Multiple chemical sensitivity

IN CLINICAL CASE STUDIES,[1] individuals have reported intolerances to low levels of common environmental exposures, and these intolerances have been associated with multiple types of acute and chronic exposures to organic chemicals. This altered pattern is known by numerous names, but the most common term is Multiple Chemical Sensitivity (MCS) syndrome. In a 1999 consensus definition of MCS syndrome,[2] 6 features are described: (1) the symptoms are reproducible with repeated chemical exposure, (2) the condition is chronic, (3) low levels of chemical exposure (i.e., lower than previously or commonly tolerated) produce the symptoms of the syndrome, (4) symptoms improve or resolve when the incitants are removed, (5) responses occur to multiple chemically unrelated substances, and (6) symptoms involve multiple organ systems.

Collectively, the heterogeneity of symptoms in MCS syndrome, which overlap symptoms of disorders (e.g., chronic fatigue syndrome, fibromyalgia[3]), and the lack of unique identifying physical findings in published investigations have increased skepticism about the existence of the syndrome[4] and have discouraged research on the condition.[1,5,6] The cause of the syndrome remains highly controversial, with some observers espousing biological[1,7-10] and others embracing psychological theories about etiology.[4,5,11-13]

Prevalence estimates, based on a broad definition such as the consensus definition cited above, have not, to our knowledge, been reported. However, telephone surveys of randomly selected adults in various states (e.g., California, New Mexico) indicated that 16% of those polled reported feeling that they were unusually sensitive to everyday chemicals.[14,15] Higher prevalence rates of chemical sensitivity (i.e., approximately 33%) emerged from studies conducted in North Carolina[16] and Arizona[17] when randomly selected adults and college students, respectively, in introductory psychology classes were asked by interviewers or via questionnaires if they were especially sensitive to everyday chemical exposures.

In specific populations, prevalence rates have also been high. Following a gasoline overexposure in one of our studies,[18] 33.3% of workers in an underground tunneling project reported either developing such sensitivities or experiencing the intensification of preexisting sensitivities, with 26.7% of these workers fitting criteria for MCS syndrome, as characterized by the consensus definition, described above. MCS syndrome was also found in 63% of 19 workers referred for examination at a University of California, San Francisco, clinic following pesticide applications for the control of cockroach infestations in a workplace setting.[19] In occupational clinics, rates of MCS syndrome in ill clinic populations have generally been high. MCS syndrome was reported by 13% of 160 solvent-exposed workers who were admitted consecutively for clinical services[20] and by 19% of 111 patients in California diagnosed by physicians as having occupational disease attributable to organophosphates.[21]

Several prevalence estimates of MCS syndrome have emerged in studies of Gulf War veterans, depending on the populations investigated and the methodology used.[22-28] Data from the largest random study, which was based on the responses to questionnaires completed by 11,216 veterans deployed to the Gulf and by 9,761 nondeployed veterans, showed a 3-fold difference in reported chemical sensitivity, favoring the deployed personnel (15%:5%).[22] The rates of MCS have been extremely high among ill veterans. Among a small random sample of Veterans Administration hospital outpatients who complained of illnesses, MCS was reported by 86% of deployed veterans, compared with 30% of the nondeployed veterans.[24] Overall, in every study known to us, more cases of MCS syndrome have been found in deployed than nondeployed individuals; Gulf War veterans have experienced more cases of MCS syndrome than have control populations.

In this article, we sought to achieve 3 objectives: (1) evaluate the psychometric properties of two sets of questionnaire-based criteria for MCS syndrome (both of which conform to the broader definition embodied by the consensus definition); (2) use the criteria to estimate the prevalence of MCS syndrome in multiple center populations; and (3) compare the current diagnostic practices of traditional physician specialists who most frequently treat such patients with those of clinical ecologists.

Method

Participants. We initially approached directors of approximately 40 medical practices informally (i.e., via telephone) about participation in this study. Follow-up information concerning the nature and requirements of study participation was forthcoming. Potential participants were recruited from those known to the authors to be interested in MCS syndrome because of published research, conference participation, or professional affiliation (e.g., membership in the Association for Occupational and Environmental Clinics or American Academy of Environmental Medicine). Directors who declined to participate did not fully explain their reasons for declining. We did not collect demographic data about the practices; therefore, we were unable to evaluate the representativeness of participating clinics.

Eleven medical centers in the United States and Canada participated in this research. There were 8 traditional medical groups (i.e., 5 occupational health clinics, 2 otolaryngology clinics, and 1 allergy practice) and 3 clinical ecological practices that participated in the study. The clinicians in the traditional medical centers tended to be currently or previously associated with academic institutions. Patients being seen for routine physical examinations or surveillance were excluded.

Procedure. We obtained human research ethics approval prior to initiating the study. Participating clinicians critiqued a preliminary version of the 2 questionnaires; the questionnaire on chemical sensitivity and health was adapted for this study from a version used during telephone interviews in an earlier study.[29] The 2 preliminary study instruments were subsequently revised for use in the present study. Participating clinicians also provided information about their diagnostic criteria for MCS syndrome. A procedures manual, developed to ensure uniform procedures by the participating clinicians and support staff, was sent to participating clinicians for their comments, and it was revised to resolve problematic issues. Materials were next field-tested during a 2-wk trial run at the University of Toronto, and they were revised a second time by the second author (ALD). The final version of the questionnaire was used during a 2-mo period for data collection at all 11 participating clinics in the fall of 1992.

