EPA/600/6-87/002a
June 1987
The Total Exposure Assessment
Methodology (TEAM) Study:
Summary and Analysis: Volume I
by
Lance A. Wallace
Project Officer
Environmental Monitoring Systems Division
Office of Acid Deposition,
Environmental Monitoring and Quality Assurance
Office of Research and Development
U.S. Environmental Protection Agency
Washington DC 20460
U.S. Environmental Protection Agency*
Region 5, Library (5PL-16)
230 S. Dearborn Sti-eet, Room 1670
Chicago, IL 60604
-------
Disclaimer
This document has been reviewed in accordance with U.S. Environmental
Protection Agency policy and approved for publication. Mention of trade
names or commercial products does not constitute endorsement or
recommendation for use.
-------
Contents
List of Figures v
List of Tables x
Acknowledgments xviii
1. Introduction 1
Pilot Study (Phase I) 1
Main Study (Phase II and III) 2
Special Studies 2
2. Recommendations 7
3. Summary and Conclusions 10
4. Overview 12
Selection of Target Chemicals 12
Study Design 14
Phase II: New Jersey, North Carolina, and North Dakota 14
Phase III California 15
Response Rates 16
Respondent Characteristics 16
Measurement Methods 18
Quality of the Data 18
Results 21
Quality Control/Quality Assurance 21
Percent Measurable 25
Concentrations 28
Indoor-Outdoor Comparisons 59
Correlations 59
Statistical Analysis of Questionnaire Data 64
Effects of Activities and Potential Sources on Exposures 73
Effects of Outdoor Concentrations on Exposures 90
Discussion 90
Comparison of New Jersey and California Results 90
Indoor versus Outdoor Air Concentrations 96
Sources of Exposure 104
Uncertainty of Estimates 105
Comparison of Weighted and Unweighted Frequency
Distributions 105
Health Effects 107
Standard Operating Procedures 107
TEAM Study Publications 108
Validity of TEAM Data 108
References 119
Appendix A: Sources of Exposure to Volatile Organic Chemicals: An
Analysis of Personal Exposures in the TEAM Study 122
-------
Contents
Page
Appendix B: Effect of Outdoor Air on Measures of Personal Exposure
in New Jersey and California 150
Appendix C: Analysis of Measurement Errors 1 60
Appendix D: Corrections to the Estimated Frequency Distributions
Due to Measurement Error 1 72
Appendix E: A Method for Comparing Weighted and Unweighted
Distributions on Probability Graph Paper, with Examples
from the TEAM Study 1 79
Appendix F: Personal versus Outdoor Air Comparisons by Seasons—
NJ 187
-------
List of Figures
No. Title Page
1. Personal monitor and vest 19
2. Schematic of breath sampling apparatus 20
3. Breath sampling system inside van with subject giving exhaled
air 20
4. Benzene: Estimated frequency distributions of personal air
exposures, outdoor air concentrations, and exhaled breath
values for the combined Elizabeth-Bayonne target population
(128,000) 32
5. Chloroform: Estimated frequency distributions of personal air
exposures, outdoor air concentrations, and exhaled breath
values for the combined Elizabeth-Bayonne target population
(128,000) 34
6. 1,1,1-Trichloroethane: Estimated frequency distributions of
personal air exposures, outdoor air concentrations, and exhaled
breath values for the combined Elizabeth-Bayonne target
population (128,000) 35
7. Tetrachloroethylene: Estimated frequency distributions of
personal air exposures, outdoor air concentrations, and exhaled
breath values for the combined Elizabeth-Bayonne target
population (128,000) 36
8. Trichloroethylene: Estimated frequency distributions of
personal air exposures, outdoor air concentrations, and exhaled
breath values for the combined Elizabeth-Bayonne target
population (1 28,000) 39
9. Carbon Tetrachloride: Estimated frequency distributions of
personal air exposures, outdoor air concentrations, and exhaled
breath values for the combined Elizabeth-Bayonne target
population (128,000) 40
10. m,p-Dichlorobenzene: Estimated frequency distributions of
personal air exposures, outdoor air concentrations, and exhaled
breath values for the combined Elizabeth-Bayonne target
population (128,000) 41
11. Styrene: Estimated frequency distributions of personal air
