|
|
| |||||||||||||||||||||||||||||||||||||||||
Background The sick building syndrome is the term given to a constellation of symptoms reported by workers in modern office buildings, hypothesized to occur when the supply of outdoor air is reduced, because of the accumulation of contaminants arising from within the building. We undertook this study to determine the effect of changing the supply of outdoor air in four office buildings on the symptoms reported by workers and their perception of the indoor environment.
Methods Within each of three consecutive two-week blocks, the ventilation systems in each building were manipulated, in random order, to deliver to the indoor environment an intended 20 or 50 ft3 (0.57 or 1.4 m3) of outdoor air per minute per person for one week at a time. Each week, the participants, unaware of the experimental intervention, reported symptoms and the indoor environment was thoroughly evaluated.
Results Of 1838 eligible workers in the four buildings, 1546 (84 percent) participated in the study. The supply of outdoor air averaged 7 percent and 32 percent in the ventilation systems and 30 and 64 ft3 (0.85 and 1.8 m3) per minute per person in the work sites at the lower and higher ventilation levels, respectively. These changes in the supply of outdoor air were not associated with changes in the participants' ratings of the office environment or in symptom frequency (crude odds ratio, 1.0; 95 percent confidence interval, 0.9 to 1.1). After work-site measures of ventilation, temperature, humidity, and air velocity were included in the regression analysis, the adjusted odds ratio was also 1.0 (95 percent confidence interval, 0.8 to 1.2).
Conclusions Increases in the supply of outdoor air did not appear to affect workers' perceptions of their office environment or their reporting of symptoms considered typical of the sick building syndrome. .
A large number of persons may be potentially affected, since more than half of all workers in industrialized nations work in offices,4 and 50 to 80 percent of approximately 9000 workers surveyed in Britain,5 Denmark,6 and Sweden7 reported symptoms typical of the sick building syndrome. On the basis of these and other prevalence surveys, it is estimated that at any one time 10 to 25 million workers in 800,000 to 1.2 million commercial buildings in the United States will have symptoms typical of the sick building syndrome8. The resultant economic impact is considerable,8 because symptomatic workers have reduced productivity9 and increased absence from work10.
It has been hypothesized that in mechanically ventilated buildings, symptoms arise because the concentrations of contaminants from indoor sources increase when the supply of outdoor air is reduced3,11,12. Despite little supportive evidence, this hypothesis has gained such widespread acceptance that the American Society of Heating, Refrigeration, and Air Conditioning Engineers recommended an increase in the minimal supply of outdoor air from 5 to 10 ft3 (0.14 to 0.28 m3) per minute per person in 198113 and another increase from 10 to 20 ft3 (0.57 m3) per minute per person in 198914. Further increases have been proposed15,16 that not only would increase energy costs,8 but could also exceed the capacity of the heating, ventilation, and air-conditioning systems in many buildings to maintain temperature and humidity in the ranges considered ideal for human comfort and well-being13,17.
This study was undertaken to test the hypothesis that symptoms of sick building syndrome could be reduced in mechanically ventilated office buildings by increasing the supply of outdoor air from 20 to 50 ft3 (1.4 m3) per minute per person.
Methods
Overall Study Design and Experimental Intervention
A randomized double-blind multiple-crossover trial was conducted to estimate the effect of changes in the outdoor-air supply in four office buildings. During three consecutive two-week blocks of time, building ventilation systems were manipulated to deliver an expected 20 or 50 ft3 of outdoor air per minute per person to the indoor environment, corresponding to indoor carbon dioxide concentrations of 1000 or 600 ppm17. Within each two-week block, the ventilation level was increased for one week and decreased for the other. The order of the increase and decrease in the ventilation level in each block was selected randomly by the study engineers and was not known by the data-collection personnel or the study participants. The dampers supplying outdoor air to the buildings were manipulated each Friday afternoon and then remained in these positions for the following week. Two buildings were studied simultaneously, with opposite ventilation levels in each building, to minimize the potential effect of reduced symptom reporting over time that was found in an earlier pilot study18.
Study Population
The study buildings had sealed windows and mechanical ventilation systems and were between 3 and 20 years old. All but one (building 4) had nonsmoking policies. Two buildings were studied in the spring and two in the fall of 1990.
A survey was conducted of all full-time workers in two buildings of 8 and 9 floors (buildings 1 and 3) and of all full-time workers on eight floors in two buildings of 23 and 25 floors (buildings 2 and 4). Eligible workers were considered to be participating if they completed at least two of the six weekly questionnaires, to have dropped out if they completed only one weekly questionnaire, and not to have responded if they completed no questionnaires.
