This study assessed closeness of interpersonal contacts between individuals in clusters of chronic fatigue syndrome (CFS
as possible evidence of transmissibility or of common exposure to the source of a noninfectious etiologic agent. Two outbreaks of CFS in discrete occupational groups (members of a symphony orchestra and teachers at an elementary school) were studied to assess whether persons with CFS had closer contact with one another than they had with those members of each group who did not have CFS or whether they had more contact with each other than noninfected members had with other noninfected members of each group. Members of the two groups felt that they contracted CFS via workplace exposure to individuals with CFS, Thus, we defined the two occupational groups as the populations at risk.

Since both clusters occurred before the development of a consensus regarding the definition of CFS, a case of CFS was determined by a physician’s diagnosis. Using current diagnostic criteria for CFS, five members of the orchestra met the research case definition of CFS. According to the latest Centers for Disease Control and Prevention (CDC) definition, three other members met Group 2 criteria. In the school cluster, three teachers met the current diagnostic criteria for CFS and four did not. Of the four not meeting the criteria, one was later found to have rheumatic fever and the other three met the criteria of Group 2 of the most recent CDC research case definition.

Members of the groups were asked about their personal contacts with the other members of their group before and during the early phase of each cluster. Contacts were assessed through a hierarchical series of I l questions that ranged from whether they knew the other person to whether they had had intimate contact with that person. Using these data, we developed methods to test the null hypothesis that there were no differences in the frequency or types of contacts between pairs of CFS patients and pairs chosen at random from the cohorts. Since the data consist of n(n – 1)/2 pairs of yes/no responses that are not independent, standard statistical methods used to compare two proportions do not apply here. Thus, for each exposure variable, we compared the observed proportions of case-to-case (C-C), case-to-noncase (C-NC), and noncase-to-noncase (NC-NC) pairs who had interpersonal contact, then computed a Knox-type test statistic and compared it with a corresponding test statistic from a simulated permutation distribution of subject pairs.

The first cluster affected eight individuals from the 67 member orchestra. Data were obtained for all eight members with CFS and 50 members without CFS. The second cluster affected seven teachers in an elementary school with a total of 38 teachers. Data were obtained for all seven teachers with CFS (one was found to have rheumatic fever) and 2 I teachers without CFS. In the orchestra, members with and without CFS were very similar in mean age, race, and marital status. There were more women than men with CFS. In the school, teachers with CFS were slightly older than those without CFS, but the two groups were similar in sex, race, and marital status.

In the orchestra, there were significantly more C-C pairs than C-NC or NC-NC pairs who reported having shared an eating utensil (P = .037) or having shared a bedroom (P = <.0001). There were more C-C pairs for the variables "playing together in a chamber group" (P = .085) and "riding in a car together" (P = .056), but neither result was significant. The data on sharing a bed with a person and on intimate sexual contact were inevalu-able because of the small numbers of such pairs (only one such pair in the C-C group). In the school investigation, there were also more contacts between C-C pairs than NC-NC pairs for the variables "riding in a car together" (P = .08), "riding on a bus together" (P = .10), and "eating a meal together" (P = .10). Although the differences observed in terms of proportions of C-C vs. NC-NC pairs with the specific shared exposure were similar in magnitude to those found for members of the orchestra, none of the differences were statistically significant, probably because of the smaller size of the school group.

These data suggest that in clusters of CFS, affected individuals appear to have had more frequent, and probably more intimate, contact with one another than with unaffected individuals, or than unaffected individuals had with each other. This result suggests interpersonal transmission or a common source of exposure to an etiologic agent.

Reprints or correspondence: Dr. Seymour Gruff'erman, Department of Family Medicine and C'linical Epidemiology, M-200 Scaife Hall, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania f 526 1.

Clinical Infectious Diseases 1994;18(Suppl 1):S54-5 O

Keywords: grufferman epidemiology cluster agent etiology pathology outbreaks transmissibility