exposures, outdoor air concentrations, and exhaled breath values
for the combined Elizabeth-Bayonne target population
(128,000) 42
12. Ethylbenzene: Estimated frequency distributions of personal air
exposures, outdoor air concentrations, and exhaled breath values
for the combined Elizabeth-Bayonne target population
(128,000) 43
-------
List of Figures (Continued)
No. Title Page
13. m,p-Xylene: Estimated frequency distributions of personal air
exposures, outdoor air concentrations, and exhaled breath
values for the combined Elizabeth-Bayonne target population
(128,000) 44
14. o-Xylene: Estimated frequency distributions of personal air
exposures, outdoor air concentrations, and exhaled breath
values for the combined Elizabeth-Bayonne target population
(128,000) 45
15. Estimated arithmetic means of 11 toxic compounds in daytime
(6:00 am to 6:00 pm) air samples for the target population of
Elizabeth and Bayonne, New Jersey, between September and
November 1981 46
16. Estimated arithmetic means of 11 toxic compounds in overnight
(6:00 pm to 6:00 am) air samples for the target population of
Elizabeth and Bayonne, New Jersey, between September and
November 1981 46
17. Estimated geometric means of 11 toxic compounds in daytime
(6:00 am to 6:00 pm) air samples for the target population of
Elizabeth and Bayonne, New Jersey, between September and
November 1981 47
18. Estimated geometric means of 11 toxic compounds in overnight
(6:00 pm to 6:00 am) air samples for the target population of
Elizabeth and Bayonne, New Jersey, between September and
November 1981 47
19. Weighted frequency distributions for 24-hour exposures of
355 NJ residents to aromatic compounds (Fall 1981) 48
20. Weighted frequency distributions of 24-hour exposures of
355 NJ residents to six chlorinated compounds (Fall 1981) 49
21. Weighted frequency distributions of day and night 12-hour
personal air exposures compared to the 48-hour average for
160 NJ residents (Fall-Summer 1981-82) 50
22. Ratios of median 12-hour indoor air concentrations to
simultaneous 12-hour outdoor air concentrations for New
Jersey homes (N=85 in Fall 1981; N=70 in summer 1981;
and N=10 in Winter 1983) 51
23. Ratios of 90th-percentile 12-hour indoor air concentrations to
simultaneous outdoor air concentrations in New Jersey homes .. .51
24. Weighted cumulative frequency distributions of overnight
personal air exposures and outdoor air concentrations of
/n,p-dichlorobenzene isomers in New Jersey 52
25. 1,1,1-Trichloroethane: Estimated frequency distributions of
personal air exposures, outdoor air concentrations, and exhaled
breath values for the target population of 360,000 persons in
the South Bay section of Los Angeles (Feb. 1984) 53
-------
List of Figures (Continued)
No. Title Page
26. p-Dichlorobenzene: Estimated frequency distributions of
personal air exposures, outdoor air concentrations, and exhaled
breath values for the target population of 360,000 persons in
the South Bay section of Los Angeles (Feb. 1984) 57
27. Benzene: Estimated frequency distributions of personal air
exposures, outdoor air concentrations, nd exhaled breath values
for the target population of 360,000 persons in the South Bay
section of Los Angeles (Feb. 1984) 58
28. Tetrachloroethylene: Estimated frequency distributions of
personal air exposures, outdoor air concentrations, and exhaled
breath values for the target population of 360,000 persons
in the South Bay section of Los Angeles (Feb. 1984) 59
29. Octane, Decane, Undecane, and Dodecane: Estimated frequency
distributions of overnight concentrations in participants'
homes compared to overnight outdoor air concentrations
(Feb. 1984) 60
30. Benzene: Estimated frequency distributions of personal air
exposures, outdoor air concentrations, and breath values for
the target population of 330,000 residents of the South Bay
section of Los Angeles (May 1984) 60
31. p-Dichlorobenzene: Estimated frequency distributions of
personal air exposures, outdoor air concentrations, and breath
values for the target population of 330,000 residents of the
South Bay section of Los Angeles (May 1984) 63
