We report the results of the first study of the chronic fatigue syndrome in the community. Previous population-based surveys have examined the prevalence of fatigue [11-18], but none have directly ascertained the prevalence of the chronic fatigue syndrome itself. Studies in the United States [23], Australia [21], and Scotland [24] have indirectly estimated the prevalence of the chronic fatigue syndrome in the community by identifying persons seeking care for fatigue through sentinel physicians. Another study in the United Kingdom used self-report on a mailed questionnaire to determine the diagnosis of the chronic fatigue syndrome [18]. In these investigations, the frequency of the chronic fatigue syndrome ranged from 3.4 to 249 cases per 100 000 persons. Other approximations of prevalence have used data from an epidemiologic study of psychiatric illness [25] or information from viral serology request forms [26]; these sources did not provide the specific information required to assess whether persons met any case definition for the chronic fatigue syndrome [19-21].

To determine the prevalence of the chronic fatigue syndrome and chronic, unexplained, debilitating fatigue in the community, we surveyed 4000 members of a large health maintenance organization. Another of our objectives was to compare the results of systematic physical and laboratory examinations and assessments of psychological distress and functional status in persons with chronic fatigue (including those with the chronic fatigue syndrome) with the results of these tests in healthy, nonfatigued, age- and sex-matched controls drawn from the same population.

Methods

Top Methods Results Discussion Author & Article Info References

The study setting was Group Health Cooperative of Puget Sound, a staff-model, nonprofit, prepaid health maintenance organization that serves 388 000 persons in Washington State. Group Health Cooperative medical records reflect essentially all medical care received by the organization's members, regardless of service site. Facilities include two hospitals, 23 medical centers, three specialty centers, and a progressive care facility. Group Health Cooperative serves a heterogeneous socioeconomic population, the age and sex of which closely resemble the age and sex of the general population of the Puget Sound region. Group Health Cooperative maintains computerized clinical information systems and supports the Center for Health Studies, a research unit experienced in conducting epidemiologic studies.

Study participants were drawn from the Group Health Cooperative central region, which administratively includes five clinic sites and 84 primary care physicians and geographically contains much of the greater Seattle area. At study commencement, approximately 108 000 persons were enrolled in this region.

Selection of Potential Cases and Controls

A screening survey was mailed to 4000 randomly selected Group Health Cooperative members, regardless of whether they had sought medical care for any reason. The survey was re-mailed to nonrespondents; those who did not respond to the second mailing were telephoned and interviewed directly by research assistants. The screening survey included the following two queries, which were intended to assess the presence of two major criteria for the chronic fatigue syndrome that were established by the original CDC case definition [19] (see Appendix): "Have you felt unusual fatigue or loss of energy, either constantly or repeatedly, for at least the past 6 months?" and "Does this state of fatigue interfere with your work or responsibilities at home such that you have had to reduce your level of activity by at least one-half?" Enrollees answering "no" to one or both queries were excluded from further consideration as cases. Those answering "yes" to both questions reviewed a list of medical and psychiatric conditions that can produce fatigue. If any of these conditions were present, members were considered ineligible for further study.

The Group Health Cooperative records of persons answering "yes" to the above questions and explicitly denying any potential cause of fatigue were abstracted and reviewed. If the medical record showed any previously unreported, unrecognized, or incompletely evaluated medical or psychiatric conditions, or regular use of medications or other substances that could result in chronic fatigue, persons were not evaluated further. Persons were not excluded for a history of a grief reaction (for example, depression after the death of a spouse). When this screening process produced a potential case, approval for study participation was sought from that person's primary care physician. After receipt of approval, eligible cases were invited to participate in the study.

Controls were chosen from among those answering "no" to both of the two major screening questions. The Group Health Cooperative records of controls were reviewed to confirm the absence of medical and psychiatric illnesses associated with fatigue. If they were eligible, potential controls were invited to participate after receipt of approval from their primary care physicians. Controls were age-matched (within 5 years) and sex-matched to the cases; 80% were also matched for clinic location. Cases and controls gave written informed consent.