Primary measures. The primary measures included a Face Sheet on Demography and Diagnoses. The participant's attending physician or his or her representative completed a 1-page face sheet. Investigators used this instrument to collect information about demographic and occupational status, as well as about primary and secondary diagnoses and assessments of the presence or absence of MCS syndrome. We used a 6-page questionnaire (Questionnaire on Chemical Sensitivity and Health [available electronically at the following website of the Department of Public Health Sciences, University of Alberta: http://www.med. ualberta.ca/phs/faculty/html]), which was completed by all eligible patients, to collect information about demographic characteristics, general health, allergies and sensitivities, responses to chemical odors, health and mental health history, current health problems, and responses to chemicals common in daily life.

Criteria for MCS syndrome. Two definitions of MCS syndrome were formulated, based on 6 domains (representing features that are widely considered characteristic of the condition, described in Table 1). Both definitions were consistent with clinical and epidemiological investigations of the condition by the authors and others[1,5,6,30] and with existing physiological and psychiatric hypotheses about the origins of the syndrome. One definition included relatively severe manifestations, whereas the other was more liberal, as described below.

Six-domain definition of MCS syndrome. The 6-domain definition of the condition required the weekly reporting of at least 1 of 8 neurobehavioral symptoms, at least 1 of 3 symptoms related to odor tolerance (hypersensitivity or hyposensitivity), and at least 1 symptom from 2 other organ systems (collectively constituting domain 1). For domain 2, an individual had to score 80 of 126 or higher on scales that measured several common environmental chemical exposures (defined roughly in terms of parameters related to close and duration of exposure), and some degree of illness had to have been produced. For domain 3, there had to have been at least 2 changed tolerances to environmental stimuli--1 for chemical odors and the other for 1 of the following: food, allergens, medications, alcohol, and sunlight.[31-33] For domain 4, recovery time from illness following chemical exposure had to have exceeded 12 hr. For domains 5 and 6, awareness of chemical odors and difficulty ignoring them, and attribution of illness to chemical exposures at least once during the past month, respectively, were the defining features. We used receiver operator curves[34] to select the cut points for domains 2 and 4 to maximize agreement between these domains and scores based on domains 1, 3, 5, and 6.

Four-domain definition of MCS syndrome. The 4-domain definition required the features described for domains 1, 5, and 6 (above), as well as a score of 20 of 126 or higher on scales that measured the number of common environmental chemical exposures, defined by dose, producing some degree of illness (domain 2, liberal criterion). Therefore, the 4-domain definition included individuals who had relatively mild manifestations of the condition and who required only a changed intolerance for chemical odors, but not necessarily other changed environmental tolerances--a concomitant of the syndrome that is rarely included in the most widely used clinical definitions.[1,5,6,40]

Overall study design

Identification of a sensitive clinic control group. We selected a sensitive clinic control to compare persons with MCS syndrome with persons with more common types of chemical sensitivities. The control subjects had to report that they felt at least "a little bad" in 3 or more situations involving exposures to odorous chemicals, but they did not satisfy either the 4- or 6-domain definitions of MCS syndrome, and they did not have a diagnosis of definite MCS syndrome by their participating physicians.

Analyses: psychometric properties of questionnaire-based definitions. To accomplish objective 1 (i.e., evaluation of the psychometric properties of the 2 questionnaire-based definitions of MCS syndrome), we performed confirmatory factor analysis on the 3 multiple-item domains (i.e., domains 1, 2, and 3). We used principal component analysis to extract the factors and the varimax rotation method for rotation. We required that each factor have at least 2 variables with loadings above 0.4 for domain 1 and above 0.5 for domains 2 and 3. Although 0.5 is the customary cut-off point, we selected 0.4 for domain 1 because it contained items that did not quite reach the 0.5 criterion, but it was biologically and clinically plausible as a constellation of symptoms. We assessed factor reliability with Cronbach's [Alpha]. A preliminary measure of test-retest reliability, based on data obtained from 5 subjects during a 2-wk interval, focused on the multiple-item domains, and we used the kappa statistic for binary responses (domains 1 and 3) and the weighted kappa statistic for ordinal responses (domain 2).[35]

We performed logistic regression analyses to determine the ability of the scores to differentiate the strength and direction of the association within each domain and the assignment of MCS syndrome case status based on our 6-domain criteria. The sensitive clinic control, which was described earlier, was used as the control population. Because domain 5 was completely predictive of being a case, it was not included in this analysis.

We examined the construct validity of the questionnaire-based domains by examining whether differences existed in physical health, chemical sensitivity, psychosocial adjustment to health, and psychiatric status variables among patients who met the 6-domain criteria, the 4-domain criteria, and the sensitive clinic control. Either a chi-square analysis or Fisher's exact test was used in the analysis. We compared continuous variables, and if they did not follow a Gaussian distribution, the Wilcoxon rank-sum test (for 2 populations) or the Kruskal-Wallis test of equality of populations (for more than 2 populations) was performed. We also investigated construct validity by examining the percentage of physician-diagnosed MCS syndrome patients who met the criteria for the various domains. Finally, to assess the face validity of the questionnaire-based domains, we compared the domains to the criteria selected as "important" by the attending physicians.

Analyses: prevalence of MCS syndrome. To estimate the prevalence of MCS syndrome in the 11 medical samples, we used the 2 questionnaire-based definitions of the condition and, in addition, we used the diagnoses of (1) all physicians, (2) traditional physicians, (3) clinical ecologists, and (4)individual traditional specialists. For all analyses, patients were considered MCS syndrome diagnosed only if the physician specified that the patient met his or her criteria "completely." Prevalence rates were calculated as simple proportions, with exact 95% binomial confidence intervals. We used either the chi-square analysis or Fisher's exact test to compare the prevalence rates of individual groups.

