Grantee Research Project Results

Final Report: Bioaerosols, Health, and Productivity in a Large Office Building

EPA Grant Number: R824797
Title: Bioaerosols, Health, and Productivity in a Large Office Building
Investigators:
Institution: Harvard University
EPA Project Officer: Chung, Serena
Project Period: October 1, 1995 through September 30, 1998
Project Amount: $439,035
RFA: Indoor Air Quality in Large Office Buildings (1995) RFA Text |  Recipients Lists
Research Category: Air Quality and Air Toxics , Air

Objective:

Building-related illnesses are of increasingly common concern. However, the causes and cures remain obscure. This project was designed to longitudinally evaluate the role of environmental factors on reported symptoms and measured working efficiency in a group of office workers.

Although a single large building was to be the focus, difficulties in obtaining access forced the use of working spaces in five different buildings. This slowed the project and increased the difficulties involved in gathering data, but expanded the range of environmental conditions available for study.

Overall we collected continuous data on CO₂, RH, and temperature from 28 sites in 5 buildings over a period of 18 months. We collected environmental samples (air samples for fungi and bacteria, dust samples for estimation of 'dustiness', culture of fungi and bacteria, and assay of allergens) on a total of 17 occasions from these sites. We also recruited 180 individuals, and recovered 729 six-week questionnaires, and 4103 weekly questionnaires. Measures of working efficiency were collected from 171 individuals, with an average of about 2 tests/person/week, and a total of 2176 tests overall.

Methods evaluations revealed that Andersen conversions were not necessary for the data set, and that MEA and DG-18 culture media produced very similar results so that the resulting data could be combined.

Using Principal Component Analysis (PCA), airborne fungal recoveries grouped into four factors which explained 53% of the variance. Airborne fungal concentrations varied seasonally with the highest median value in August and the lowest in January. Total airborne fungal concentrations were negatively correlated with CO₂ and positively related to relative humidity. PCA factors 1 and 2 were negatively correlated with CO₂ and positively correlated with RH

Similar studies for dustborne fungi. These populations clustered into three factors using PCA. Concentrations of dust fungi were related to temperature and CO₂ levels. Cat and dust mite allergens were consistently present in dust in all buildings.

Analysis of this large data set is continuing, with current emphasis on relationships between weekly and daily symptom reporting and environmental variables. The data are the subject of two doctoral theses, and have stimulated creation of a bioaerosols analysis working group at the School.

Summary/Accomplishments (Outputs/Outcomes):

Overview of the problem
Office buildings are increasingly the focus of complaints related to air quality. Because air quality is rarely studied in well-maintained, non-complaint buildings, there are no reference guidelines on which to base recommendations for remedial actions when complaints do occur. These data are especially lacking for airborne biological agents such as bacteria and fungi, so that the presence of any of these common organisms is often interpreted as constituting a problem.

In addition to the lack of data on exposures in non-complaint buildings, it is not clear how air quality is related to perception, comfort, and productivity in the office environment. Thus, decisions on whether or not to attempt to improve air quality are often based solely on the expense related to the improvement rather than the overall equation balancing remediation expense against potentially more costly losses in productivity.

Goals of these studies
The Large Building Study is designed to extend the protocol developed for the EPA-BASE studies by more intensively studying a single building over time, with expansion of the assessments for biological agents, and by adding a component designed to evaluate the effects of air quality parameters on worker productivity. To accomplish this overall goal:

1. We have established baseline levels and patterns of variability for important indoor air quality measures over time in non-complaint office buildings. These measures include temperature, relative humidity, carbon dioxide levels, ventilation rates, and water activity in materials. We have collected sequential samples from each building over 18 months for analysis of dust and airborne fungi and bacteria, dustborne arthropod and animal allergens, and dust and airborne endotoxin.
2. We have used brief questionnaires administered bimonthly over 12 months to monitor comfort, air quality perception, and symptoms known to be related to specific air contaminants in a group of workers in each building. We used a computer test developed by NASA to measure productivity in these same workers.
3. We will compare symptom prevalence and proficiency on the computer test with environmental measurement data taking repeated measures and co-variation into account.
4. We will these results to similar data collected during the EPA BASE studies in a large cross-section of office buildings.

