Details of the population-based study to estimate the prevalence and incidence of CFS in the adult population of Wichita, Kansas, have been published [11]. In brief, we used a computer-assisted telephone interviewing system to screen ~90 000 persons. Respondents with severe fatigue for at least 1 month (n = 3 528) and randomly selected non-fatigued (n = 3 634) respondents ranging in age from 18 to 69 years completed detailed telephone interviews concerning fatigue, other symptoms, and medical history.

Respondents were queried as to a variety of medical and psychiatric illnesses and stratified on the basis of the absence or presence of conditions that could explain their symptoms and thus exclude a diagnosis of CFS [1]. Reported exclusionary medical conditions included cancer within 5 years of the interview (except basal skin cancer), emphysema, chronic hepatitis, rheumatoid arthritis, acquired immunodeficiency syndrome, systemic lupus erythematosus, Sjögren's syndrome, multiple sclerosis, organ transplantation, pregnancy or major surgery within the past year, and any previous medical condition for which a resolution had not been documented. Stroke, heart attack, heart failure, and a heart condition limiting the ability to walk were exclusionary if they occurred within 2 years of the interview.

Exclusionary psychiatric conditions included those that would prevent a subject from accurately reporting symptoms (e.g., schizophrenia and bipolar disorder) and those with fatigue as a reasonably anticipated symptom (e.g., bulimia or anorexia nervosa, major depressive disorder with melancholia, and alcohol or substance abuse within the 5 years prior to the onset of fatigue).

On the basis of responses to the detailed telephone interview, respondents were classified as having "No Fatigue" if they did not report fatigue of at least 1 month's duration. Respondents reporting fatigue lasting at least 1 month were considered fatigued and stratified into three groups: 1) "Prolonged Fatigue" (ie, those whose fatigue lasted between 1 and 6 months); 2) "Chronic Fatigue" (ie, those who reported fatigue of at least 6 month's duration but with insufficient symptoms or fatigue severity to meet the case definition of CFS); 3) "CFS-like" (ie, respondents whose reported symptoms and fatigue severity met the case definition of CFS).

CFS-like respondents who reported medical or psychiatric conditions that could have explained their fatigue (n = 511) were classified as having "Explained Syndromic Fatigue." The remaining 456 CFS-like respondents, who did not report an exclusionary condition, were invited to participate in a clinical evaluation to confirm a diagnosis of CFS, as recommended in the 1994 International CFS Research Case Definition [1]. Those who agreed to be clinically evaluated received a comprehensive physical examination, psychiatric evaluation, and routine laboratory tests. Because placement in the CFS-like classification group was based on telephone responses, those who declined clinical evaluation retained the classification of CFS-like, as did those who no longer reported sufficient symptoms or fatigue severity during their clinical evaluation. Those who had an exclusionary condition identified upon clinical evaluation were placed in the Explained Syndromic Fatigue category. Last, participants who received a clinical evaluation and reported sufficient symptoms and fatigue severity were classified as having CFS if we were unable to find a medical or psychiatric condition that might explain their fatigue.

We measured functional status by using data from the detailed telephone interview (Table 1). Participants were asked how many hours per week they currently spent on work, household chores, and other activities, such as hobbies, schooling, or volunteer work. We created a variable to represent the total hours of meaningful activity per week, which was the sum of the three individual activity variables for each respondent. Fatigued subjects were also asked how many hours they had spent on these activities prior to the onset of fatigue and if they were currently unemployed due to their fatiguing illness. In addition, all subjects were asked, "During the past 4 weeks, where would you place yourself in terms of energy, wellness, and ability to complete your everyday activities on a scale from 1 to 100, with 1 being the worst you could feel and 100 being the best you could feel?"

We classified subjects into one of six fatigue categories, as outlined above (No Fatigue, Prolonged Fatigue, Chronic Fatigue, Explained Syndromic Fatigue, CFS-like, or CFS). We further stratified the first three groups by whether an exclusionary condition was or was not reported. Three subjects who were evaluated clinically were excluded from this analysis because the results of their psychiatric evaluations were inconclusive, and thus we were unable to classify them.

Statistical analyses were done using SAS version 8.01 (SAS Institute, Cary, NC). Dichotomous variables were compared using Pearson's chi-square test, while the Wilcoxon rank sum and Kruskal-Wallis tests were used to compare continuous variables. The Wilcoxon signed ranks test was used to compare current and prior responses for reported hours spent on activities per week. The Cochran-Armitage test was used to test for trend. All statistical tests were two-tailed, and significance was determined at an α-level of .05.

Figure 1 summarizes the composition of the fatigued and non-fatigued groups. The study population was mostly white (88%) and female (65%) and had a median age of 42 years. Demographic characteristics were similar across fatigue categories, with the exception of female sex and employment (data not shown). The percentage of women was lowest among the non-fatigued group without medical or psychiatric conditions (56%) and progressively increased with each level of fatigue, the CFS group having the highest percent of females (93%, test for trend P < .01). There was also a significant trend for employment. Employment was highest among the non-fatigued group (78%) and decreased with each level of fatigue, with only 54% of those with CFS being employed (test for trend P < .01). Unemployment due to fatigue is discussed below.

