Health symptoms and the work environment in four nonproblem

Objectives The objective of this study was to quantify health sylnptorn reports in four "nonproblem" buildings and to assess the relationship between symptolns and air quality measures, workstation characteristics, and psychosocial aspects of the workplace. Methods Environmental sampling was conducted in four office buildings occupied by employees worltillg for the state of Washington in March of 1992. A questionnaire was concurrently administered to buildillg occu-pants. While measured contaminant levels were low, symptoms were frequent. Fifty-five percent of the 646 respondents reported recent symptoms which affected the eyes, nose, or throat and improved when away from work. Sympton~s were not associated with measured contaminant levels, but, rather, with perceptions about air movement, dryness, odors, and noise. Psychosocial factors were less strongly assnciated with symptoms. C O ~ C ~ U S ~ O ~ ~ S Even in nonproblem buildings, symptom reports can be frequent and may represent overall satisfaction with the work environment. In response to symptoms ascribed to air quality problems, it may be appropriate to address employee perceptions regarding the work environment in addition to evaluating environmental characteristics relating to chemicals, biological contamination, air movement, temperature, and humidity.

In recent years, studies of the office work environment have increased dramatically in number: the United States (US) National Institute for Occupational Safety and Health (NIOSH) completed over 500 nonindustrial health hazard evaluations in the 1971-1988 period; in 1990, NIOSH received more than 2000 telephone calls requesting health hazard evaluations concerning indoor air quality in office buildings (1,NIOSH,unpublished results). Several investigations examining office workplace characteristics and employee health symptoms have been conducted in buildings in which large numbers of workers have experienced adverse health effects, sometimes termed "problem buildings," where employees frequently reported respiratory, skin, or central nervous system (CNS) symptoms. In many cases, no single exposure hazard could be identified, but sometimes improvements made to ventilation systems appeared to relate to a reduction in symptoms. A recent study suggested that changes in ventilation rate alone may not lead to a reduction in symptom reporting (at least not when ventilation levels are already adequate) (1,2). In other studies carried out in notlproblern buildings in Great Britain and Denmark, similar symptoms have been observed (3).
The current study examines relationships between employee health and measures of indoor air quality, workstation characteristics, and psychosocial factors in a large group of state government employees working in four "nonproblem" buildings without know11 exposure hazards or high rates of complaints or symptom reporting. The objective was to quantify symptom reports, characterize physical conditions, and identify risk factors that are associated with respiratory, dermal, and CNS symptoms in an environment experienced by employees working in US office buildings.
ington; these employees worlced in four buildings, termed A through D in the following report. The agency functions carried out in these buildings included the administration of the state workers' compensation claiin system and the worlcplace inspection plan. Employee job responsibilities included clerical, administrative, and professional support. Approximately 60% of the employees were women; two-thirds were in the 30-to 50-year age range.

Study buildings
Three buildings were located in semiurban areas; building D was an older facility located in a dense urban area. The structures varied in age and size; three had operable windows (table 1). Building D was extensively renovated in 1987, including the addition of a third floor (the study space). Smoking was prohibited inside all four buildings. Worlcspaces included private offices, shared rooms, and open spaccs with or without partitions. The buildings were included in the Indoor Air Building Assessment Survey and Evaluation program of the US Environmental Protection Agency (EPA) and as such had to fulfill selection criteria stating that they were free of obvious water damage, obvious contaminant sources, or physical characteristics that would significantly influence indoor air quality or occupant comfort (4). Because they fulfilled these criteria and because they had not generated sufficient conlplaints to lead to formal indoor air quality investigations, as suggested by others (3,5), they were considered to be "nonproblem" buildings.

Questionnaire
Potential study participants were invited to complete a self-administered questionnaire in March of 1992. Participation in the survey was encouraged but voluntary. The study was approved by the State of Washington Department of Social and Health Services Hurnan Subjects Researcll Review Board.
The questionnaire items included questions regarding respiratory, mucosal, CNS, and dermal symptoms; demographic characteristics; physical and psycl~osocial attributes of the worlc environment; job satisfaction; worlcpace and influence over responsibilities; and nonworkrelated stressors. The questionnaire was adapted from a variety of instruments used by other researchers, including the EPA Indoor Environmental Quality Survey, the NIOSH National Center for Health Statistics building and Library of Congress building study questionnaires, and the NIOSH Job Stress Instrument (6)(7)(8). The questionnaire responses were anonymous.
The following four groups of symptoms conlmonly associated with complaints about indoor air quality were selected for analysis according to groupings suggested by other investigators: upper respiratory or rnucosal irritation (dry, itching, irritated or watery eyes; sore throat; nasal symptoms or sneezing; tired or strained eyes; dry throat; cough), lower respiratory problems (tight chest; difficulty breathing; shortness of breath; wheeze), CNS symptoms (headache; depression; tiredness; tension or irritability; difficulty remembering; dizziness or lighthcadedness; mental fatigue), and dermal problems (skin rash or dry skin) (3). A case was defined as any individual reporting at least onc of the symptoms The other constructed variables included several scales to score psychosocial job characteristics that have been associated with symptoms reported to be related to indoor air in other studies (ie, overall job satisfaction, decision latitude, hard or fast work, or conflicting supervisory instructions) (8)(9)(10)(11)(12). A scale for "job stress," combining scores for the perception of influence over work and having to work hard (according to the job demands-control model) was also constructed (1 3).