Collection of Environmental Data
The supply of outdoor air at each work site was estimated by measuring the carbon dioxide concentrations at 8 to 12 work sites per floor in midafternoon on the day that the weekly questionnaire was completed. The values for the carbon dioxide concentrations were converted to cubic feet per minute per person with the following accepted formula14: carbon dioxide in cubic feet per minute per person = [(0.75) (1,000,000 - indoor carbon dioxide)/60]/(indoor carbon dioxide - outdoor carbon dioxide), where 0.75 ft3 (0.02 m3) per minute is the amount of carbon dioxide produced by the office occupant, 1,000,000 ppm is the maximal concentration of pure carbon dioxide, indoor carbon dioxide is the average concentration (in parts per million) of carbon dioxide at each work site on the afternoon of the test day, and outdoor air is the concentration (in parts per million) of outdoor air on the test day.
The outdoor-air supply calculated for each participant was the carbon dioxide value obtained at the nearest work site. The carbon dioxide concentrations in the supply air, return air, and outdoor air of the building's ventilation system were used to verify the level of ventilation in each building14 (the percentage of outdoor air = 100 - [(supply air - outdoor air)/(return air - outdoor air)] x 100).
The major determinants of human comfort -- air velocity, temperature, and relative humidity13,17 -- were measured in the morning and afternoon on the same day and at the same sites as carbon dioxide was measured. The following indoor contaminants were measured with methods previously reported18,19 once weekly at two to four sites per floor: formaldehyde, nitrogen oxides, carbon monoxide, total volatile organic compounds, total and viable airborne fungal spores, and total airborne dust.
Questionnaires
The workers completed self-administered questionnaires at base line (one to three weeks before the study began) that provided data on their personal, smoking, medical, and work histories and their state of mind, according to the Bradburn index of emotional well-being20. To minimize recall bias, in midafternoon on Wednesday or Thursday of each of the six study weeks, they completed a five-minute questionnaire on their environment and the presence or absence of symptoms experienced that day. The symptoms asked about -- headache, fatigue, difficulty concentrating, cough, and irritation of the eyes, nose, or throat -- were those reported most frequently in our pilot study18 and in other studies6,7,21. The participants' overall impressions of the office environment (good, poor, or variable) were reported, and changes in their impressions from the previous week (better, the same, or worse) were noted. They also rated nine environmental conditions (temperature, humidity, air circulation, lighting, noise, space, dust, smoke, and odors) on an ordinal scale. The scores for the nine items were added and rescaled into a total environmental dissatisfaction score, in which a score of 0 indicated that all conditions were ideal, and a score of 100 that all conditions were terrible.
Data Analysis
The primary outcomes were weekly reports of any symptom, mucosal symptoms (irritation of the nose or throat or a cough), and systemic symptoms (headache, fatigue, or difficulty concentrating)5,6,22. The primary analysis was a comparison within subjects of the frequency of symptoms under the different ventilation conditions. Combining only discordant responses from the same participants within each two-week block, the crude odds ratio represents the average number of participants who were symptomatic only during the weeks of increased ventilation, divided by the average number of participants who were symptomatic only during the periods of decreased ventilation23. The odds of reporting symptoms at different carbon dioxide concentrations, adjusted for other environmental measurements and ratings, were calculated with conditional logistic-regression analysis, in which the symptoms and actual work-site concentrations of carbon dioxide during all six weeks were analyzed24. A crude or adjusted odds ratio of less than 1.0 would indicate a reduction in symptom reporting during periods of increased outdoor-air supply.
Additional analyses between subjects were conducted to assess the relation of the participants' characteristics to symptoms, the association of differences between work-site environmental conditions and symptoms (logistic-regression analysis), and the variability of environmental conditions within and between work sites (analysis of components of variance)25,26.
