Environmental Sensitivities: Prevalence of Major Symptoms in a Referral Center: The Nova Scotia Environmental Sensitivities Research Center Study

Environmental Health Perspectives, Feb, 2001 by Michel R. Joffres, Tim Williams, Brenda Sabo, Roy A. Fox

Open-ended questions provided space for patients to list the type of exposures that might have been associated with occurrence of their symptoms. This was followed by their family history, demographic characteristics, employment history, socioeconomic status, and information related to the completion of the questionnaire.

The Environmental Health Center Questionnaire was based on the University of Toronto Health Survey questionnaire (22,23). Several modifications were made based on suggestions from the Toronto investigators and feedback from focus groups involving patients, from practitioners, and from pilot testing. Face validity was achieved by designing the symptoms around the six definitions of ES (22,23). To restrict the length of the questionnaire, we placed less emphasis on attempts to identify exposures that patients perceived as provoking symptoms. An open-ended question was added to provide an opportunity to document perceived exposure-symptom relationships. Further modifications included changes in wording from that of the Toronto questionnaire. Complex terminology and politically sensitive language were avoided. Pilot testing indicated that the modifications were relevant and reliable. Ethical approval of study and of consent forms was obtained from the faculty of medicine ethics committee of Dalhousie University.

Content validity for the Environmental Health Center questionnaire was achieved in several ways. Questions relating to ES were derived from consultation with ES practitioners and patients. Because recent definitions state that ES are a multisystem, multisymptom, multifocus health problem (21), the initial questionnaire was subdivided into different organ systems to ensure identification of those systems involved in the illness. Additional symptoms were added to increase content validity after feedback from patients and clinicians. Demographic and social variables were obtained from census questionnaires used by Statistics Canada Standards Division. Questions on general health were modeled on those used in previously validated health surveys (24).

At this point, it is difficult to assess construct validity, predictive validity, and concurrent validity. We evaluated test-retest reliability for 19 individuals who completed the same questionnaire after an interval of approximately 2 weeks. Kappa-values for each section ranged from 0.6 to 0.8, with the lowest value being 0.4 for one question. For the initial Toronto questionnaire, test-retest reliability showed K-values around 0.4-0.6 for most systems and overall good agreement on symptoms (23).

We computed scores as the frequency of occurrence of symptoms since the beginning of the illness (rarely, from time to time, most of the time, all the time, rated as 1-4) multiplied by the severity (low, moderate, high, rated 1-3). Therefore the maximum score for each question is 12, the minimum being 0. The global score is a mean score computed as the sum of all scores divided by the number of symptoms. Statistical tests ([X.sup.2] statistic for proportions and two-sample t-test statistic for means) were used only for select comparisons to limit multiple comparison problems and increase clarity of the tables: Standard errors are provided to simplify comparisons.