Problems with building selection
The protocol, as originally written, involved the use of a large building owned by a Boston-based investment company employing more that 1500 people in the building. The advantages of this building were:

1. the large numbers of people working in more or less open office spaces, and all doing similar jobs;
2. the fact that the company monitors productivity by monitoring telephone usage;
3. the fact that upper management of the company understood the value of the study and agreed (in writing) to collaborate with our group.

After the study was funded, the company management introduced our protocol to their attorneys who strongly vetoed any involvement by the company. It spite of considerable efforts we were unable to obtain access to the building. We then explored other large commercial buildings. Unfortunately, the attorneys uniformly considered the study to present a legal risk, and we were unsuccessful.

Instead, we approached non-profit institutions in the Boston area and were able to obtain permission to use 4 University office buildings and a downtown office building partially occupied by a State agency. Implications of the use of these multiple buildings are the following:

Numbers of participants: There was no single building with enough employees in units where supervisors were willing to cooperate to allow us to use a single building. We have thus been forced to use sites in a total of four different buildings to obtain the necessary number of participants.

Ventilation systems: Each building has its own ventilation system. This forced us to do the ventilation assessments ourselves, because we had only arranged for our outside contractor to do a single building. We do have consulting access to the same company (that does many of the BASE studies), and have followed their protocols exactly but used our own personnel.

Number of employees available/sampling site: The maximum number of employees available at any single sampling site in the chosen buildings is 10. This means that we have had to increase the number of sampling sites, the number of samples collected, and therefore the cost of the sampling protocol.

Productivity: Neither the University or the State agency maintains productivity records on office personnel. Fortunately, through our contacts in NASA, we have been able to gain access to a set of computer tests that are used to assess the response of shuttle astronauts to their environment. We are using two tests. The first uses pattern recognition (recognizing whether or not a letter or number is in an appropriate orientation); the second tests short term memory. The set of tests has been evaluated by the US Air Force in 5000 office workers, and this control data base is available for these studies. We feel that the use of this test is an improvement to the project. The test will eventually be commercially available, and could be used widely in future studies. It also is likely to more generally and more accurately measure the effects of changes in air quality than the productivity measures used by investment companies.

Because of the unavailability of the original commercial building, we have had delays in recruiting, have had to significantly increase the number of sampling sites, and have had to instigate an external measure of working efficiency. These changes all have been costly, and we have had to make changes in our original protocol to reduce costs.

1. We have eliminated Burkard spore trap sampling following collection of pilot data that indicated that this sampler is insufficiently sensitive for use in our relatively clean buildings.
2. We are archiving dust samples for future analysis of ergosterol and endotoxin.
3. We have reduced the frequency of integrated sampling to once each six weeks (from monthly).
4. We are performing working efficiency measures on only 3/4 of total participants.
5. On the other hand, we have added a weekly questionnaire that addresses relatively short-term changes in symptoms (See attachments). Even with these economies, we have collected important and reliable data with respect to the longitudinal effects of building conditions (especially with respect to bioaerosols) on perception, symptoms, and working efficiency of occupants.

Approaches for completion of the study
Because of our initial difficulties in recruitment, we were not able to complete this study during the grant period. However, we are continuing (and will continue) to bring the studies to completion. To facilitate this effort, we have involved two doctoral students in the project. Both have participated in sample/data collection, and are using the studies as the basis of their doctoral theses. We have divided the work into two large sections to facilitate this approach.

1. Jasmine Chao is using the data from the University buildings, and is focusing on the physical environmental parameters and bacteria and fungi as well as the questionnaire and working efficiency data from all the buildings.
2. Jenny Lee is analyzing all of the allergen data, and is focusing on the physical environmental parameters from the State agency and will do the EPA/BASE comparisons.

Both of these students are required to prepare at least 3 publishable manuscripts to satisfy their thesis requirements. As these students prepare each paper for publication, copies of the final manuscripts will be submitted to the EPA.