32. Benzene: Estimated frequency distributions of personal air
exposures, outdoor air concentrations, and breath values for
91,000 residents of Antioch and Pittsburg, CA 64
33. p-Dichlorobenzene: Estimated frequency distributions of
personal air exposures, outdoor air concentrations, and breath
values for 91,000 residents of Antioch and Pittsburg, CA 65
34. Octane, Decane, Undecane, and Dodecane: Estimated frequency
distribution of overnight concentrations in participants' homes
compared to overnight outdoor air concentrations (June 1984)... .65
35. Unweighted cumulative frequency distributions of benzene
concentrations in the breath of current smokers vs. nonsmokers
(NJ. Fall 1981) 76
36. Unweighted cumulative frequency distributions of benzene
concentrations in the air in homes with at least one smoker
vs. homes with no smokers (NJ, Fall 1981) 77
37. Median breath concentrations of 21 chemical plant workers
vs. 330 other participants (NJ, Fall 1981) 78
38. Median breath values for 28 paint workers vs. 320 other
participants (NJ, Fall 1981) 78
-------
List of Figures (Continued)
No. Title Page
39. Median breath values for 11 plastics manufacturing workers
vs. 340 other participants (NJ, Fall 1981) 79
40. Median breath values for 19 petroleum plant workers vs. 330
other participants (NJ, Fall 1981) 79
41. Median breath values for 9 printing plant workers vs. 340
other participants (NJ, Fall 1981) 80
42. Median breath values for 11 persons visiting dry cleaning shops
on the day they were sampled vs. 340 other participants (NJ,
Fall 1981) 80
43. Median breath values for 67 persons visiting a service station
the day they were sampled vs. 270 other participants (NJ,
Fall 1981) 81
44. Median breath values for 62 persons exposed to automobile
or truck exhaust on the day they were sampled vs. ^280 other
participants (NJ, Fall 1981) 81
45. Median breath concentrations of 150 smokers compared to
150nonsmokers(NJ, Fall 1981) 82
46. Median breath concentrations for 20 persons using pesticides
vs. 330 other participants (NJ, Fall 1981) 82
47. Comparison of unweighted 75th percentile concentrations of
11 prevalent chemicals in overnight outdoor and personal air
in NJ (Fall 1981) with outdoor air measured in a number of
U.S. cities between 1970-1980 95
48. Comparison of unweighted 99th percentile concentrations of
11 prevalent chemicals in overnight outdoor air and overnight
personal air in NJ (Fall 1981) 95
49. Weighted vs. unweighted frequency distributions for 1,1,1-
trichloroethane 106
50. Comparison of outdoor air measurements in Los Angeles by
the California Air Resources Board (CARB) and the TEAM
Study 115
51. Comparison of median outdoor air concentrations 116
D-1. Cumulative frequency distribution of geometric means of 62
pairs of duplicate measurements of overnight personal air
exposures to 1,1,1-trichloroethane: (NJ, Fall 1981) 173
D-2. Cumulative frequency distribution of geometric means of 62
pairs of duplicate measurements of overnight personal air
exposures to benzene: (NJ, Fall 1981) 174
D-3. Cumulative frequency distribution of measurement errors
(defined as the ratio of one measurement to the geometric mean
of the pair) for 62 pairs of duplicate overnight personal air
samples: 1,1,1-trichloroethane (NJ, Fall 1981) 175
-------
List of Figures (Continued)
No. Title Page
D-4. Cumulative frequency distribution of measurement errors for
62 pairs of duplicate overnight personal air samples: benzene
(NJ, Fall 19811) 175
D-5. Effect of correction factor of 0.93 (Table D-1) on observed
cumulative frequency distribution of ovenight personal air
exposures to tetrachloroethylene for 350 residents of
Bayonne-Elizabeth, NJ 176
E-1. Weighted vs. unweighted frequency distributions for
m,p-dichlorobenzene 181
E-2. Weighted vs. unweighted frequency distributions for
1,1,1-trichloroethane 182
E-3. Weighted vs. unweighted frequency distributions for benzene ... 183
E-4. Weighted vs. unweighted frequency distributions for styrene .... 184
E-5. Weighted vs. unweighted frequency distributions for
tetrachloroethylene 185
F-1. 24-hour personal exposures to 1,1,1-trichloroethane compared