Index Visit Evaluation

All 74 cases and 74 controls had a medical evaluation consisting of a questionnaire (148 items for cases; 124 items for controls) about past and current medical problems; the questionnaire included items that would elicit the information required to diagnose the chronic fatigue syndrome. Cases and controls also had a standardized physical examination and a battery of blood tests directed toward detecting causes of fatigue. The physical examiner was blinded to whether a person was a case or a control. Routine tests included a complete blood count with manual differential; a test for erythrocyte sedimentation rate; SMA-24 (to assess electrolyte concentrations and levels of blood urea nitrogen, creatinine, glucose, uric acid, phosphate, calcium, magnesium, cholesterol, triglycerides, total protein, globulin, albumin, total and direct bilirubin, alkaline phosphatase, -glutamyl transpeptidase, alanine aminotransferase, aspartate aminotransferase, lactate dehydrogenase, and iron); and tests for thyroid function and antinuclear antibodies. Additional laboratory testing was done on the basis of the medical history, the physical examination, and the laboratory results, according to an explicit, previously developed algorithm.

Functional status and psychological status were assessed by the Medical Outcomes Study Short-Form Health Survey (SF-36) and the Hopkins Symptom Checklist-90 (SCL-90). The SF-36 assesses physical and mental health, role and social functioning, pain, and health perception [27]. The SCL-90 was used to determine current psychological distress by measuring the presence and severity of specific symptoms [16]. Of particular interest were the dimensions of depression, anxiety, and somatization.

Psychiatric diagnoses among persons meeting the criteria for the chronic fatigue syndrome were determined using the National Institute of Mental Health (NIMH) Diagnostic Interview Schedule Version III-A [28], a structured interview administered by a trained research assistant that assigns current and lifetime diagnoses on the basis of DSM-III-R criteria [29]. Additional information was obtained during the administration of the Diagnostic Interview Schedule to determine the onset and resolution of psychiatric disorders relative to the onset of chronic fatigue. The Diagnostic Interview Schedule was not given to all persons because structured psychiatric interviews in the assessment of the chronic fatigue syndrome were not formally recommended by the National Institutes of Health (NIH), the NIMH, or the CDC until 2 to 4 years after study commencement.

Follow-up Evaluations

At 12 and 24 months, cases and controls completed the SCL-90, the SF-36, and a brief fatigue inventory that explicitly inquired about the presence of chronic, debilitating fatigue during the follow-up period. Cases were also seen in person at 1 year and reassessed using a questionnaire (which included questions about symptoms required by the original CDC criteria for the chronic fatigue syndrome) and repeated laboratory and physical examinations identical to those done at the index visit. The physical and laboratory examinations were not repeated for controls. Lastly, the medical records of the 74 cases and the 74 controls were reviewed approximately 1 year after enrollment and at completion of the study protocol to record information collected by Group Health Cooperative providers during regular visits.

Definitions of Chronic Fatigue, Chronic Fatigue Alone, and the Chronic Fatigue Syndrome

Cases who met the two major criteria from the original CDC case definition [19, 30] on the basis of the screening survey and medical record review described above were considered to have chronic fatigue. Cases classified as having the chronic fatigue syndrome fully met the major and minor criteria of the CDC case definition after the initial evaluation (see Appendix). Those who met the major but not the minor criteria were designated as having chronic fatigue alone. Thus, the group of cases with chronic fatigue comprised both those with the chronic fatigue syndrome and those with chronic fatigue alone.

A history of a preexisting psychiatric illness (other than somatoform pain disorder) present for more than 1 year before the onset of the chronic fatigue syndrome or any previous substance abuse precluded the diagnosis of the chronic fatigue syndrome. The diagnosis was not excluded if the onset of a nonpsychotic psychiatric disorder occurred at the same time as or after the onset of the chronic fatigue syndrome.