Final study design

Buildings
We have used spaces within five different buildings for these studies.

University buildings
The four University buildings used are in busy urban areas. Buildings 1, 2, and 3 are 14, 4, and 10 stories respectively, forming an inter-connected campus building complex. Four different air-handling units controlled the 7 administrative offices selected from these three buildings. All four are constant air volume systems. Fourteen sampling sites were selected from Building 4, a 10-story office building on a different campus. The building is supplied by 2 air handling units and 380 fan coil units. Temperature sensors are installed in almost every room, and are monitored by the computer system in the building manager's office.

State office building
The State office building is situated in downtown Boston surrounded by numerous historical buildings. The building itself is a newly renovated modern high-rise (48 floors) air-conditioned building. The collaborating agency is located on the first 8 floors with its own entrance. The remaining 40 floors are residential condominiums with a separate entrance.

Sampling site and participant selection
Meetings were held to acquaint employees and union representatives with the study. All attendees were given information regarding the study and consent forms. In addition, walkthroughs were conducted and groups of people in especially suitable locations (5-10 people in open areas) were encouraged to review our literature and to participate. Based on consent forms received, a number of sites in each building were chosen. These are summarized in Table 1.

Table 1. Initial site and participant recruitment

Building	# Sites	# Participants
University buildings	20	105
State office building	8	75
Totals	28	180

Environmental samples

	Sampling	Analysis	Frequency
CO2		Direct reading	Continuous
Temp.		Direct reading	Continuous
RH	Wet/dry bulb temp.	Wet/dry bulb calculation	Continuous
Aw	RH, Surface temp	Calculation	6-week events*
Dust Fungi	Vacuum samples, floor and chairs	Culture: DG18, MEA	6-week events*
Dust Bacteria	Vacuum samples, floor and chairs	Culture: R2A	6-week events*
Air Fungi	N-6 Andersen	Culture: DG18, MEA	6-week events*
Air Bacteria	N-6 Andersen	Culture: R2A	6-week events*
Allergens	Vacuum samples Floor and chair dust	Immunoassay	6-week events*
Surface dust	Adhesive film contact	Densitometry	6-week events

*Each event: 1 week, 2 days/week, 2 sample sets/day/site=4 sample sets/site/event

Human studies

Confidentiality

The human studies protocol was assessed by our IRB and approved. Briefly, each participant was given a code number known only to the investigator collecting data who maintains a code key. The code investigator for each building collected base and 6-week questionnaires. Weekly questionnaires were deposited in a locked box for collection by the code investigator. Working efficiency tests are performed and data stored only by code in the computer.

Table 3. Questionnaire types

Base	EPA base questionnaire
6-week	Base questionnaire minus questions on parameters not likely to change (e.g., sex)
Weekly	Symptoms experienced over the past week

Working efficiency tests
A standardized computer test, Nova Scan A (NTI, Inc., Dayton OH) was used to evaluate work performance. Nova Scan A was designed to test higher cognitive functions that might be applicable to jobs that involve high degrees of information processing. This test consists of three specific tasks: spatial visualization, continuous memory and attention monitoring. The first test uses pattern recognition (recognizing whether or not a letter or number is in an appropriate orientation); the second tests short term memory. The two tests are taken together, with each screen presenting one test or the other randomly. After about 20 repetitions to establish an individual baseline for each worker, the test is used to assess excursions from the baseline by assessing scores (% correct answers) and the time it takes to do the test.

Schedule for data collection

Table 4. Summary of the Sampling week schedule

Time	Monday	Tuesday	Thursday
AM	Distribute BASE or 6-week questionnaire	Collect airborne fungi and bacteria	Collect airborne fungi and bacteria
PM		Collect airborne fungi and bacteria	Collect airborne fungi and bacteria
After hours			Dust sampling in chairs and floors Measure water activity Surface dust sampling

Data collected

Environmental measures

Table 5: 6-Week site and measurement summary University buildings

Event	1st	2nd	3rd	4th	5th	6th	7th	8th	9th	10th
	1997						1998
Date	5/5	6/23	8/4	9/15	10/27	12/8	1/19	3/2	4/13	5/25
# of Sites	20	20	20	20	21	19	15	14	15	15
Air organisms	20	20	20	20	21	19	15	14	15	15
Dust (floors)	20	20	20	20	21	19	15	14	15	15
Dust (Chairs)	20	20	20	20	21	19	15	14	15	15
Water Activity?	20	20	19	20	21	19	15	14	15	15
Surface Dust?	NA	NA	18	18	19	17	9	12	13	15

? Water Activity is the amount of water available in a substrate for microorganisms' growth.