Current diagnostic practices. To investigate current diagnostic practices, we compared patients diagnosed with MCS syndrome by traditional specialists with those so diagnosed by clinical ecologists on physical health, psychosocial adjustment to health, chemical sensitivity, and psychiatric variables. We also compared patients diagnosed with MCS syndrome (by physicians) with the sensitive clinic controls on the bases of physical health, chemical sensitivity, psychosocial adjustment to health, and psychiatric variables. We then used the 6- and 4-domain definitions (i.e., chi-square and Fisher's exact tests) to evaluate the sensitivity and specificity of the diagnoses of the traditional physicians and clinical ecologists. Exact binomial confidence intervals of 95% were used for proportions. In all cases for which traditional specialist physicians were collapsed into 1 category, those specialists included occupational physicians (n: 5), otolaryngologists (n = 2), and allergists (n = 1).

Factor and reliability analyses were carried out with the Statistical Package for the Social Sciences (SPSS 7.5, SPSS Inc. [Chicago, Illinois[36]]). In the remaining analyses, we used STATA version 5.0.[37]

Results

Demographic characteristics of participants and nonparticipants. A total of 1,294 face-sheets and/or questionnaires were completed. Participants were excluded from analyses for the following reasons: discrepancies of more than 2 y between their ages on the face sheet and questionnaire (n = 29), age less than 18 y (n: 99); and incomplete face sheet or questionnaire (n = 151). Data from patients who had completed only a questionnaire (n = 45) or completed only a face sheet (n = 106) were included in some analyses.

The demographic characteristics of the clinic participants were compared with characteristics of the nonparticipants (all aged [is greater than or equal to] 18 y) on the basis of information on completed face sheets, which were available on both groups. Age and gender compositions were similar for the two groups, but participants were more likely than nonparticipants to have achieved higher socioeconomic levels (i.e., judged by education level and occupation), and participants were more likely than nonparticipants to have been seen by a clinical ecologist during their most recent visit. Nonparticipants were more likely than participants to attribute their visit to chemical illness or injury. Diagnosed MCS syndrome (designated by the physician as either "completely" or "partially" meeting her or his personal criteria) was not significantly different between participants and nonparticipants.

Questionnaire-based definitions of MCS syndrome. The 6 domains of questionnaire-based criteria for MCS syndrome that constituted the 6- and 4-domain definitions are shown in Table 1.

With respect to the 6-domain definition, we constructed receiver operator curves and used 4 diagnostic criteria (1, 3, 5, and 6) as the response variables (as was described earlier). For domain 2, the constructed receiver-operator curve indicated that a score of 80 (of a maximum of 126) optimized the percentage of true-positive diagnoses for a fixed percentage of false-positive diagnoses. Given the categorical responses in domain 4, the receiver operator curve indicated that a 12hr requirement was optimal--providing a relatively low percentage of false-positive diagnoses at the expense of only slightly lowering the percentage of true-positive diagnoses. The 12-hr requirement for domain 4 was consistent with our goal of using the 6-domain definition to identify the most severely affected subjects.

With respect to the 4-domain definition of MCS syndrome, the criterion for domain 2 was arbitrarily set at 20 or higher. Domains 3 and 4 were dropped because they are least often considered essential for defining the condition. Patients who met only the 4-domain definition (excluding the subset of patients that met both the 4-domain definition and 6-domain definition) were compared with those who met only the 6-domain definition of MCS syndrome on physical health, chemical sensitivity, psychosocial adjustment to health, and psychiatric status. Given that 30 comparisons were run, the p level for significance was set at .05/30 or .0016 (not shown). The two groups differed at the .0016 level or beyond on 0/8 physical health, 2/4 chemical sensitivity (severity of sickness induced by chemical exposures, recovery time after exposures), 1/7 psychosocial adjustment to health (work limited because of health), and 0/10 psychiatric variables. Patients diagnosed by the 6-domain definition were somewhat more likely to report more illness, sensitivity, and disability than those diagnosed by the 4-domain definition.

We used univariate analyses to compare patients (n = 30) who met either set of criteria for MCS syndrome with the sensitive clinic control group. A conservative adjustment for the multiple comparisons required a p value of .05/30 = .0016 to reach significance. Patients who met either definition for MCS syndrome differed significantly from the sensitive clinic control group on 8/8 physical health, 5/5 chemical sensitivity, and 7/7 psychosocial adjustments to health variables. Patients who met the 6-domain criteria differed from the sensitive clinic control group on 2 of 10 psychiatric variables: (1) percentage reporting burning sensation in sexual organs at least weekly (26% vs. 11%, p [is less than] .001); and (2) good health during their first 30 y (associated negatively with psychosomatic illness) (72% vs. 90%, p [is less than] .001). The patients who met the 4-domain criteria differed from the sensitive clinic control group on 6/10 psychiatric status characteristics: total number of psychiatric diagnoses (1.09 vs. 0.63, p [is less than] .0001); percentage who experienced burning sensation in sexual organs at least weekly (25% vs. 11%, p [is less than] .001) and fainting (5% vs. 2%, p [is less than] .001); percentage reporting ever having been treated for distress and adjustment difficulties (40% vs. 24%, p [is less than] .001); and percentage reporting ever having been treated for general depression (35% vs. 19%, p [is less than] .001). Moreover, patients who met the 4-domain MCS syndrome definition were less likely than clinic controls to rate their general health during their first 30 yr as "good" or "better" (73% vs. 90%, p [is less than] .001).