Statistical Analysis

Because the number of persons with the chronic fatigue syndrome was small, these cases were combined with the cases with chronic fatigue alone for comparisons with controls. Chi-square analyses were used to compare dichotomous variables. Differences in the distributions of normally distributed continuous variables were compared using t-tests; those with skewed distributions were examined using the Wilcoxon rank-sum test. All P values of 0.05 or less are shown in the tables. However, because multiple tests were done, only those with P values of 0.01 or less were considered significant.

Results

Top Methods Results Discussion Author & Article Info References

Identification of Cases with Chronic Fatigue

Cases were recruited between January 1990 and March 1991. Four thousand surveys were mailed, and responses were received by mail or telephone from 3066 persons (77%). Of these 3066 persons, 323 were unable to participate further (for example, because they did not speak English), 419 could not be contacted, and 192 refused to participate. Among 638 persons who reported having fatigue of at least 6 months' duration, 590 (19%) had fatigue that significantly interfered with normal activities (CDC major criterion 1). Self-report (169 persons) or review of the Group Health Cooperative medical record (219 persons) identified exclusionary medical or psychiatric conditions in 388 persons (66%) (CDC major criterion 2). Among the 202 eligible cases (6.6%) who met both of the major criteria, 121 refused to participate, physician approval was not obtained for 3, and 4 could not be contacted. Of these 202 persons, 140 (69%) were women; information on race and ethnicity was not available. Thus, 74 (37%) cases with chronic fatigue were enrolled.

The most common reason that members gave for declining to participate in the study was lack of time or interest. Nonparticipants were significantly younger (median age, 40.2 years compared with 44.4 years; P < 0.001) and more likely to be women (75% compared with 60% were women; P = 0.025) than were participants.

Identification of Controls

Potential controls were chosen from among the 2428 respondents who denied having fatigue, being pregnant, using steroids, or having a health condition that could have produced fatigue on the screening survey. On the basis of this process, the Group Health Cooperative medical records of 400 potential controls were chosen for review; 258 of these potential controls (64%) were found to have no evidence of any medical or psychiatric illnesses. The primary care physicians did not permit 3 potential controls to participate in the study. Using this pool of eligible members, 5 potential controls were randomly chosen for each of the 74 cases with chronic fatigue. Each control was similar to his or her assigned case in age (within 5 years) and sex, and most attended the same clinic as their assigned patients. Potential controls identified by this protocol were sequentially contacted until 1 agreed to participate and was matched to a case. During the recruitment process, 2 controls died, 110 (43%) refused to participate, and 69 were not needed. The sex of the participating controls was similar to that of the eligible nonparticipants (60% of controls and 62% of eligible nonparticipants were women); however, the mean age of the controls was significantly greater than that of the eligible nonparticipants (44.9 years compared with 39.9 years; P = 0.0027).

Identification of Cases with the Chronic Fatigue Syndrome from among Cases with Chronic Fatigue

On the basis of history, physical examination, and laboratory tests, 6 of the 74 cases with chronic fatigue appeared to meet the CDC criteria for the chronic fatigue syndrome. The Diagnostic Interview Schedule was administered to these individuals, who had no evidence on self-report or chart review of preexisting psychiatric illness. Five subjects had a total of 10 current or lifetime diagnoses. One had a history of alcohol and substance abuse and preexisting major depression, and 2 others had preexisting dysthymia; these 3 cases were no longer regarded as having the chronic fatigue syndrome. Thus, 3 cases with the chronic fatigue syndrome were detected among the 74 cases with chronic fatigue: One had no lifetime psychiatric disorder and 2 had had the simultaneous onset of somatoform pain disorder and the chronic fatigue syndrome. The other 71 cases (designated as having chronic fatigue alone) met the CDC major criteria for the chronic fatigue syndrome on the basis of their responses to the screening survey queries about the severity and duration of fatigue (major criterion 1) and the absence of self-reported or documented medical or psychiatric conditions that could produce fatigue (major criterion 2).