Table 6: 6-Week site and measurement summary: State Office Building

Event	1st	2nd	3rd	4th	5th	6th	7th
1998							1999
Date	4/27	6/8	7/20	8/24	10/5	11/30	1/11
# Sites	8	8	8	8	7	7	7
Air organisms	8	8	8	8	7	7	7
Dust (floors)	8	8	8	8	7	7	7
Water Activity	8	8	8	8	7	7	7
Surface dust	8	8	8	8	7	7	7

All air and dust samples have been weighed and cultured for fungi and bacteria, and assayed for dust mite and cat allergens

Human studies

Table 7: Enrolled subjects and loss over time by building: University buildings

Event	1st	2nd	3rd	4th	5th	6th	7th	8th	9th	10th	Totals
1997							1998
Date	5/5	6/23	8/4	9/15	10/27	12/8	1/19	3/2	4/13	5/25
# of Sites	20	20	20	20	21	19	15	14	15	15
6-Week Quest.	82	70	56	51	56	51	45	41	40	38	530
Weekly Quest.	347	329	301	300	319	264	272	239	-	-	(2371)
Computer Tests	226	244	189	177	181	120	172	154	-	-	(1463)

Table 8: Enrolled subjects and loss over time: State office building

Event	1st	2nd	3rd	4th	5th	6th	7th
Date	4/27/98	6/8/98	7/20/98	8/24/98	10/5/98	11/30/98	1/11/99
# of Sites	8	8	8	8	7	7	7
6-Week Quest.	62	37	37	30	21	23	21
Weekly Quest.	301	211	190	166	134	118	138
Computer Tests	81	63	58	53	68	56	41

Table 9: Questionnaire and computer tests: overall totals

	University	State agency	Totals
6-wk questionnaires	498	231	729
Weekly questionnaires	2845	1258	4103
Computer tests	1756	420	2176

Table 10: Weekly participation rates: University buildings

	(initial: 105)	Weekly Questionnaires		Computer tests
Event	# recruits remaining	#	#/person	#	#/person
1	98	442	4.51	250	2.55
2	100	393	3.93	255	2.55
3	99	308	3.11	232	2.34
4	97	301	3.10	197	2.03
5	101	313	3.10	175	1.73
6	92	314	3.41	150	1.63
7	71	272	3.83	179	2.52
8	68	242	3.56	160	2.35
9	79	260	3.31	158	2.01

Table 11: Weekly participation rates: State Agency

	Initial:75	Weekly Questionnaires		Computer tests
Event	# recruits remaining	#	#/person	#	#/person
1(6w)	71	301	4.24	81	1.14
2(6w)	66	211	3.2	63	.95
3(5w)	62	190	3.06	58	.94
4(6w)	60	166	2.77	53	.88
5(8w)	46	134	2.91	68	1.48
6(6w)	43	118	2.74	56	1.3
7	40	138	3.45	41	1.03

Status of data entry
All data entry and checking is complete for the University building data. All data is being maintained as PC, mainframe, and portable disc files. Status of data on the State office building is as follows:

• Allergen data have been entered into the database.
• Questionnaire data have been entered into the database.
• Computer test data were downloaded and entered into the database.
• Environmental data on carbon dioxide, temperature and relative humidity were downloaded and entered into the database.

Additional data collection to be done

Analysis of environmental samples
Samples have not been analyzed for endotoxin or ergosterol

Data, entry, development of final data sets

• Environmental data on water activity were calculated but have not yet been entered.
• Environmental data on fungi and bacteria have not yet been entered.