to outdoor air in New Jersey—first three seasons 188
F-2. 24-hour personal exposures to tetrachloroethylene compared
to outdoor air in New Jersey—first three seasons 189
F-3. 24-hour personal exposures to styrene compared to outdoor
air in New Jersey—first three seasons 190
F-4. 24-hour personal exposures to chloroform compared to outdoor
air in New Jersey—first three seasons 191
F-5. 24-hour personal exposures to benzene compared to outdoor
air in New Jersey—fall season 192
-------
List of Tables
No. Title Page
1. Summary of TEAM Studies 4
2. Target Compounds Selected for Monitoring in Environmental
Media: New Jersey 13
3. Target Compounds Selected for Monitoring in Environmental
Media: California 14
4. Sites Visited in the Main TEAM Study 16
5. Response Rates: All TEAM Sites 17
6. Respondent Characteristics 17
7. Samples Collected at All TEAM Sites 21
8. Blank Values and Recovery Efficiencies for Air and Breath
Samples: New Jersey 22
9. Ranges of Mean Recoveries and Backgrounds for Field
Controls and Blanks—Air and Breath Samples: California 23
10. Recoveries and Backgrounds for Field Controls and Blanks—
Water Samples 24
11. Coefficients of Variation (%} for Duplicate Air and Breath
Samples in New Jersey—Season I 24
12. Duplicate Air and Breath Samples—Median Relative Standard
Deviations (%): California 25
13. Duplicate Water Samples—Median Relative Standard
Deviations (%) 26
14. Comparison of TEAM and CARB Co-located Sampling Results
(in ppb) 26
15. Target Compounds Sorted by Percent Measurable in Breath
and Air Samples—All Three Seasons: New Jersey 27
16. Target Compounds Sorted by Percent Measurable in Water
Samples—NJ—All Three Seasons 28
17. Target Compounds Sorted by Percent Measurable in Air and
Breath Samples—NC and ND 29
18. Target Compounds Sorted by Percent Measurable in Water
Samples—NC and ND 30
19. Target Compounds Sorted by Percent Measurable in Air and
Breath Samples 31
20. Target Compounds Sorted by Weighted Percent Measurable in
Drinking Water Samples 32
-------
List of Tables (Continued)
No. Title Page
21. Weighted Estimates of Air and Breath Concentrations of 11
Prevalent Compounds for 130,000 Elizabeth-Bayonne Residents
(Fall 1981); 110,000 Residents (Summer 1982); and 49,000
Residents (Winter 1983) 33
22. Weighted Arithmetic Mean Overnight Personal Exposures
(Indoor Air) Compared to Outdoor Air Concentrations
(/jg/m3): New Jersey, All Three Seasons 37
23. Maximum Concentrations (/jg/m3) of Organic Compounds in
Air and Breath of 350 NJ Residents 38
24. Arithmetic Means and Maxima (fjg/L) of Organic Compounds
in New Jersey Drinking Water 38
25. Indoor/Outdoor Ratios in Greensboro, NC 54
26. Indoor/Outdoor Ratios in Devils Lake, ND 54
27. Weighted Estimates of Air and Breath Concentrations of
Nineteen Prevalent Compounds for 360,000 Los Angeles
Residents (February 1984), 330,000 Los Angeles Residents
(May 1984), and 91,000 Contra Costa Residents (June 1984) 55
28. Estimates of Drinking Water Concentrations for California
Residents 58
29. Indoor/Outdoor Comparisons for Matched Samples: Median
Overnight Concentrations (fig/m3): California 61
30. Spearman Correlations Between Breath Concentrations and
Preceding Daytime 12-Hour Personal Exposures to Eleven
Compounds in New Jersey, North Carolina, and North Dakota 66
31. Spearman Correlations Between Breath and Preceding Air
Concentrations (Measurable Amounts Only): California 67
32. Spearman Correlations > 0.5 Between Prevalent Compounds
in Air and Breath: TEAM Study, New Jersey, Fall 1981;
Summer 1982; Winter 1983 68
33. Variables Included in Statistical Analysis 70
34. Questionnaire Items Associated with Significantly
Increased Exposures (p < 0.001): New Jersey, Fall 1981 71
35. Variables Ranked by Number of Pairwise Associations with
Significantly Increased or Decreased Exposures (p <0.05)
(All New Jersey and California Visits) 74
36. Breath Concentrations (fjg/m3) of Selected Chemicals: Smokers
vs. Non-smokers: Unweighted Geometric Means 75
37. Overnight Indoor Air Concentrations (/ug/m3) in Homes With
and Without Smokers: Weighted Geometric Means 75
