were evaluated in the present study, the BDI and SF-36.
Components of each of these instruments were found to discriminate CFS,
MDD, and healthy controls. Items measuring self-reproach on the BDI were
found to distinguish CFS from MDD and controls. Applied use of this
information could help prevent cases of CFS being diagnosed as primary
depression and vice versa. With respect to the SF-36, five of the eight
scales were found to distinguish CFS from MDD and controls. Clinical use
of these SF-36 scales could help clinicians evaluate the diagnostic
criteria for CFS, which require "substantial reductions" in functioning,
as well as aid them in distinguishing cases of CFS form other conditions
on the basis of levels of functioning. The above findings support
previous research by Johnson et al. (1995) examining the BDI and research
by Buchwald et al. (1996) examining the SF-36. These results strongly
suggest the inclusion of standardized measures in the diagnostic
assessment process for CFS.

A limitation of the present study involves the small sample size. Small
sample size typically leads to problems with low statistical power,
failure to obtain normal distributions, and failure to obtain a sample
that is representative of the larger population. These problems may
threaten the accuracy of statistical findings and limit the external
validity of results. Even when sample size is not large, mild to moderate
violations of the assumptions of normality are not strong reasons for
choosing a nonparametric test over a parametric test (McCall, 1996). In
the present study, because the violations of normality were only mild to
moderate in nature, parametric tests were used to analyze the data (i.e.,
ANOVA).

A second concern regarding the results of the present study involves the
problem of experiment-wise type I error. Experiment-wise type I error
concerns the overall level of type I error for an experiment. According
to the Bonferroni inequality, the overall alpha level for a study will be
less than or equal to the sum of alpha levels associated with each
individual test (Grimm and Yarnold, 1998). In the present study and alpha
level of 0.01 was used to try and reduce type I error.

The current case definition for CFS was developed by consensus, that is,
based on the opinions and anecdotal experience of a group of "experts"
assembled by the CDC in 1994 (Komaroff et al., 1996). Since this time a
number of studies have found varying degrees of empirical support for the
case definition ( Hartz et al., 1998 and Jason and Taylor, 2002; Jason et
al., 2002a and Jason et al., 2002b). However, many clinicians and
researchers are not completely satisfied with the current case definition
and have questioned the specificity and sensitivity of the current
criteria. Future research should therefore focus on the development of an
empirical case definition for CFS.

The development of an empirical case definition for CFS would entail a
large scale study with many sites across the country. Important
methodological considerations for such a study would include: (1) the use
of a randomly collected community-based sample, (2) use of at least two
physicians to examine each participant and confirm his or her diagnosis
to ensure the diagnostic accuracy and validity of the study sample, and
(3) use of multiple assessment measures (e.g., physical examination,
clinical interview, and standardized self-report measures). Statistical
methods would likely include the use of factor analysis to identify the
most salient predictors of CFS, followed by discriminant function
analyses to examine the ability of the predictors to discriminate and
classify cases of CFS within the context of other medical conditions.


References
Beck, A.T., Steer, R.A. and Garbin, M.G., 1988. Psychometric properties
of the Beck depression inventory: twenty-five years of evaluation.
Clinical Psychology Review 8, pp. 77-100. Abstract | Abstract +
References | PDF (1980 K)

Buchwald, D., Pearlman, T., Umali, J., Schmaling, K. and Katon, W., 1996.
Functional status in patients with chronic fatigue syndrome other
fatiguing illnesses and healthy individuals. The American Journal of
Medicine 101, pp. 364-370. Abstract | PDF (624 K)

Denche, L., Friedberg, F., MacKenzzie, M., Fontanetta, R., 1996. An
typology for chronic fatigue syndrome, unpublished manuscript.

First, M.B., Spitzer, R.L., Gibbon, M., Williams, J.B.W., 1996.
Structured Clinical Interview for DSM-IV Axis I Disorders, Clinician
Version (SCID-CV). American Psychiatric Press, Inc., Washington, DC.

Fukuda, K., Straus, S.E., Hickie, I., Sharpe, M.C., Dobbins, J.G. and
Komaroff, A., 1994. The chronic fatigue syndrome: a comprehensive
approach to its definition and study. Annals of Internal Medicine 121,
pp. 953-959.

Grimm, L.G., Yarnold, P.R., 1998. Reading and Understanding Multivariate
Statistics. American Psychological Association, Washington, DC.

Hartz, A., Kuhn, E.M. and Levine, P.H., 1998. Characteristics of fatigued
persons associated with features of chronic fatigue syndrome. Journal of
Chronic Fatigue Syndrome 4, pp. 71-97.