Results
Ms. Chao and Ms. Lee are currently preparing the following manuscripts to satisfy their thesis requirements.

Environmental measurements

Manuscript 1: Longitudinal studies of airborne fungi in office buildings

Positive hole conversions (comparisons of Andersen data with and without conversions).

Characterizing fungal populations

Comparisons between recoveries on DG-18 and MEA

Principal Component analyses (4 principal components)

Summary Statistics

Overall distributions

Airborne fungi

Temperature

Relative Humidity

CO₂

Distributions over time

Airborne fungi

Temperature

Relative humidity

CO₂

Fungal concentration modeling (GAM models)

Total fungi

Principle components group 1

Principle components group 2

Draft conclusions

1. Positive hole conversions for Andersen data were not necessary in this data set.
2. MEA recovered slightly more total fungi, and more fungal taxa than DG-18. However, the differences were too small to be important, and data from the two culture media were combined for further analysis.
3. Overall, airborne fungal concentrations (CFU/m3 of air) for 10 sampling events were:
   
   Mean: 42, SD 70
   Median: 22
   Min/Max: 1/618
4. Four principal component factors were found explaining 53% of the variance
5. Airborne fungal concentrations varied seasonally with the highest median value in August and the lowest in January.
6. Distributions over time for PCA factors 1 and 2 differed.
7. Total airborne fungal concentrations were negatively correlated with CO₂ and positively related to relative humidity.
8. PCA factors 1 and 2 were negatively correlated with CO₂ and positively correlated with RH

Manuscript 2: Longitudinal studies of dustborne fungi in office buildings Plate loading and dilution effects

Characterizing populations

Comparisons between recoveries on DG-18 and MEA

Principle Component analyses

Summary statistics

Overall distributions for all environmental variables

Chair and floor dust fungi

Amount of chair and floor dust

Surface dust

CO₂

Temperature

Relative humidity

Water activity

Distributions Over time for these same variables

Fungal concentration modeling

Total floor fungi

Floor principal component group 1

Total chair fungi

Chair principal component group 1

Draft conclusions

Environmental factors predicting fungal concentrations in floor and chair dust

	Total fungi floor	PCA factor 1 floor	PCA factor 2 floor
Time	Yes	Yes	Yes
CO2	Yes	Yes	Yes
Temperature	Yes	Yes	No
Water activity	No	No	Yes
Grams of dust in chairs	No	No	No
	Total fungi chairs	PCA factor 1 chairs	PCA factor 2 chairs
Time	Yes	Yes	Yes
CO2	No	No	No
Temperature	No	No	No
Water activity	No	No	No
Grams of dust in floor cover	Yes	Yes	Yes

Manuscript 3: Dust as a source for airborne fungi in office buildings

Air-dust relationships for total fungi

Correlations between PCA factors for air and dust

Floors

Chairs

Air dust relationships for individual taxa

Effects of environmental factors on air dust relationships

Manuscript 4: Factors controlling allergen prevalence in office buildings (This is the first paper proposed for Ms. Lee's thesis).

Summary statistics

Overall distributions of environmental variables

Overall distributions of allergen concentrations

Floor dust

Chair dust

Distributions over time by allergen

Floor dust

Chair dust

Modeling of allergen concentrations

Human studies
These data are currently being analyzed. Proposed manuscript outlines are presented below. Because of the early stage of these analyses, these are provisional and are likely to change.

Manuscript 5: Environmental predictors for monthly symptom reporting in office workers

Distributions of Demographic Data

Symptom Prevalence over Time
Symptom groupings

Nonspecific/ Central Nervous System (CNS): (headache, fatigue, tension, pain in back, shoulders, or neck, difficulty remembering things or concentrating, dizziness, feeling depressed, and nausea).

Upper Respiratory/ Mucosal: (dry eyes, tired eyes, sore or dry throat, stuffy or runny nose, sneezing, and cough).

Lower Respiratory: Includes wheezing, chest tightness, shortness of breath, and cough.