38. Median Concentrations (/jg/m3) of Chemicals Significantly
(p < .05) Higher in Breath of Persons Exposed to Potential
Sources During the Day (Week) They Were Monitored 83
-------
List of Tables (Continued)
No. Title Page
39. Chemicals Showing Significantly (p < .05) Higher
Concentrations in Air and Breath of Persons Recently Exposed
to Potential Sources Compared to Persons Not Exposed to
That Source 85
40. Chemicals with Significantly (p < .05) Higher Concentrations
in Air and Breath of Persons Recently Exposed to Potential
Sources Compared to Persons Not Exposed to Any Source 87
41. Effects of Smoking on Breath Concentrations of Benzene and
Other Hydrocarbons 89
42. Effect of Outdoor Air Concentrations on Measures of Personal
Exposure (NJ): Coefficients of Stepwise Regressions 91
43. Effect of Overnight Outdoor Air Concentrations on Indoor Air
Concentrations (CA): Coefficients of Stepwise Regressions 92
44. Control and Blank Data for Tenax Cartridges Used in New
Jersey and California: TEAM Study 93
45. Median Coefficients of Variation (%) for Duplicate Personal Air
Samples in New Jersey and California: TEAM Study 94
46. Volatile Organic Compounds in Overnight Indoor Air m New
Jersey and California: TEAM Study, 1981-84 97
47. Volatile Organic Compounds in Overnight Outdoor Air in New
Jersey and California: TEAM Study, 1981-84 98
48. Maximum Overnight Concentrations Indoors and Outdoors for
Homes with Outdoor Monitors: TEAM Study, 1 981 -84 99
49. Median Indoor-Outdoor Differences (//g/m3) for Persons Who
Did Not Leave Their Homes During the 12-Hour Monintoring
Period 101
50. Mean Indoor-Outdoor Differences (/ug/m3) for Persons Who Did
Not Leave Their Homes during the 12-Hour Monitoring
Period 102
51. Approved SOPs for Phase III TEAM Study 109
52. TEAM Study Publications 111
53. Comparison of Outdoor Measurements of Toxics by TEAM
Study and by California Air Resources Board 114
A-1. Stepwise Regression Results: Daytime Personal Air—New
Jersey, Fall 1981 123
A-2. Stepwise Regression Results: Overnight Personal Air—New
Jersey, Fall 1981 126
A-3. Stepwise Regression Results: Breath (Daily Exposures)—
New Jersey, Fall 1981 129
A-4. Stepwise Regression Results: Breath—New Jersey, Summer
1982 132
-------
List of Tables (Continued)
No. Title Page
A-5. Stepwise Regression Results: Overnight Personal Air—New
Jersey, Summer 1982 133
A-6. Stepwise Regression Results: Daytime Personal Air—New
Jersey, Summer 1982 135
A-7. Stepwise Regression Results: Breath—New Jersey (Winter
1983) 137
A-8. Stepwise Regression Results: Overnight Personal Air—
New Jersey (Feb. 1983) 138
A-9. Stepwise Regression Results: Daytime Personal Air—
New Jersey (Feb. 1983) 138
A-10. Stepwise Regression Results: Breath—Los Angeles
(Feb. 1984) 140
A-11. Stepwise Regression Results: Overnight Personal Air—Los
Angeles (Feb. 1984) 141
A-1 2. Stepwise Regression Results: Daytime Personal Air—Los
Angeles (Feb. 1984) -. 142
A-13. Stepwise Regression Results: Breath—Los Angeles
(May 1984) 144
A-14. Stepwise Regression Results: Overnight Personal Air—Los
Angeles (May 1984) 145
A-15. Stepwise Regression Results: Daytime Personal Air—
Los Angeles (May 1984) 146
A-16. Stepwise Regression Results: Breath—Contra Costa,
(June 1984) 147
A-17. Stepwise Regression Results: Overnight Personal Air—
Contra Costa (June 1 984) 148
A-18. Stepwise Regression Results: Daytime Personal Air-
Contra Costa (June 1984) 149
B-1. Stepwise Regression Results for 87 New Jersey Homes
with Outdoor Measurements: Overnight Personal Air—
Fall 1981 151
B-2. Stepwise Regression Results for 87 New Jersey Homes
with Outdoor Measurements: Daytime Personal Air—
Fall 1981 152
B-3. Stepwise Regression Results for 87 New Jersey Homes
with Outdoor Measurements: Breath—Fall 1981 153
B-4. Stepwise Regression Results for 71 New Jersey Homes
with Outdoor Measurements: Overnight Personal Air—
Summer 1982 154
B-5. Stepwise Regression Results for 71 New Jersey Homes
with Outdoor Measurements: Daytime Personal Air—
Summer 1982 155
-------
List of Tables (Continued)
No. Title Page
B-6. Stepwise Regression Results for 71 New Jersey Homes