Factor structure, reliability, and validity: questionnaire-based MCS syndrome domains. We performed exploratory factor analysis on the variables within domains 1, 2, and 3 (i.e., the only multi-item domains) to determine if one or several underlying factors were being measured. Regardless of whether a face sheet had been completed, we used data for all individuals who completed a questionnaire on chemical sensitivity to perform factor analysis. First, confirmatory factor analysis was performed on 49 symptoms, which clinic patients were asked to report if they occurred at least once a week during the past month (domain 1) (Table 2). Factor 1, accounting for 22% of the variance, included the following: (1) neurobehavioral symptoms, such as confusion or difficulty with memory or concentration; (2) chemical odor intolerance; (3) systemic symptoms, such as chronic, overwhelming fatigue; (4) gastrointestinal symptoms; and (5) visual changes (e.g., blurring). Factor 1, conceptualized as the most characteristic features of MCS syndrome, had an internal consistency of 0.86, as measured by Cronbach's [Alpha]. The remaining 3 factors described symptom complexes dominated by the musculoskeletal system (factor 2, accounting for 4% of the variance); the larynx/oral cavity and skin, indicative of allergic disorders (factor 3, accounting for 4% of the variance); and the upper respiratory system, especially the sinuses, nose, and eyes (factor 4, accounting for 3% of the variance). Factors 2, 3, and 4 may represent clusters of symptoms commonly seen in occupational health, allergy, and otolaryngology clinics. The total cumulative percentage of variance accounted for was 33%. Cronbach's [Alpha] ranged from 0.65 to 0.86. Test-retest reliability for the 49 items of domain 1 was .83 (95% CI = 0.75, 0.92 [kappa statistic]).

We also used confirmatory factor analysis to examine a priori hypotheses regarding dose of exposure and severity of effects (domain 2), as well as changes in tolerance to environmental stimuli (domain 3). Domain 2 yielded 8 factors: factor 1 represented response severity to brief, low-level environmental exposures. Factor 2 represented response severity to 2 organic chemical exposures inside: paint and pesticides. Factors 3-8 indicated that people reported that they responded consistently to 2 different durations of specific chemical exposures (e.g., roadside exposures, perfume/cologne, tobacco smoke, gas stoves, shopping, wall-to-wall carpet). The finding of 8 separate factors for domain 2 suggested that people responded to most situations independently, rather than responding with a fixed bias (such as "yeasaying" or "naysaying"). Cronbach's [Alpha] ranged from 0.89 to 0.94, with a test-retest reliability of 0.75 (95% CI = 0.61, 0.88 [weighted kappa]). Factor 1 accounted for 51% of the variance, factor 2 accounted for 8%, factor 3 for 6%, factors 4-7 for 4% each, and factor 8 for 3%, for a cumulative total of 84% of the variance accounted for by these 8 factors.

With respect to domain 3, 4 factors were extracted, representing changes in intolerance for food, classical immunoglobulin E-mediated allergens (e.g., to mold, dust), alcohol and medications, and sunlight. Cronbach's [Alpha] suggested that both of the first 2 factors were highly internally consistent domains ([Alpha] = 0.86 and 0.85), but that factor 3 (i.e., alcohol and medications) had relatively low internal consistency ([Alpha] = 0.59), thus precluding the assumption of one underlying construct. Factor 4 comprised a single item: sunlight. Factor 1 accounted for 40% of the variance, factor 2 for 13%, factor 3 for 7%, and factor 4 for 6%; therefore, these 4 factors accounted for a cumulative total of 66% of the variance. Test-retest reliability for domain 3 yielded a kappa statistic of 0.83 (95% CI = 0.66, 0.99).

To determine which individual variables within each of 5 domains were most predictive of MCS syndrome, we undertook separate logistic regression analyses for each domain. We repeated each regression: in the first instance, we used the 6-domain criteria, and in the second instance, the 4-domain criteria as cases and the sensitive clinic population controls as noncases. Domain 5 was not included in these analyses because, by definition, hypersusceptibility to chemical odors was completely predictive of case status. Hence, a total of 10 logistic regression models were analyzed. Results of the most parsimonious 5 individual models in which the 6-domain criteria were used are shown in Table 3. Twenty-one items identified the variables most closely associated with being defined as a case. These items included 7 symptom, 7 exposure, and 4 changed tolerance variables; 3 additional predictive questions related to recovery time after chemical exposure, hypersusceptibility to chemical odors, and frequency of impairment after chemical exposure. Affirmative responses to 4 questions appeared to lower the risk of being defined as a case.

We determined that both the duration of recovery time and the frequency of impairment resulting from chemical odors were significant independent and incremental predictors of case status, using both the 6-and 4-domain criteria, despite exclusion of recovery time as a defining feature of the 4-domain criterion.

We assessed the construct validity of the domains of the questionnaire by looking at the factor analysis and by analyzing the percentages of physician-diagnosed MCS syndrome patients who met the criteria for each domain (Table 4). The domain criteria appeared to have been used implicitly or explicitly by physicians in the case of all domains, except 4 (i.e., recovery time). Domain 5 (i.e., hypersusceptibility to odorous chemical exposures) also showed somewhat weaker acceptance, as reflected in physicians' diagnoses. Moreover, physicians' diagnoses were more likely to meet our criteria for the 4-domain definition than for the 6-domain definition for reported degree of sensitivity and illness from exposures (domain 2).

The face validities of the domains of the questionnaire were supported by their agreement with physicians' reported personal criteria for diagnosing MCS syndrome (not shown). Criteria that were declared "important" by more than 50% of clinician participants included the following: "symptoms are elicited by chemicals of diverse structural classes and toxicologic modes of action" (by 100% of clinician participants), "exposures that elicit symptoms must be very low" (91%), "no single widely available test of organ function can explain symptoms" (82%), "symptoms recur and abate in response to predictable stimuli" (73%), "symptoms involve more than one organ system" (64%), "affected individuals show heightened awareness of chemical odors" (55%), and "affected individuals demonstrate avoidance of 2 or more incitants" (55%).