Modeling symptom prevalence

Generalized estimating equations (GEE) and generalized linear mixed models (GLIMMIX) are used to correlate symptom presence (a binomial distribution) with psychosocial factors and environmental measurements. The autocorrelations resulting from repeated self-reported symptoms are accounted for by GEE/GLIMMIX modeling. Appropriate models are constructed for each symptom group.

(1) Nonspecific/ Central Nervous System (CNS)

(2) Upper Respiratory/ Mucosal

(3) Lower Respiratory

(4) Dry Skin

Manuscript 6. Environmental predictors for daily symptom reporting in office workers

Distributions of Demographic Data

Distributions for participants' gender, job categories, education and other demographic factors are listed.

Symptom Prevalence over Time

Daily symptoms were collected, including 4 groups:

1. Headache, unusual tiredness, fatigue, drowsiness, or difficulty remembering things or concentrating.

2. Shortness of breath, or wheezing.

3. Dry or itchy skin, or rash.

4. Dry, itching, or irritated eyes.

Symptom prevalence for each group over the sampling period is listed.

Distributions of Daily Environmental Variables

CO₂

Temperature

Relative humidity

Dustiness (surface, floor, chair)

Effects of Psychosocial Factors and Environmental Exposures on Daily Health Symptoms

GEE and GLMMIX are used to explore the correlations between daily symptoms and environmental and psychosocial factors. Models are obtained for each symptom group.

(1) Headache, unusual tiredness, fatigue, drowsiness, or difficulty remembering things or concentrating.

(2) Shortness of breath, or wheezing.

(3) Dry or itchy skin, or rash.

(4) Dry, itching, or irritated eyes.

Manuscript 7. Environmental factors predicting working efficiencies for occupants of large office buildings

Distributions of Demographic Data

A subset of the participants took the computer test. Demographic data for those who took the computer test regularly are shown including gender, age, job categories, education and etc.

Distributions of Number of Computer Tests over Time

Number of participants who took the computer test and number of computer tests taken every six weeks are displayed over the sampling year.

Correlations between Working Efficiencies and Environmental Exposures

Generalized additive models (GAM) in S-Plus are employed to model correlations between computer test scores and environmental variables. Serial correlation (autocorrelation) resulting from repeated measurements for same subject will be examined using partial autocorrelation functions in S-Plus and will be accounted for by including fixed subject effect in the models.

Manuscript 8: Allergens, symptoms, and working efficiency in office buildings

Manuscript 9: Relationships between cross-sectional and longitudinal data for large office buildings (Ms. Lee)
Outlines being formulated by Ms. Lee

Journal Articles on this Report : 3 Displayed | Download in RIS Format

Publications Views

Other project views:	All 3 publications	3 publications in selected types	All 3 journal articles

Publications

Type	Citation	Project	Document Sources
Journal Article	Chao HJ, Schwartz J, Milton DK, Burge HA. Populations and determinants of airborne fungi in large office buildings. Environmental Health Perspectives 2002;110(8):777-782.	R824797 (Final)	Full-text from PubMed Abstract from PubMed
Journal Article	Chao HJ, Milton DK, Schwartz J, Burge HA. Dustborne fungi in large office buildings. Mycopathologia 2002;154(2):93-106.	R824797 (Final)	Abstract from PubMed Abstract: Springer Abstract Exit
Journal Article	Chao HJ, Schwartz J, Milton DK, Burge HA. The work environment and workers' health in four large office buildings. Environmental Health Perspectives 2003;111(9):1242-1248.	R824797 (Final)	Full-text from PubMed Abstract from PubMed

Supplemental Keywords:

Health, Air, Scientific Discipline, Geographic Area, Health Risk Assessment, EPA Region, Risk Assessments, State, Biology, indoor air, Atmospheric Sciences, Biochemistry, air toxics, workplace, indoor air quality, hvac, longitudinal variation, building related illness, human exposure, MA, buildings, particulates, Region 1, ambient air, sick building syndrome, bioaerosols, human health risk, inhalation, Massachusetts

Progress and Final Reports:

Original Abstract

1996

1997