By the time of the initial evaluation, several weeks to months after participants had completed the screening survey, 46 of the 71 cases with chronic fatigue alone no longer met the original CDC major criterion 1 (that is, they no longer reported having debilitating fatigue). The minor criteria required by this case definition were not reported by 67 of the 71 cases with chronic fatigue alone. Overall, 5 cases with chronic fatigue alone were found to have a previously undiagnosed organic disease: As directed by our algorithm dictating an appropriate work-up, 1 was found to have anemia secondary to Crohn disease as a result of the initial study evaluation and 1 was found to have anemia as a result of the follow-up study evaluation. Chart review documented medical conditions in 3 subjects (1 had hypothyroidism, 1 had breast cancer, and 1 had sleep apnea). In another 9 persons with chronic fatigue alone, findings detected at the index or follow-up visit or on medical record review suggested but were inadequate to determine conclusively whether a medical condition that could have caused fatigue was present or had subsequently developed. During the study, possible medical conditions associated with fatigue were documented in nine controls, and definite medical conditions associated with fatigue were documented in four controls.

Demographic Characteristics

Table 1 shows the demographic characteristics of the 3 cases with the chronic fatigue syndrome, the 71 cases with chronic fatigue alone, and the 74 age- and sex-matched controls. The mean age of the cases with the chronic fatigue syndrome and chronic fatigue alone was similar to that of controls; sex and racial and ethnic distributions, marital status, educational level, and employment status were also similar. Of note, rates of full-time or part-time employment were high among cases with the chronic fatigue syndrome (100%), cases with chronic fatigue alone (86%), and controls (88%).

A comparison of the demographic profiles of cases with chronic fatigue and Group Health Cooperative members with those of the populations of the Seattle area and the United States showed no striking differences with regard to age, sex, or race. Women were not substantially overrepresented among those with chronic fatigue: They accounted for 59% of chronic fatigue cases, 55% of Group Health Cooperative members, and 51% of the United States population. Surprisingly, whites were underrepresented, accounting for 82.5% of chronic fatigue cases but 91% of Group Health Cooperative members, 90% of the Seattle area community, and 83% of the United States population. Cases with chronic fatigue, Group Health Cooperative members, and Seattle area residents tended to be more educated than the national population as a whole.

Symptoms and Physical Examination Findings

Table 2 shows the clinical features of the cases with the chronic fatigue syndrome, the cases with chronic fatigue alone, and controls. Cases with the chronic fatigue syndrome and chronic fatigue alone were similar in age at onset and in duration of fatigue. The average severity of fatigue during the preceding months was mild (that is, cases were able to attend to their primary responsibilities at home and at work) in almost all cases with the chronic fatigue syndrome and chronic fatigue alone. The onset of illness followed an acute viral infection in 19% of cases with chronic fatigue alone compared with 33% of cases (1 of 3) with the chronic fatigue syndrome. The 74 cases with chronic fatigue (those with the chronic fatigue syndrome plus those with chronic fatigue alone) were significantly more likely than controls to have muscle weakness, muscular pain, neuropsychological complaints, and sleep disturbances (P < 0.001). The higher frequency of symptoms in the cases with the chronic fatigue syndrome is due to the fact that these symptoms define the syndrome.

Abnormalities on the initial physical examination were uncommon across groups and did not contribute to the diagnosis of the chronic fatigue syndrome in any person. Enlarged or tender anterior cervical, posterior cervical, or axillary lymph nodes were found in 33% of cases (1 of 3) with the chronic fatigue syndrome, 14% of cases with chronic fatigue alone, and 1% of controls. These findings were significantly more frequent among the 74 cases with chronic fatigue than among the controls (P < 0.01).

Psychological and Functional Status

Table 3 shows the most salient findings of the functional status and psychological assessment tests. On parametric tests, functional status showed a striking trend for all eight subscales: The highest SF-36 scores were seen in controls, the lowest scores were seen in cases with the chronic fatigue syndrome, and the intermediate scores were seen in cases with chronic fatigue alone. For all subscales, significant differences were noted when controls were compared with the 74 cases who had the chronic fatigue syndrome or chronic fatigue alone (P 0.001). Similarly, the SCL-90 subscale scores manifested a consistent inverse trend: The lowest scores were seen in controls, the midrange scores were seen in cases with chronic fatigue alone, and the highest scores (indicating the most distress) were seen in the cases with the chronic fatigue syndrome. The use of nonparametric tests did not alter our interpretation of the results (P < 0.001 for all comparisons of cases with the chronic fatigue syndrome and chronic fatigue alone with controls) (Table 3). At 12 and 24 months, scores on the SF-36 and SCL-90 subscales among the 74 cases with the chronic fatigue syndrome or chronic fatigue alone were still higher than those among controls (P 0.01 for all comparisons).