Prevalence of MCS syndrome. Estimates of the prevalence of MCS syndrome depended on the criteria used (Table 5). As expected, the less-stringent 4-domain definition yielded higher prevalence estimates than the 6-domain definition. Overall, a total of 70 individuals (6.6%) met the criteria for the 6-domain definition and 247 individuals (23.3%) met the criteria for the 4-domain definition. Prevalence was highest among patients seen by clinical ecologists (11%, 6-domain definition; 37%, 4-domain definition) and lowest among patients seen by otolaryngologists (1%, 6-domain definition; 5%, 4-domain definition). Prevalence estimates among patients seen by occupational physicians (8%, 6-domain definition; 27%, 4-domain definition) exceeded prevalence among patients seen by allergists (2% and 13%, respectively). The prevalence rates, based on physicians' diagnostic criteria for MCS syndrome, were lower than the rates based on the questionnaire data. This difference is described in more detail in the section that follows.

Diagnosis of MCS syndrome. Patients diagnosed by traditional specialists were compared with those diagnosed by clinical ecologists with respect to physical health, psychosocial adjustment to health, chemical sensitivity, and psychiatric status variables. We used a p value of .0016 to compensate for the 30 comparisons (.05/30), and only one statistically significant difference was found between the patients diagnosed by the two specialist groups ("seen by clinical ecologist prior to this visit"). Moreover, no statistically significant differences were found between the MCS syndrome patients diagnosed by the two groups of clinicians with respect to any demographic variable. Patients diagnosed as having MCS syndrome were significantly different from the sensitive clinic controls on 8/8 physical health variables, 5/5 chemical sensitivity variables, 7/7 psychosocial adjustment to health variables, and 3/10 psychiatric variables examined (even when we used a p value of .0016 to compensate for the 30 comparisons [.05/30]).

We compared patients diagnosed with MCS syndrome by specialist physicians with the 4- and 6-domain questionnaire-based definitions of the condition to assess the sensitivity and specificity of clinician diagnoses (Table 6). The MCS syndrome diagnoses of traditional physician specialists achieved greater specificity than did those of clinical ecologists when evaluated by both the 6-domain definition (97% vs. 82%, respectively; p [is less than] .001) and 4-domain definition (98% vs. 90%, respectively; p [is less than] .001). Clinical ecologists achieved greater sensitivity than traditional physician specialists, as judged only by the 4-domain criterion of MCS syndrome (41% vs. 16%, respectively; p [is less than] .001). Traditional specialists failed to identify 60% and 84% (assessed by 6- and 4-domain questionnaire-based definitions, respectively) of their MCS syndrome patients, whereas clinical ecologists failed to identify 50% and 59%, respectively, of their MCS syndrome patients.

Discussion

Evaluating 6- and 4-domain questionnaire-based criteria for identifying MCS syndrome. The 6-page self-report questionnaire, which was used for the identification of MCS syndrome, required between 15 and 20 min to complete. Preliminary data suggest that the domains of MCS syndrome derived from the study instrument possess excellent preliminary test-retest reliability ([is greater than] 0.80) and good internal consistency (0.59-0.94). Construct validity was supported by the finding that the factors extracted appeared to match domains or features of domains identified a priori as criteria for MCS syndrome by clinicians in the published literature and to possess good internal consistency. For example, symptoms especially characteristic of MCS syndrome (e.g., neurobehavioral symptoms, odor intolerance, systemic symptoms [e.g., chronic overwhelming fatigue], changes in eyesight, gastrointestinal symptoms) were identified within 1 factor with excellent internal consistency (Cronbach's [Alpha] = 0.84). The association of more disability and more symptoms with a more rigorous definition of MCS syndrome (one demanding a greater degree of perceived illness in proximity to environmental chemicals, a recovery time of at least 12 hr, and the perception of changed tolerances for both odor and another environmental stimulus) was likewise supportive of the questionnaire's construct validity. The construct validity of the questionnaire was further supported by the implicit or explicit use of the criteria defining the domains by physicians diagnosing MCS syndrome. The face validity of the questionnaire was supported by the agreement between physician-reported diagnostic criteria and the domains comprising the questionnaire.

The 90-question questionnaire used to define MCS syndrome (6-domain criteria), and especially the 21-question subset identified by logistic regression, have significant advantages over questionnaires developed previously for the screening of MCS syndrome by virtue of combining three characteristics: (1) brevity, (2) a quantitative approach, and (3) coverage of major clinical criteria. Several existing questionnaires focus only on one domain, such as the number of odorous chemical exposures that elicit symptoms or the frequency with which specific chemical odors cause illness,[16,38,39] and they do not attempt to include other facets of the illness. Other questionnaires propose idiosyncratic criteria.[40] Others are long and are intended for hypothesis testing and, therefore, are not appropriate for screening.[41] A recent formulation[14] proposes 15 questions, with a branching structure for the screening of "MCS sufferers in a medical setting," and it is defined in terms of 3 criteria: (1) physician-diagnosis of MCS, (2) self-assessed unusual sensitivity to chemicals, and (3) a health problem that restricts the patient's daily activities. Our research identified problems with the requirement that a physician provide a diagnosis for defining the syndrome.

Prevalence estimates of MCS syndrome. Most studies of population-based prevalence have addressed chemical sensitivity or multiple chemical sensitivity--defined primarily in terms of self-reported adverse responses to varied chemicals, rather than MCS syndrome, defined more stringently--so that the estimates have not been strictly comparable. Researchers who have conducted investigations that examine frequent sufferers provide information that can be compared more justifiably with our data. Two previous studies of frequent sufferers in the general population found MCS syndrome prevalence rates of approximately 12%.[14,16] Our overall MCS syndrome prevalence rate (i.e., 23%) exceeded this rate, most likely the result of the at-risk nature of the group we studied.