Results
Study Population
In the four buildings studied, 1838 eligible workers were identified, of whom 182 did not respond to the questionnaires, 110 agreed to participate but subsequently dropped out of the study, and 1546 (84 percent) participated (Table 1). As compared with the participants, the workers who did not respond to the survey were more likely to be men, and those who dropped out were less likely to have a history of atopic illness or to work in clerical positions. All subsequent results refer to the 1546 participants, of whom an average of 82 percent completed questionnaires each week. The 637 who completed all six weekly questionnaires were more likely to be female, younger, and work in open areas in clerical positions, than participants who completed two to five questionnaires.
|
The average supply of outdoor air at work sites during the study was 64 and 30 ft3 (1.8 and 0.85 m3) per minute per person when outdoor-air dampers were open and closed, respectively, and the percentage of outdoor air in the supply air of the ventilation system was 32 percent and 7 percent (i.e., 68 percent and 93 percent of the air was recirculated), respectively (Table 2). In all buildings, reducing the supply of outdoor air resulted in considerably higher concentrations of formaldehyde and volatile organic compounds. The experimental changes in building ventilation conditions were also associated with changes in carbon dioxide concentrations at work sites, as intended, but were not associated with any significant changes in the temperature, air velocity, or humidity of the work sites.
|
|
|
On average, a little more than half the participants reported at least one symptom each week, and the proportion reporting symptoms within each two-week block was virtually identical at the two levels of ventilation (Table 4). There was a strong temporal trend in symptom reporting, since the frequency of symptoms declined from 55 percent to 35 percent from the first to the sixth weeks in buildings 1 and 2, and from 66 percent to 55 percent in buildings 3 and 4.
|
|
|
The following personal and work-site characteristics were significantly associated with the reporting of any symptom: female sex (odds ratio, 1.7; 95 percent confidence interval, 1.5 to 1.9), atopic illness (odds ratio, 1.4; 95 percent confidence interval, 1.3 to 1.6), use of the computer for more than four hours each day (odds ratio, 1.3; 95 percent confidence interval, 1.1 to 1.5), working in an open area (odds ratio, 1.1; 95 percent confidence interval, 1.0 to 1.2), and younger age. These characteristics were also associated with the reporting of mucosal or systemic symptoms. No extrabinomial variation was found in the frequency distribution of discordant responses within trials, and symptoms were not associated with the level of ventilation in more susceptible groups, such as female participants or those with a history of atopic illness (i.e., no subgroup could be identified in whom responses were associated with the level of ventilation).
The variation in temperature and air velocity during the six-week study period was minimal within the work sites, but substantial between work sites, even those located on the same floor. These differences in temperature and air velocity between sites were associated with differences between subjects in symptom reporting, as shown in Figure 1.
Discussion
In this study, increasing the supply of outdoor air in mechanically ventilated office buildings was not associated with either improved environmental ratings or a reduction in the number of symptoms reported by the participants. Key features of the design were that three randomized crossover trials of experimental manipulation of outdoor-air supply were conducted, during which the participants, who were unaware of the study intervention, completed standard questionnaires under different environmental conditions. The occurrence of reporting bias28 should have been reduced by keeping the participants unaware of the ventilation conditions, and the effect of reduced reporting over time, noted in our pilot study,18 should have been reduced by counterbalancing the experimental sequence in pairs of buildings studied simultaneously. The estimate of the effect within subjects should have reduced the effect on symptoms of differences in personal and work-site characteristics5,6,7 and the effect of nonparticipation bias (which was already minimized by the participation rate of 84 percent). The likelihood of potentially misclassifying participants' exposures22 to outdoor-air supply was reduced by measuring carbon dioxide at more than 250 work sites (i.e., that of every sixth participant).
The failure to detect a significant association between the level of ventilation and the number of symptoms reported may have occurred for several reasons. The participants may not have had the sick building syndrome. On average, however, over 50 percent of the participants reported at least one symptom each week, comparable to the frequency of symptoms reported in other studies of sick buildings5,6,7. In addition, the symptomatic participants had the same personal and work-site characteristics that have been associated with the reporting of symptoms in other studies5,6,7.
The buildings studied may not have been sick. The World Health Organization has defined a sick building as one with an "excess" frequency of symptoms among occupants,1 but there are no established norms for the frequency of these symptoms among office workers. In studies in Britain,5 Denmark,6 and Sweden,7 the reported frequency of symptoms among workers in "problem" buildings was similar to that among workers in "non-problem" buildings. In the absence of any clear definition, the buildings used in this study were not selected on the basis of a previous identification of the building as sick, but rather because the characteristics of the buildings were similar to those of buildings in which the sick building syndrome has been described5,6,7,21. Some buildings, but not the four we studied, may have increased concentrations of certain contaminants that lead to symptoms among workers. In such buildings, increasing the supply of outdoor air could potentially reduce the frequency of symptoms. We studied buildings constructed 3 to 20 years ago, typical of the majority of existing office buildings in North America. In new buildings, which have increased concentrations of contaminants released from construction materials and new furnishings,29 increasing the supply of outdoor air may be beneficial.