Follow-up at 1 Year

Twelve months after the initial study visit, 16 cases (25%) with chronic fatigue alone reported resolution of their fatigue. Of those recovering, 94% had had resolution of the fatigue by the 6-month follow-up visit. Rates of recovery did not differ between men and women. By 1 year after enrollment, none of the cases initially classified as having chronic fatigue alone met the criteria for the chronic fatigue syndrome, but 3 had had onset of fibromyalgia. In contrast, none of the 3 cases with the chronic fatigue syndrome reported resolution of their fatigue, developed fibromyalgia, or had a documented medical disorder that could account for their symptoms in the year since enrollment. Two of the 74 controls developed fatigue but not the chronic fatigue syndrome.

Prevalence Estimates

Estimates of prevalence are affected by assumptions about the frequency of the chronic fatigue syndrome and chronic fatigue alone among nonrespondents and nonparticipants. Of the 4000 persons originally surveyed, 934 did not or could not respond (nonrespondents). Of the respondents who reported having chronic fatigue on the initial survey and were thus eligible for the complete study evaluation, 202 did not participate further (nonparticipants). Only 3 of the 74 participants with chronic fatigue who had had the study evaluation subsequently met the CDC criteria for the chronic fatigue syndrome. If we choose as our denominator only survey respondents and assume that no nonparticipants had chronic fatigue (a conservative assumption), the observed crude point prevalences are as shown in Table 4.

These observed rates have a well-defined lower bound and a reasonable upper bound. If we assume that none of the nonrespondents or nonparticipants had either the chronic fatigue syndrome or chronic fatigue alone, the lowest possible estimates derived from our data would be 75 cases of the chronic fatigue syndrome and 1775 cases of chronic fatigue alone per 100 000 persons. On the other hand, if the rates among nonrespondents and nonparticipants equaled the rates observed among study participants, the estimated prevalence would be 267 cases of the chronic fatigue syndrome and 6321 cases of chronic fatigue alone per 100 000 persons in this community. These seem to be reasonable upper bounds of the prevalence estimates because 1) nonparticipants typically declined to participate in the study because of a lack of interest or time; 2) the evaluation was free; and 3) a plausible reason why the rates among nonrespondents and nonparticipants would be higher than the observed rates is not apparent.

Discussion

Top Methods Results Discussion Author & Article Info References

To our knowledge, this is the first community-based investigation of the prevalence of unexplained debilitating fatigue lasting at least 6 months (referred to as chronic fatigue) and of the chronic fatigue syndrome as defined by the original CDC case definition [19] and subsequent modifications [30]. Previous studies of the chronic fatigue syndrome have, for the most part, been conducted among patients in general medical settings. For example, David and colleagues [9] screened 611 ambulatory patients in the United Kingdom and found only 1 with a chronic fatigue syndrome-like illness who had had at least 6 months of significant fatigue as determined by self-report. In a subsequent investigation in which patients had a comprehensive medical and psychological evaluation, the investigators [8] detected 17 persons with the chronic fatigue syndrome among 686 patients attending a primary care practice. Our recently completed prospective cohort study in a primary care practice identified 3 cases of the chronic fatigue syndrome, as defined by the CDC, among 1000 patients seeking care for any reason [7]. Thus, by extrapolation to standardized rates, the point prevalence of the chronic fatigue syndrome in these investigations ranged from 164 to 2500 cases per 100 000 persons among general medical patients.