Studies of prevalence of multiple chemical sensitivities in particular clinics vary widely according to selection criteria. Most studies have been based on small cohorts with specific exposures and do not present overall perspectives about prevalence in clinic populations as a whole.[16,18,41] We know of only one pertinent study. On the basis of a clinic population size of 705 clinic patients seen for surveillance, but using a single factor definition of MCS syndrome, investigators found that 15% of occupational medicine patients were affected and 20% of general medicine clinic patients were affected.[38] Our overall prevalence rate of 23% was likely elevated by two factors: (1) restriction of the population to nonsurveillance patients (our population was seeking help for medical problems) and (2) the inclusion of clinical ecological practices. Among traditional physicians, we found the lowest rate of 5% (95% CI = 3, 9) in the otolaryngology clinics, an intermediate rate of 13% (95% CI = 9, 1 7) in the allergy practice, and the highest rate of 27% (95% CI = 20, 35) among the occupational medicine clinics. The rates found in our investigation may also have been inflated by the fact that participating study physicians had a more than average interest in MCS syndrome and, therefore, may have attracted a relatively large proportion of such patients.

Diagnostic practices for identifying MCS syndrome. Whether specialists were traditional physicians or ecologists, the specificity for diagnosing MCS syndrome was very high, whereas the sensitivity for diagnosing the condition was poor. Evaluation against the 4-domain, but not the 6-domain, questionnaire-based definition resulted in significantly differing sensitivity of diagnosing the syndrome among traditional specialist physicians and clinical ecologists. The 25% discrepancy in sensitivity between the two groups, which favored the ecologists, may derive from the fact that clinical ecologists are more likely than traditional physicians to ask questions about chemical sensitivities because such sensitivities are central to their interests. It is noteworthy, however, that neither group performed well with respect to identifying the presence of the condition.

Assuming our questionnaire-based identification of MCS syndrome was accurate in identifying the syndrome, clinicians' judgments about the absence of the syndrome were especially problematic. The overall false-negative rate of physicians with an interest in this syndrome was high (53%) with the 6-domain questionnaire-based criteria and was 68% with the 4-domain questionnaire-based criteria.

Overall, these observations underscore the need for both unified criteria for identifying the syndrome and for an instrument that can achieve high sensitivity while retaining high levels of specificity. We emphasize that the more rigorous 6-domain criteria, which identified subjects with more illness and disabilities, also improved sensitivity in identifying MCS syndrome.

Traditional specialist physicians were more specific than clinical ecologists, regardless of whether the 4- or 6-domain questionnaire-based criteria were used. These findings suggest that clinical ecologists are more likely than traditional specialist physicians to falsely diagnose MCS syndrome. The relatively high proportion of false-positive diagnoses by clinical ecologists may be responsible, in part, for the ongoing criticism and distrust of clinical ecology by traditional medical practitioners.

MCS syndrome patients visiting traditional medical specialists and clinical ecologists, however, showed more similarities than differences in health (physical or psychiatric), adjustment to health, and chemical sensitivity. These findings imply that similar populations of patients visit both groups of physicians--a finding that accords with the trend toward increasing acceptance by the general public of nontraditional providers of health care. In fact, there was a trend toward more-educated persons visiting clinical ecologists, but this did not reach statistical significance.

Chemical sensitivity versus MCS syndrome. It was clear from our findings that more and less rigorous questionnaire-based criteria for MCS syndrome resulted in very similar manifestations of general health, chemical hypersensitivity, psychosocial adjustment to illness, and psychiatric qualities. Patients meeting the 6- and 4-domain definitions of MCS syndrome differed primarily in terms of degree of illness and disability, with those diagnosed with the 6-domain definition manifesting more pathology and disability.

Both 6- and 4-domain variants of MCS syndrome were significantly different from manifestations of chemical sensitivity in a sensitive clinic control. In contrast to the sensitive controls, the MCS syndrome patients reported more illness from more common environmental chemical exposures, more changed tolerances, longer recovery times after illness-inducing exposures, more odor related symptoms (including odor intolerance), more illness related to chemical exposures, and more symptoms, especially of the following types: neurobehavioral (e.g., confusion); systemic (e.g., overwhelming fatigue); gastrointestinal (e.g., stomach pain); odor intolerance; changes in vision (e.g., blurring); and upper respiratory symptoms (e.g., stuffy nose).

Psychiatric aspects of MCS syndrome. Our data allow comment on psychiatric factors in MCS syndrome, an issue that we did not set out to investigate. Compared with the clinic controls, the patients who met the questionnaire-based definition of MCS syndrome reported that they were treated more often for psychopathology of several types: depression, adjustment-related disorders, possible psychosomatic symptoms (e.g., burning sensation in sexual organs); and they less frequently described their health during the first 30 y as "good" or "excellent." However, the majority (60-79%) of people who were defined as meeting criteria for MCS syndrome did not report treatment for depression, anxiety, or distress/adjustment issues. Moreover, the majority (73%) of individuals who were defined as meeting criteria for MCS syndrome reported "good" or "excellent" health during the first 30 y of life. Hence, our self-reported data are not supportive of the idea that MCS syndrome is primarily a psychiatric or psychosomatic condition.

Two additional findings in our study are of interest. People who, by definition, met criteria for MCS syndrome were 5-6 times more likely to seek treatment for psychological issues secondary to chemical sensitivity than were the sensitive clinic controls. Moreover, when fit into a multivariable model, only 2 psychiatric variables (i.e., involvement in a support group as a result of chemical sensitivities [6-domain definition] and psychological counseling for chemical sensitivities [4-domain definition]) were independent predictors of satisfying our criteria for MCS syndrome. These data are congruent with the hypothesis that much of the psychopathology seen in MCS syndrome is secondary to feeling ill.