The supply of outdoor air in the work sites may have been poorly estimated from measurements of carbon dioxide. Current ventilation standards for buildings in North America are expressed in terms of cubic feet per minute per person, calculated from carbon dioxide concentrations,14 which can be measured rapidly and directly with portable instruments. As a result, the carbon dioxide concentration is the most widely used, commonly reported, and easily understood measure for estimating the level of ventilation in a building, although it is produced only by the human occupants, who contribute just 13 percent of the total indoor load of other contaminants15. Estimating the outdoor-air supply with tracer gas decay is more accurate, but is also complex and time consuming30 and could not have been done at many sites. The practice of basing the key exposure variable -- the change in ventilation conditions from week to week -- on measurements of carbon dioxide should have been valid, because the buildings' occupants did not change and other sources of contaminants should have remained constant.
The reduction in the supply of outdoor air may have been insufficient to produce symptoms. Although outdoor air composed an average of only 7 percent of the total air supply in the ventilation system (i.e., 93 percent was recirculated air) during the weeks in which the level of ventilation was reduced, it proved impossible in the three older buildings to reduce average work-site levels of outdoor air to less than 30 ft3 per minute per person, because of the infiltration of outdoor air through the buildings' shells. However, all participants were exposed to substantial changes in the supply of outdoor air, which resulted in significant changes in the concentrations of formaldehyde and volatile organic compounds -- contaminants that are produced almost entirely from indoor sources15,29 and removed by the exchange of indoor air with outdoor air11,12,29.
The experimental intervention may have been confounded by the effects of temperature, relative humidity, and air velocity, all of which have important effects on workers' perceptions of indoor-air quality17,31 and on symptoms13,17,31. We measured the temperature, air velocity, and relative humidity at the work site of every sixth worker twice daily each week and used multivariable analysis to adjust for any potential effect of these variables. These measurements also varied independently of the experimental changes in ventilation conditions. Finally, temperature and air velocity should not have contributed to differences in symptoms within subjects, because they varied little at the same work sites throughout the study.
In this study, temperature, humidity, and air velocity varied significantly between work sites and contributed to differences between subjects in the number of symptoms reported. This finding and the findings of others22 suggest that the microenvironmental conditions of individual employees' work sites may be important determinants of exposure.
In conclusion, the supply of outdoor air was experimentally increased, from an average of 7 percent of the total air supply in the ventilation system and 30 ft3 of outdoor air per minute per person at work sites to an average of 32 percent and 64 ft3 of outdoor air per minute per person, respectively, and these increases were associated with significant changes in the concentrations of contaminants at different work sites. These changes were not associated with the participants' ratings of the environment or with the number of symptoms reported that were considered typical of the sick building syndrome. We believe that this research method can be used in further studies to establish a scientific basis for ventilation and contaminant standards to ensure the health and safety of the majority of North American workers.
Supported by the National Health Research Development Program and the Respiratory Health Network of Centres of Excellence.
We are indebted to the owners and occupants of the four buildings involved, to Mr. Bob Robb and Mr. Pierre Marcotte for technical assistance, and to Drs. Margaret Becklake, Pierre Ernst, and Peter Macklem for critical review of the manuscript.
Source Information
From the Departments of Epidemiology and Biostatistics (R.M., Robyn Tamblyn, J.H.) and Medicine (R.M., Robyn Tamblyn), the Respiratory Epidemiology Unit (F.M.), and the School of Occupational Health (P.F.), McGill University, Montreal, and Engineering Interface, Toronto (Robert Tamblyn).
Address reprint requests to Dr. Menzies at the Respiratory Epidemiology Unit, McGill University, 1110 Pine Ave. West, Montreal, QC H3A 1A3, Canada.
References
| |||||||||||||||||||||||||||||||||||||||||
Related Letters:
Sick Building Syndrome
Edelman P. A., Hethmon T., Boatman J. F., McCarthy J. F., Farnham J. E., Small G. W., Bromberg J. S., Menzies R., Tamblyn R.
Extract |
Full Text
N Engl J Med 1993;
329:503-504, Aug 12, 1993.
Correspondence
Building-Related Illnesses
Johanning E., Auger P. L., Reijula K., Kroenke K., Menzies D., Bourbeau J.
Extract |
Full Text
N Engl J Med 1998;
338:1070-1071, Apr 9, 1998.
Correspondence
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | TERMS OF USE | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2009 Massachusetts Medical Society. All rights reserved. |