Exposure assessment
Environmental sampling was conducted in the four buildings, concurrent with questionnaire administration. The sampling strategy followed a protocol prepared in cooperation with the EPA (4). For each building, one floor was designated as the sampling zone, and three individual sites were monitored. The sites were considered to be representative of the entire building; additional data were collected from other floors to determine if this assumption was valid. Air samples were also collected outside each building to compare indoor and outdoor contaminant levels.
The sampling methodology has been described in detail elsewhere (4). Each sampling survey was conducted for one day between 0800 and 1700. Each sampling station was equipped to measure total particulates, selected aldehydes and other volatile organic compounds, carbon dioxide, and carbon monoxide. Approximately 60 volatile organic compounds, including aliphatic, aromatic and chlorinated hydrocarbons, ketones, chlorofluorocarbons, and esters, were inventoried. Exposure to such compounds was expressed as the sum of the levels of each compound identified as present in a particular sample. Samples for particulates, volatile organic compounds, and aldehydes were analyzed in an EPAapproved laboratory. Measurements for temperature, relative humidity, noise, and illumination were also recorded.
Viable bacterial and fungal bioaerosols were collected with methods commonly employed in NIOSH investigations of indoor air quality and recommended by the American Conference of Governmental Industrial Hygienists (4). The sampled bacteria included thermophilic (cultured at 55OC) and mesophilic (cultured at 35OC) species. For each sample, the number of colony-forming units per cubic meter (cfu . m-3) was determined. Biological samples were analyzed by the laboratories of the University of Nevada.
Physical characterization of two buildings and their heating, ventilation, and air conditioning (HVAC) systems was also carried out (not reported here).

Statistical methods
Relationships between health outcomes and potential risk factors were examined with the computation of correlation coefficients (Pearson product moment and Spearman), parametric and nonparainetric analyses of variance (ANOVA and Wilcoxon ranked sum tests), a coinputation of odds ratios for categorical comparisons, and multivariate logistic regression (including stepwise) for developing final models. Data from all four buildings were combined for the multivariate statistical analyses; indicator variables were included in the models to identify the building in which a person worked. Potential risk factors that were statistically significantly related to health outcomes in two-way analyses were included in subsequent nlultivariate analyses. Models that included health outcomes, exposure variables, potential confounders, and terms for interactions between exposure and other variables were examined with forward stepwise logistic regression procedures. The criteria used to determine the variables that would remain in the final models included significance levels of 0.05 for entry into and removal from models. SAS (statistical analysis system) software was used (SASPC version 6.04). Table 1 presents the characteristics of the workers and buildings under study. The study population included 646 participants at the four locations, with a total response rate of 70%. It was not possible to determine response rates by age and gender group, as this information was not available by building at the time of the study. (However, when the study was repeated in 1993 and the respondents were compared with the target population, the response rates were slightly higher for the women and the response rate appeared to increase with age.) The study group was 71% female, with an average age of 42 years. Tables 2 and 3 summarize the responses to selected questions from the questionnaire. Fifty-four percent of the employees were at least somewhat satisfied with their overall workstation environment (table 2). Less than half thought their worlcstations were reasonably clean, and substantial numbers felt the environment had problems with temperature, noise, odors, or air circulation (table 2). Respondents from building C gave the poorest evaluation of their facility. Overall, 83% were at least somewhat satisfied with their jobs (table 3). Job demands were high in that 77% felt their jobs required hard or fast work.

Questionnaire findings
A large percentage of the group reported health symptoms. Over half of the reported upper respiratory symp-   Table 5 shows selected environmental measurements made for the four facilities. The formaldehyde, acetaldehyde, and particulate levels were below the legal and recommended exposure limits. The summed levels of the volatile organic compounds were also below the a Symptoms occurred 1-3 d a week in the last four weeks (or more) and symptoms improved after work.