Figure 2 shows the median reported number of weekly hours spent on work, chores, and other activities and the total reported hours of meaningful activity for each fatigue category. Fatigue groups without exclusionary conditions, including the No Fatigue group, were similar with respect to the reported number of hours spent on chores. However, they were significantly different with respect to the reported number of hours spent on work and other activities, and the total meaningful activity decreased significantly across the fatigue groups without exclusionary conditions (Kruskal-Wallis P-values < .01).

With regard to other activities, the Prolonged Fatigue (no exclusions), Chronic Fatigue (no exclusions), and CFS-like groups were statistically similar (Kruskal-Wallis P = .30). The No Fatigue group without exclusions reported spending significantly more time on other activities than did the three middle groups without exclusions (Wilcoxon rank sum P < .01), and the CFS group reported significantly less time spent on other activities (Wilcoxon rank sum P = .01). With respect to the reported number of hours spent on work and the reported total number of meaningful activity hours per week, the No Fatigue and Prolonged Fatigue groups without exclusions were statistically similar (Wilcoxon rank sums P = .37 and .19, respectively) and the Chronic Fatigue (no exclusions), CFS-like and CFS groups were statistically similar (Kruskal-Wallis P = .22 and .06, respectively). The more severely fatigued groups without exclusions reported spending significantly less time working and fewer total hours on meaningful activity than the No Fatigue and Prolonged Fatigue groups without exclusions (Wilcoxon rank sums both P < .01).

However, when unemployed subjects were excluded from these analyses, differences in work hours and total hours were no longer significant: all groups without exclusionary conditions, including the No Fatigue group, reported spending a median of 40 to 45 hours working (P = .49) and a median of 65–70 hours on total meaningful activity (P = .70) (data not shown). Therefore, it appears that the differences in reported hours of work and total hours between fatigued and non-fatigued subjects were most likely due to differences in the number of unemployed subjects in each group.

In comparisons of people with and without exclusionary conditions for each fatigue category, several differences were demonstrated. Among all groups, persons with exclusionary conditions reported spending significantly fewer hours per week working than those without exclusionary conditions in the same category (P < .01). The exception was the comparison between those with CFS and those with Explained Syndromic Fatigue, for which the difference was not significant. After unemployed subjects were dropped from the analyses, differences were no longer significant except for the No Fatigue (P = .03) and Prolonged Fatigue (P = .04) groups.

Subjects with Explained Syndromic Fatigue also reported spending fewer hours per week on chores than did those with CFS and those with CFS-like illness (P = .01). People with Chronic Fatigue accompanied by an exclusionary condition and those with Explained Syndromic Fatigue reported spending fewer hours on other activities than did people with Chronic Fatigue without such conditions (P = .02) and those with CFS-like illness (P = .02), respectively.

With regard to the total reported hours of meaningful activity per week, persons with exclusionary conditions in all fatigue categories reported significantly fewer hours (P < .01) except for those in the Prolonged Fatigue and CFS groups. However, when unemployed subjects were excluded, only difference in hours between the CFS-like and the Explained Syndromic Fatigue groups remained significant (P = .01).

Figure 3 shows the change in reported hours per week for work, chores, other activities, and total meaningful activity from before onset of fatigue to time of interview. The change in hours of activity reported is striking. All fatigued groups with and without exclusions reported that they spent significantly less time on all activities after the onset of their fatiguing illness (P ≤ .01), with the exception of the Prolonged Fatigue group without exclusionary conditions, which reported no change in the number of hours worked per week. The magnitude of the decrease in activity was greatest among the Explained Syndromic Fatigue and CFS groups (P < .01); however, the decreases in these two groups did not differ from each other (P = .16–.79).

We asked subjects to express how they felt in terms of energy, wellness and ability to complete everyday activities on a composite scale from 1 to 100 (Figure 4). There was a clear downward trend in energy scores associated with severity of the fatigue category. The No Fatigue groups reported median energy scores between 80 and 85; Prolonged Fatigue, Chronic Fatigue, and CFS-like groups reported median energy scores of 50 (P < .01 compared with No Fatigue groups); CFS and Explained Syndromic Fatigue groups were also significantly lower, with median energy scores of 40 (P < .01 compared with the middle groups).

There was a significant trend of increasing fatigue severity associated with increasing unemployment due to fatigue for subjects with and without exclusionary conditions (P < .01 for both), although differences were more pronounced among fatigue categories with exclusionary conditions (Figure 5). In all fatigue categories, more than twice as many people with accompanying exclusionary conditions reported that they were unemployed because of their fatigue compared with those in the same category without such exclusionary conditions (P < .01). In particular, approximately 40% of subjects with Explained Syndromic Fatigue reported being unemployed due to their fatiguing illness, compared with 15% of subjects with CFS (P < .01).