Several studies have attempted to determine the population-based prevalence of the chronic fatigue syndrome. In the two most rigorous investigations, cases with the chronic fatigue syndrome were identified only if they sought medical care from sentinel physicians; members of the community were not surveyed directly. Thus, cases not seeking care were undetected. In a largely rural area of Australia, Lloyd and colleagues [21] estimated a crude prevalence of 37 cases per 100 000 persons using a unique definition of the chronic fatigue syndrome to identify cases through local medical practitioners. Similarly, the CDC has conducted a study in four states, in which sentinel physicians identify potential cases who are subsequently referred for a standardized, intensive medical and psychological evaluation. In a preliminary report of this study, Gunn and colleagues [23] state that they found a minimum of 2.0 to 7.3 cases of the chronic fatigue syndrome (as defined by the CDC) per 100 000 persons. These estimates are probably conservative for several reasons. First, the investigators assumed that all cases of the chronic fatigue syndrome in the communities studied had come to the attention of a sentinel physician. Second, the CDC study excluded all patients with preexisting affective, anxiety, and somatization disorders present before the illness onset, whereas a recent conference sponsored by the NIH and the NIMH recommended that only specific psychiatric disorders (primarily psychoses) be considered incompatible with the diagnosis of the chronic fatigue syndrome [30]. We used these latter guidelines; the new CDC case definition of the chronic fatigue syndrome includes similar exclusionary criteria [22].

Ho-Yen surveyed general practitioners [26] and reviewed data contained on viral serology requisitions (diagnoses such as "post-viral fatigue") sent to a regional laboratory [21] to approximate the prevalence of the chronic fatigue syndrome. Using this indirect method, he found that the number of cases ranged from 51 to 131 per 100 000 residents of Britain. In the United Kingdom, Pawlikowska [18] reported finding 6 cases of the chronic fatigue syndrome per 100 000 patients identified from the rosters of six general practices. However, the diagnosis relied on self-report, and persons who reported having the chronic fatigue syndrome did not have systematic medical and psychological evaluations. Lastly, using data from a psychiatric epidemiology study, Price and colleagues [25] estimated the prevalence of the chronic fatigue syndrome to be 7 cases per 100 000 persons [25]. This lower prevalence is probably a result of the strict exclusion of persons with any lifetime psychiatric disorders and the use of data collected with a survey instrument that did not include questions devised to define the chronic fatigue syndrome.

Unlike these previous efforts, our study was designed specifically to ascertain the prevalence of chronic fatigue and the chronic fatigue syndrome in the community, regardless of health care use. Hence, we did not identify cases through medical providers, requests for laboratory tests, or other means that required contact with the medical system. This reduced the likelihood of introducing potential biases secondary to help-seeking behaviors and socioeconomic factors that could influence the presentation of illness. For example, in contrast to studies relying on sentinel physicians, which have found nonwhites to be underrepresented [23], blacks were overrepresented in our study, accounting for 9.5% of cases with chronic fatigue but only 3% of Group Health Cooperative members and 4% of the general population of Seattle. Similarly, we observed only a slight increase in the proportion of women with chronic fatigue: Fifty-five percent of Group Health Cooperative members and 58% of cases with chronic fatigue were women.

It is likely that our sampling strategy also resulted in the representation of a broader spectrum of illness, including milder cases of the chronic fatigue syndrome that nonetheless met the original CDC case definition. That the illness experienced by our cases was less severe is suggested by several observations. First, 62% of cases who reported having had at least 6 months of fatigue and a 50% reduction in activity on the screening survey did not report having one or both of these items at the index visit, suggesting that the severe fatigue was transient or had improved substantially. Second, in our consultative practices of patients with the chronic fatigue syndrome, 35% to 45% are unemployed (unpublished data), and approximately 60% of cases in the study by Lloyd and colleagues [21], which used sentinel physicians, were not working. In contrast, in this study, only 14% of cases with chronic fatigue and none of those with the chronic fatigue syndrome were unemployed. The conservative application of the psychiatric exclusionary criteria by Gunn and colleagues [23] and differences in sampling strategies may explain the greater representation of mild cases and the several-fold higher prevalence of the chronic fatigue syndrome in our study compared with studies by the CDC and by Australian researchers [21, 23].