with Outdoor Measurements: Breath—Summer 1982 156
B-7. Stepwise Regression Results for 24 Homes with Outdoor
Measurements: Overnight Personal Air—Los Angeles
(February 1984) 157
B-8. Stepwise Regression Results for 24 Homes with Outdoor
Measurements: Overnight Personal Air—Los Angeles
(May 1984) 158
B-9. Stepwise Regression Results for 10 Homes with Outdoor
Measurements: Overnight Personal Air—Contra Costa
(June 1984) 159
C-1. Coefficients of Variation (%) of Relative Response Factors
and Recovery Efficiencies: California 162
C-2. Estimated Magnitude of Errors Associated with Air
Measurements 164
C-3. Coefficients of Variation (%) of Measurement Errors:
TEAM—California Study 166
C-4. Comparison of Total Variance with "Chemical-Specific"
Component of All Measurable Duplicate Samples: NJ,
Fall 1981 167
C-5. Comparison of Total Variance with "Chemical-Specific"
Component of All Measurable Duplicate Samples: NJ,
Summer 1982 168
C-6. Comparison of Total Variance with "Chemical-Specific"
Component of All Measurable Duplicate Samples: NJ,
Winter 1983 169
C-7. Comparison of Total Variance with "Chemical-Specific"
Component of All Measurable Duplicate Samples: Los
Angeles, Winter 1984 170
C-8. Comparison of Total Variance with "Chemical-Specific"
Component of All Measurable Duplicate Samples: Contra
Costa, June 1984 171
D-1. Correction Factors Due to Measurement Errors—
Fall 1981 176
D-2. Correction Factors Due to Measurement Errors—
Summer 1982 177
D-3. Correction Factors for Estimated Frequency Distributions
Based on Measurement Errors—Winter 1983 178
E-1. Weighted and Unweighted Overnight Personal Exposures
(Indoor Air Concentrations) and Geometric Standard
Deviations Calculated for Selected Percentiles 186
-------
Acknowledgments
The TEAM Study was conceived in 1979. The Assistant Administrator of
EPA's Office of Research and Development, Dr. Steven Gage, overcame the
many obstacles in the way of this new approach to measuring human
exposure. He obtained the necessary resources and delegated the necessary
authority to carry out the project. Chuck Brunot, Director of the Environmental
Monitoring Division, ORD, was an unfailing source of support throughout
the entire six years of this study.
Local and state officials in New Jersey, North Carolina, North Dakota, and
California gave essential support to this study. In New Jersey, special efforts
were made by Dr. John Sakowski and Mr. David Roach of the Bayonne
Department of Health, Mr. John Surmay and Mr. Robert Travisano of the
Elizabeth Health, Welfare and Housing Department, Dr. Thomas Burke of
the New Jersey Department of Environmental Protection, and employees
of EPA's Region 2. In North Carolina, the Guilford County Health Department
prepared the way for community involvement. In North Dakota, the Ramsey
County Department of Health gave essential advice (including avoiding the
start of duck hunting season, which would have caused our response rate
to plummet). Finally, in California, the South Coast Air Quality Management
District of the California Air Resources Board (CARB) provided useful
information on emissions and the best seasons to sample in the Los Angeles
area. The Bay Area Air Quality Management District of CARB did the same
for the Contra Costa area. EPA's Region 9 provided helpful press relations.
Dr. Robert Ziegenfus of Kutztown University of Pennsylvania provided the
nonparametric statistical tests and corresponding computer graphics. The
author prepared this summary while on appointment as a Visiting Scientist
to the Harvard University School of Public Health, and would like to thank
Prof. J. D. Spengler for providing the opportunity to reflect on these matters
in a stimulating atmosphere. Many discussions with Dr. Wayne Ott and Dr.
David T. Mage of EPA helped clarify the author's thinking. Ann Desmond
of the Harvard staff prepared the manuscript through many revisions with
great efficiency and good humor.
We are most indebted to the hundreds of citizens in four states who
conscientiously wore monitors, kept diaries, and answered questions about
their activities.