Environmental measures
Includes occurrence of any of the symptoms appearing below it.

Relationships between health symptoms and work environment characteristics
Upper respiratory or mucosal symptoms. In the two-way analyses, upper respiratory or mucosal irritation was sta-Scand J Work Envlron Health 1995, vol21, no 1 tistically significanlly related to the personal characteristics of gender (women being approximately twice as likely as men to report symptoms), age and education level (likelihood of symptoms decreasing with increasing age and education level), and contact lens use (wearers of contact lenses being twice as likely to reporl symptoms). Cigarette smoking was not related to the occurrence symptoms. Several self-reported underlying medical conditions were also related to the development of symptoms, including a history of migraine headaches, asthma, hay fever, or allergy to dust or molds.
The medical symptoms differed by building. The rates were significantly higher for buildings A and C. The syrnptoins did not appear to be related to the environmental air levels for aldehydes, carbon dioxide, humidity, or bacteria. The symptoms were statistically signifi- cantly related to increases in volatile organic compounds, particulates, and noise (and a decrease in illunlination), although the changes in risk were small. Several job characterislics were related to the development of upper respiratory symptoms, including perceptioils of workstation cleanliness, comfort, privacy, and overall satisfaction with the physical workstation. Many variables related to psychosocial job characteristicsincluding overall job satisfaction, influence over job tasks, working hard or fast, conflict in supervisory instructions, and job stresswere statistically significantly related to the development of upper respiratory or mucosal irritation, but the odds ratios were not large.
Stress from personal life outside the workplace, including stress from childcare responsibilities, volunteer work, part-time schooling, or lengthy commutes, did not appear to contribute to the reporting of upper respiratory or mucosal issitation.
Final multivariate analyses were carried out to determine which factors were the most strongly related to the development of respiratory symptoms when all were considered simultaneously. The building in which a person worked, the air contaminant levels, temperature, and noise were not related to the development of respiratory symptoms; however, factors related to a person's perception of work environment characteristics did appear to be related to respiratory symptoms. Employees who thought their work environment was not clean or had too little air movement or thought the atmosphere was too dry or noisy were more likely to report upper respiratory symptoms than those who did not (table 6).
Other factors that continued to be related to the development of symptoms in multivariate analyses were contact lens use, overall satisfaction with the physical work environment, report of sensitivity to chemicals found in the workplace, and job stress.
Central nervous system symptoms. Statistical analyses similar to those for upper respiratory or mucosal irritation were carried out to determine which factors were the most highly predictive of CNS symptoms. In the multivariate analyses, the reporting of CNS symptoms was related to several factors (table 6). The strongest was gender; women were 2.6 times more likely to report symptoms than men.
Lower respiratory and dermal symptoms. In the multivariate analyses, lower respiratory symptoms were statistically significantly associated with the perception of too little air movement, odors, and job stress (table 6). Factors related to the reporting of dermal symptoms included gender, perception of workplace odors, and workstation discomfort (table 6).