Conclusions

To our knowledge, our study sample has constituted the largest sample of patients that have met strict, widely agreed upon criteria for MCS syndrome studied to date. We believe that the instrument developed for this study, with questions that operationalized 6 domains of MCS syndrome, was useful for research because it was relatively brief and provided quantitative information about commonly observed characteristics of the syndrome. We have provided preliminary evidence that this questionnaire has desirable psychometric properties, including internal consistency, test-retest reliability, and construct and face validity.

Our findings did not suggest that either the 6- or 4-domain definition of MCS syndrome was superior to the other for all research purposes. Rather, we believe that both were satisfactory and that the desirability of one or the other definition would depend on the intended research purpose. In research for which definition of the syndrome should be conservative and have a high specificity, at the expense of lower sensitivity (e.g., characterizing physiological concomitants), the 6-domain definition would be preferable. In research for which definition of the syndrome should have a high sensitivity, at the expense of lower specificity (e.g., that which attempts to characterize the condition at an early phase), use of the 4-domain definition would be preferable.

We have provided preliminary evidence that suggests that MCS syndrome tends to be both commonly misdiagnosed and underdiagnosed by physicians--even by physicians who have a special interest in the condition. The data also suggest that the MCS patients seen by both traditional medical specialists and clinical ecologists are similar in health status. Both groups of physicians tend to frequently overlook the condition. Moreover, our observations indicate MCS syndrome is prevalent. Collectively, these findings underscore the need for objective criteria and an objective screening instrument to guide diagnoses. Such criteria and such an instrument were described.

Future investigations will extend the information that has emerged from this study. Investigators will find it fruitful to examine the prevalence and nature of MCS-like symptoms in a large sample undergoing surveillance, without health complaints, and to isolate demographic trends (gender and socioeconomic differences) in large samples with MCS syndrome. It will also be valuable to researchers to address differences between those with exposure-related versus chronic or life-long sensitivities to chemicals (the later being prevalent among those with allergy, sinusitis, or rhinitis).

References

[1.] Ashford NA, Miller CS. Chemical Exposures. Low Levels and High Stakes, 2nd ed. New York: Van Nostrand Rheinhold, 1998.

[2.] Multiple chemical sensitivity: a 1999 consensus. Arch Environ Health 1999; 54:147-49.

[3.] Buchwald D, Garrity D. Comparison of patients with chronic fatigue syndrome, fibromyalgia, and multiple chemical sensitivities. Arch Intern Med 1994; 154:2049-53.

[4.] Terr Al. Environmental illness: a clinical review of 50 cases. Arch Intern Med 1986; 146:145-49.

[5.] Cullen MR. The worker with multiple chemical sensitivities. An overview. Occup Med State Art Rev 1987; 2:655-68.

[6.] Report of the Ad Hoc Committee on Environmental Hypersensitivity Disorders. Toronto, Canada: Office of the Minister of Health, 1985; pp 17-18.

[7.] Meggs WJ. Neurogenic inflammation and sensitivity to environmental chemicals. Environ Health Perspect 1993; 101:234-38.

[8.] Meggs WJ, Elsheik T, Metzger WJ, et al. Nasal pathology and ultrastructure in patients with chronic airway inflammation (RADS and RUDS) following an irritant exposure. Clin Toxicol 1996: 34:383-96.

[9.] Bell I, Miller C, Schwartz G. An olfactory limbic model of Multiple Chemical Sensitivity syndrome: possible relationships to kindling and affective spectrum disorders. Biol Psych 1992: 32:218-42.

[10.] Callender TJ. Metabolic brain imaging abnormalities in chemically exposed individuals who developed multiple chemical sensitivity syndrome or chronic fatigue syndrome. Session on Multiple Chemical Sensitivity and the Environment. II. Diagnosis and Therapy. Paper presented at the American Public Health Association, Atlanta GA: November 10-14, 1991.

[11.] Brodsky CM. "Allergic to everything": a medical subculture. Psychosomatics 1983; 24:731-34.

[12.] Staudenmeyer HS, Selner JC. Neuropsychophysiology during relaxation in generalized universal "allergic" reactivity to the environment: a comparison study. J Psychosomat Res 1990; 34: 259-70.

[13.] Bolla-Wilson K, Wilson RJ, Bleecker ML. Conditioning of physical symptoms after neurotoxic exposure. J Occup Med 1988; 30:684-86.

[14.] Kreutzer R, Neutra RR, Lashuay N. Prevalence of people reporting sensitivities to chemicals in a population-based survey. Am J Epidemiol 1999; 150:1-12.

[15.] Voorhees RE. Memorandum from New Mexico Deputy State Epidemiologist to Joe Thompson, Special Counsel, Office of the Governor, 13 March 1998.

[16.] Meggs WJ, Dunn KA, Bloch RM, et al. Prevalence and nature of allergy and chemical sensitivity in a general population. Arch Environ Health 1996; 51:275-82.

[17.] Bell IR, Schwartz GE, Peterson JM, et al. Self-reported illness from chemical odors in young adults without clinical syndromes or occupational exposures. Arch Environ Health 1993; 48:6-13.

[18.] Davidoff AL, Keyl PM, Meggs WJ. Development of multiple chemical sensitivity in laborers after acute gasoline fume exposure in an underground tunneling operation. Arch Environ Health 1998; 53:183-89.

[19.] Cone JE, Sult TA. Acquired intolerance to solvents following pesticide/solvent exposure in a building: a new group of workers at risk for multiple chemical sensitivities? Toxicol Ind Health 1992; 8:29-40.