Finally, although our study was not intended to address prognosis, our findings are consistent with previous longitudinal studies of the natural history of chronic fatigue and the chronic fatigue syndrome. Moderate to complete recovery at 1 year has been found in 22% to 60% of patients with chronic fatigue [1, 5-7]. Similarly, 26% to 57% of patients with chronic fatigue syndrome-like illnesses also improve or recover over varying periods of follow-up [7, 31-34].

One limitation of our study is the possibility of bias due to the differential participation of eligible subjects. Despite a response rate to the screening survey of 77%, only 37% of eligible respondents with chronic fatigue agreed to further evaluation of their fatigue. This is curious, because study eligibility required the presence of substantial disability as a result of the fatigue, and respondents were offered an extensive, free evaluation. The high rate of nonparticipation may be explained by at least three phenomena. First, cases may have overstated the severity of their fatigue on the screening survey: The high rate of employment and the subsequent lack of reports of disabling fatigue on the index visit questionnaire among our enrolled sample suggest that the fatigue may have been viewed as acceptable, normative, or easily explained by a fast-paced life. Second, our offer of a free evaluation may have been less attractive to members of a health maintenance organization (who could obtain such services at little or no cost) than to fatigued persons in fee-for-service settings. Third, some individuals may have realized that psychological distress, substance abuse, or other circumstances explained their fatigue and may have declined an evaluation out of the fear of being stigmatized. A final caveat is that the frequency of chronic fatigue might have been higher among nonparticipants than among participants, making the upper boundary of our prevalence estimates too conservative. Although it is conceivable, we think it is unlikely that the nonparticipants, if truly sicker, would have chosen not to pursue a medical evaluation.

Although chronic fatigue is a widespread symptom in the community, unexplained debilitating fatigue of at least 6 months' duration is considerably less common. The chronic fatigue syndrome occurs even less frequently: A conservative estimate of the point prevalence ranges from 98 to 267 cases per 100 000 persons. Persons with either the chronic fatigue syndrome or chronic fatigue alone have greater psychological distress and lower functional status than healthy persons. Future studies should elucidate the differences between persons seen in clinical settings and those detected in epidemiologic studies.

Appendix: 1988 CDC Criteria for the Chronic Fatigue Syndrome

Major Criteria

1. New onset of debilitating fatigue that does not resolve with bedrest and that impairs average daily activity below 50% of premorbid levels for at least 6 months.

2. Exclusion of conditions that may cause fatigue, including malignancy; autoimmune and chronic inflammatory diseases; infections; neuromuscular or endocrine diseases; specific psychiatric disorders; chronic use, abuse, or side effects of prescription or illicit drugs; and exposure to toxins.

Minor Criteria

Symptom Criteria

To fulfill a symptom criterion, a symptom must have begun at or after the time of onset of increased fatigability and must have persisted or recurred over at least 6 months.

1. Mild fever—oral temperature between 37.5 °C and 38.6 °C—or chills.

2. Sore throat.

3. Painful axillary, anterior, or posterior cervical nodes.

4. Unexplained generalized muscle weakness.

5. Muscle discomfort or myalgia.

6. Prolonged (24 hours or more) generalized fatigue after levels of exercise that would have been easily tolerated in the premorbid state.

7. New or different generalized headaches.

8. Migratory arthralgia without joint swelling or redness.

9. Neuropsychological problems (including photophobia, transient visual scotomata, forgetfulness, excessive irritability, confusion, difficulty thinking, inability to concentrate, and depression.)

10. Sleep disturbance (hypersomnia or insomnia).

11. Onset of the main symptom complex over hours to a few days.

Physical Examination Criteria

Physical criteria must be documented by a physician on at least two occasions at least 1 month apart.

1. Low-grade fever—oral temperature of 37.6 °C to 38.6 °C or rectal temperature of 37.8 °C to 38.8 °C.

2. Nonexudative pharyngitis.

3. Palpable or tender anterior or posterior cervical or axillary lymph nodes.