Discussion
The air contaminant levels were generally low in the buildings under study in con~parison with current permissible exposure levels. [It can be noted, however, that some persons have argued that US permissible exposure levels are too high to protect the health of those exposed (14)l. The contaminant levels did not appear to be associated with health symptoms; this finding is similar to the results of other studies (15, 16). The measured chemical levels did not vary according to the study participants' perceptions regarding environmental conditions. There were no differences in the mean measured contaminant levels for the participants who did or did not rate their work areas as clean or as having too much or too little air movement or as having chemical or other odors. There may be several reasons for this lack of differences. First, environmental samples were collected in three locations on one floor of each building. It is possible that contaminant levels fluctuate considerably by site and that these sampling stations did not represent individual worksites. However, when the analysis was limited to only those working on the floors where the air samples were collected, the results did not change. Of course it is possible that the floor-specific information was still insufficient to describe conditions at individual workstations. Second, it is possible that intermittent exposures or point sources were particularly effective in inducing symptoms and our monitoring techniques did not record at this level of detail (as results were recorded as time-weighted averages over an 8-h sampling period). Human beings may be sensitive to minute environlnental changes that cannot be measured by usual industrial hygiene monitoring methods. Third, we may have failed to measure some unltnown but importa~~t environlnental exposure or characteristic. Finally, and most likely, it is possible that the observed levels were too low and invariable for any relationship with symptoms to be observed.
Seventy percent of the eligible employees participated in the survey, the response rates varying from 57 to 82% by building. It is possible that the respondents were not representative of the entire eligible group, if those who experienced health problems were more (or less) likely to participate than those who did not. There was some indication that the response rates were positively correlated with the symptom report rates for these buildings (although correlations were based on only four observations and were not statistically significant). In addition, it should be pointed out that these symptoms were selfreported; thus they represent subjective evidence of health problems that might be over-or under-reported. For example, 12.4% of the study group reported a personal history of asthma; this percentage appears high in comparison with population-based estimates of this disorder (17).
Other studies have also examined relationships between health symptoms and environmental characteristics in office buildings. Some of these were initiated because of worlters' complaints about their worltplaces. Such buildings are generally referred to as "problem" buildings, where symptom report rates are high. In the cuwent study, nonproblem buildillgs were selected for examination. While it is difficult to compare the results with those from other studies because of differences in workplace characteristics and ways in which health outcomes are defined, our rates were similar to those observed in other groups. In a study by Sltov et a1 (5) approximately 25, 16, and 11 % of the study group reported mucous membrane, nasal, and eye irritation, respectively. In another study of a "problem" building, 30% reported mucous membrane irritation and 35% reported headaches (18).
Other studies have shown several factors to be related to the development of "indoor air quality" symptoms, including female gender, age, hay fever, contact lens use, exposure to biological agents, inadequate ventilation, worlt with copiers or carbonless paper, type of worltstation, and a number of factors related to job stress and satisfaction (5, 10, 18-24). Our study had many similar findings (although the relationships were not strong and often became statistically nonsignificant when a number of factors were considered simultaneously); some of the similar findings are an increase in symptom reporting for younger age groups, female gender, contact lens wearing, self-reported allergies and hay fever, and associations with several factors related to job stress and satisfaction and perceived physical workspace conditions (worltstation cleanliness and comfort, air movement, noise, odors, etc). Our study was also similar in that no consistent relationships between symptoms and measured air contaminant levels were observed. Symptoms were generally the most strongly associated with perceptions about worltstation comfort and cleanliness, air movement, dryness, odors, and noise. Psychosocial aspects of the work environment were less strongly associated with symptoms.
It is unexpected and notable that, in the current study, cigarette smoking did not appeas to be associated with respiratory symptoms. In other studies in which the effects of smoking were considered, the results varied. In some, smoking was associated with respiratory symptoms (10, 1 I), whereas in another, it was not (21).
In a study such as ours, in which the health outcomes of interest relate to the self-reporting of symptoms and many potential causative factors are examined, it is sometimes difficult to determine relationships among them (25). Because of the cross-sectional nature of the survey, it is possible that the observed associations may not represent cause-effect relationships, but rather may be measures of the same perception. As has already been mentioned, many employecs reported more than one indoor air-related symptom group, a finding which might be expected. In addition to these interrelationships, the liltelihood of indoor air-related symptoms was associated with the reporting of musculoslteletal problems. For example, those with upper respiratory or rnucosal irritation were eight times as likely to report back problems as those without such irritation. One would expect the two types of problems to be independent, unless they were associated with the same relatively nonspecific causal factors. Other observed associations included relations between health symptoms and perceived environmental characteristics (such as problems with air movement or dryness). For these associations in particular, it seems possible that both may be measures of the same perception (eg, eye irritation may be caused by air dryness, but irritation might also lead a person to believe the environment must be dry). Thus, although we have identified a number of factors that are related to reported symptoms, it is not clear that these factors are causal.
A comparison of the questionnaire responses with the results of more objective measures of health outcomes, such as the neuropsychological test battery used by Middaugh et a1 (26), and the measures of bioinicroscopic eye manifestations used by Franclc et a1 (27), might help to determine the sensitivity and specificity of symptom survey instruments. It is of interest that, in both of these studies, objective measures were associated with reports of subjective symptoms, but the relationships were not necessarily strong. For example, in the study by Franck et a1 (27) the strongest relationship between an objective sign (absence of foam at the inner eye canthus) and the reporting of eye complaints had an odds ratio of 2.9.
In conclusion, our results suggest that, even in nonproblem buildings, reports of health symptoms can be frequent. Although our study did not suggest that symptoms were associated with air contaminant levels or other measured aspects of the physical environment, it is possible that the health symptoms were related to microenvi-ro~lmental factors that we were not able to measure. It is also possible that symptoms reported in a questionnaire represent perceptions regarding workplace physical comfort and cleanliness rather than objective measures of health.