[20.] Gyntelberg F, Vesterhauge S, Fog P, et al. Acquired intolerance to organic solvents and results of vestibular testing. Am J Ind Med 1986; 9:363-70.

[21.] Tabershaw IR, Cooper WC. Sequelae of acute organic phosphate poisoning. J Occup Med 1968; 8:5-20.

[22.] Kang HK, Mahan CM, Lee KY, et al. Prevalence of chronic fatigue syndrome among U.S. Gulf War veterans. Boston, MA: Fourth International AACFS Conference on CFIDS, 10 October 1998 (abstract and presentation).

[23.] Fiedler N, Kipen H, Natelson B. Civilian and veteran studies of multiple chemical sensitivity. Boston MA: 21 6th Annual meeting of the American Chemical Society, Symposium on Multiple Chemical Sensitivity: Problems for Scientists and Society; 26 August 1998.

[24.] Bell IR, Warg-Damiani L, Baldwin CM, et al. Self-reported chemical sensitivity and wartime chemical exposures in Gulf War veterans with and without decreased global health ratings. Milit Med 1998; 163:725-32.

[25.] Black DW, Doebbing BN, Voelker MD, et al. Multiple Chemical Sensitivity syndrome: symptom prevalence and risk factors in a military population. Atlanta, GA: The Health Impact of Chemical Exposures during the Gulf War--A Research Planning Conference. 28 February 1999 (presentation; manuscript submitted for publication).

[26.] Canadian Department of National Defense (CDND). Health Study of Canadian Forces Personnel Involved in the 1991 Conflict in the Persian Gulf. Ottawa, Canada: Goss Gilroy, 20 April 1998. [Online at: www.DND.ca/menu/press/Reports/Health/ health-study-e-vol1-TOC.htm]

[27.] Unwin C, Blatchley N, Coker W, et al. Health of U.K. service men who served in the Persian Gulf War. Lancet 1999; 353: 169-78.

[28.] Fukuda K, Nisenbaum R, Stewart G, et al. Chronic multisymptom illness affecting Air Force veterans of the Gulf War. JAMA 1998; 2280:981-88.

[29.] Davidoff AL, Keyl PM. Symptoms and health status in individuals with Multiple Chemical Sensitivities syndrome from four reported sensitizing exposures and a general population control. Arch Environ Health 1996; 51:201-13.

[30.] Nethercott JR, Davidoff LL, Curbow B, et al. Multiple Chemical Sensitivities syndrome: toward a working case definition. Arch Environ Health 1993; 48:19-25.

[31.] Rossi JR. Sensitization induced by kindling and kindling-related phenomena as a model for multiple chemical sensitivity. Toxicology 1996; 111:87-100.

[32.] Miller CS. White paper: chemical sensitivity--history and phenomenology. Toxicol Ind Health 1994; 10:253-75.

[33.] Bell IR. White paper: neuropsychiatric aspects of sensitivity to low-level chemicals--a neural sensitization model. Toxicol Ind Health 1994; 10:277-309.

[34.] Zweig MH, Campbell G. Receiver-operating characteristics (ROC) plots: a fundamental evaluation tool in clinical medicine. Clin Chem 1993; 39:561-77.

[35.] Nunnally JC, Bernstein IH. Psychometric Theory, 3rd ed. New York: McGraw-Hill, 1994.

[36.] SPSS Base 7.5 for Windows. User's Guide. Chicago, IL: SPSS Inc., 1997.

[37.] Stata User's Guide. College Station, TX: Stata Press, 1997.

[38.] Kipen HM, Hallmann W, Kelly-McNeil K, et al. Measuring chemical sensitivity prevalence: a questionnaire for population studies. Am J Public Health 1995; 85:574-77.

[39.] Schwartz GE, Bell IR, Dikman ZV, et al. EEG responses to low-level chemicals in normals and cacosmics. Toxicol Ind Health 1994; 10:633-43.

[40.] Simon GE, Katon WJ, Sparks PJ. Allergic to life: psychological factors in environmental illness. Am J Psychiatry 1990; 147:901-06.

[41.] Miller C, Mitzel H. Chemical sensitivity attributed to pesticide exposure versus remodeling. Arch Environ Health 1995; 50: 119-29.

Financial support for the project was provided by the Ontario Ministry of Health. The results and conclusions, however, are those of the authors, and no official endorsement by the Ministry is intended or should be inferred. The senior author was supported by a Fellowship Award from the Medical Research Council of Canada and the Alberta Heritage Foundation for Medical Research.

Submitted for publication May 22, 2000; revised; accepted for publication November 28, 2000.

Requests for reprints should be sent to D. J. Kutsogiannis, M.D., M.H.S., Division of Critical Care Medicine, University of Alberta, Edmonton, Room Royal Alexandra Hospital, 10240 Kingsway Avenue, Edmonton, Alberta T5H 3V9, Canada.

D. J. KUTSONGIANNIS (*) Department of Public Health Sciences and Division of Critical Care Medicine University of Alberta, Edmonton Edmonton, Alberta, Canada

ANN L. DAVIDOFF([dagger]) Sheppard Pratt Health System Baltimore, Maryland

(*) Dr. Kutsogiannis was formerly affiliated with the Department of Epidemiology, School of Hygiene and Public Health, Johns Hopkins University, Baltimore, Maryland.

([dagger]) Dr. Davidoff was formerly affiliated with the Division of Occupational Health, Department of Environmental Health Sciences, School of Hygiene and Public Health, Johns Hopkins University, Baltimore, Maryland.

COPYRIGHT 2001 Heldref Publications
COPYRIGHT 2001 Gale Group