Holmes, G.P., Kaplan, J.E., Gantz, N.M., Komaroff, A.L., Schonberger,
L.B., Strauss, S.S., Jones, J.F., Dubois, R.E., Cunningham-Rudles, C.,
Pahwa, S., Tosato, G., Zegans, L.S., Purtilo, D.T., Brown, W., Schooley,
R.T. and Brus, I., 1988. Chronic fatigue syndrome: a working case
definition. Annals of Internal Medicine 108, pp. 387-389.

Huber, S., Freidenberg, D., Paulson, G., Shuttleworth, E. and Christy,
J., 1990. The pattern of depressive symptoms varies with progression of
Parkinson’s disease. Journal of Neurology, Neurosurgery, and Psychiatry
53, pp. 275-278.

Huber, S., Rammohan, K., Bornstein, R. and Christy, J., 1993. Depressive
symptoms are not influenced by severity of multiple sclerosis.
Neuropsychiatry, Neuropsychology, and Behavioral Neurology 6 3, pp.
177-180.

Jason, L.A., Richman, J.A., Friedberg, F., Wagner, L., Taylor, R. and
Jordan, K.M., 1997. Politics, science, and the emergence of a new
disease: the case of chronic fatigue syndrome. American Psychologist 52,
pp. 973-983. Abstract | Abstract + References | PDF (2679 K)

Jason, L.A., King, C.P., Taylor, R.R. and Kennedy, C., 2000. Defining
chronic fatigue syndrome: methodological challenges. Journal of Chronic
Fatigue Syndrome 7 3, pp. 17-32.

Jason, L.A. and Taylor, R.R., 2002. Applying cluster analysis to define a
typology of chronic fatigue syndrome in a medically-evaluated, random
community sample. Psychology and Health 17, pp. 323-337.

Jason, L.A., Taylor, R.R., Kennedy, C.L., Jordan, K., Huang, C.,
Torres-Harding, S., Song, S. and Johnson, D., 2002. A factor analysis of
chronic fatigue symptoms in a community-based sample. Social Psychiatry
and Psychiatric Epidemiology 37, pp. 183-189.

Jason, L.A., Torres-Harding, S.R., Carrico, A.W. and Taylor, R.R., 2002.
Symptom occurrence in persons with chronic fatigue syndrome. Biological
Psychology 59, pp. 15-27. Abstract | Full Text + Links | PDF (94 K)

Johnson, S.K., DeLuca, J. and Natelson, B., 1995. Depression in fatiguing
illness: comparing patients with chronic fatigue syndrome, multiple
sclerosis and depression. Journal of Affective Disorders 39, pp. 21-30.

Komaroff, A.L., Fagioli, L.R., Geiger, A.M., Doolittle, T.H., Lee, J.,
Kornish, R.J., Gleit, M.A. and Guerriero, R.T., 1996. An examination of
the working case definition of chronic fatigue syndrome. The American
Journal of Medicine 100, pp. 56-64. Abstract | Full Text + Links | PDF
(831 K)

Matarazzo, J., 1983. The reliability of psychiatric and psychological
diagnosis. In: Hooley, J., Neale, J, Davidson, G. (Eds.), Readings in
Abnormal Psychology. Wiley, New York, pp. 36-65.

McCall, R.B., 1996. Fundamental Statistics for Behavioral Sciences.
Harcourt Brace Jovanovich Publishers, New York, NY.

Nisenbaum, R., Reyes, M., Mawle, A.C. and Reeves, W.C., 1998. Factor
analysis of unexplained severe fatigue and interrelated symptoms: overlap
with criteria for chronic fatigue syndrome. American Journal of
Epidemiology 148, pp. 72-77.

Norusis, M.J., 2000. SPSS 10.0 Guide to Data Analysis. Prentice Hall
Inc., Englewood Cliffs, NJ.

Stewart, A.L., Greenfield, S., Hays, R.D. et al., 1989. Functional status
and well-being of patients with chronic conditions. JAMA 262, pp.
907-913.

Ware, J.J. and Sherbourne, C.D., 1992. The MOS 36-item short-form health
survey (SF-36). I. Conceptual framework and item selection. Medical Care
30, pp. 473-483.

Wasser T.E., in press. Statistical correction of Hollingshead’s four
factor index of social status. Psychological Health.


Copyright © 2004 Elsevier B.V. All rights reserved. ScienceDirect ® is a
registered trademark of Elsevier B.V.