We would like to thank the reviewers who read over the many pages
of materials and made helpful comments:
Gerald Akland David T. Mage
Joseph Behar Milton Russell
Patricia Buffler Lars Mtflhave
Howard Crist John A. Moore
Larry T. Cupitt Samuel C. Morris III
Joan Daisey Demetrios Moschandreas
Stan Dawson William C. Nelson
-------
Charles Delos Maurice Owens
Robert Fegley Karen Pederson
Ben G. Ferris, Jr. Thomas Purcell
Richard B. Gammage Everett S. Plyler
Edwin Johnson James Repace
Patrick Kennedy R. c. Rhodes
Donna Kuroda Vivian Thomson
Paul Lioy Philip Walsh
Jack McGinnity Elizabeth K. Weisburger
The entire project was under the superb technical direction of Dr. Edo
Pellizzari of Research Triangle Institute, Principal Investigator, and his
coworkers:
T. Hartwell C. Sparacino
C. Leininger R. Whitmore
K. Perritt H. Zelon
L. Sheldon R. Zweidinger
We would also like to thank the following chemists, statisticians, and other
technical workers at Research Triangle Institute whose work was essential
to this project:
P. Blau J. Keever
J. Bursey T. Pack
N. Castillo R. Porch
S. Cooer D. Smith
L. Dang K. Thomas
P. Elkins D. Whitaker
S. Frazier F. Williams
We thank also Dr. Jack O'Neill and Dr. Sid Gordon of NT Research Insti-
tute for providing quality assurance.
-------
Section 1.
Introduction
The TEAM Study was planned in 1979 and completed in 1985. The goals
of this study were: (1) to develop methods to measure individual total exposure
(exposure through air, food, and water) and resulting body burden of toxic
and carcinogenic chemicals, and (2) to apply these methods within a
probability-based sampling framework to estimate the exposures and body
burdens of urban populations in several U.S. cities. To achieve these goals,
the following approach was adopted:
1. A small personal sampler was developed to measure personal exposure
to airborne toxic chemicals;
2. A specially-designed spirometer was developed to measure the same
chemicals in exhaled breath; and
3. A survey design involving a three-stage stratified probability selection
approach was adopted to insure inclusion of potentially highly exposed
groups.
Pilot Study (Phase I)
A pilot study was conducted between July and December 1980 to test 30
sampling and analytical protocols for four groups of chemicals potentially
present in air, water, food, house dust, blood, breath, urine, and human
hair.
The four groups of chemicals were:
1. Volatile organics (1 5 target chemicals including benzene, vinyl chloride,
chloroform, and tetrachloroethylene)
2. Semivolatile organics (8 target pesticides and PCBs)
3. Metals (lead, cadmium, arsenic)
4. Polyaromatic hydrocarbons (6 compounds including benzo-a-pyrene)
In this pilot study, nine subjects from New Jersey and three from North
Carolina collected environmental and biological samples for several days
on three separate visits over the six-month period. They also filled out a
series of household questionnaires and activity recall questionnaires that
had been approved by the Office of Management and Budget (OMB).
The results of the pilot study (1,2) indicated that the TEAM goals could
be met at present for only one group of compounds: the volatile organics.
Adequate methods existed to determine their concentrations in personal air,
ambient air, exhaled breath, and drinking water. They were not present in
food (with the exception of chloroform in beverages), so that food could safely
be ignored.
Each of the other three groups of chemicals had measurement method
problems. Both metals and pesticides have a major route of exposure in
-------
solid foods—yet the sampling and analytical protocols for measuring
individual meals do not exist. For the PAHs, no personal air monitor capable
of collecting sufficient amounts to analyze existed.
Thus, it was decided to concentrate the main TEAM Study on the volatile
organics. This group of some hundreds of compounds includes a dozen or
so known or suspected human carcinogens, including many organics
contained in the list of 37 potential Hazardous Air Pollutants that EPA's
Office of Air and Radiation must decide whether to regulate or not; several
solvents of interest to the Office of Toxic Substances; and compounds that
the Office of Drinking Water will soon regulate.
Main Study (Phases II and III)
The main TEAM Study measured the personal exposures of 600 people
to a number of toxic or carcinogenic chemicals in air and drinking water.
A total of 20 target chemicals were selected on the basis of their toxicity,
carcinogenicity, mutagenicity, production volume, presence in preliminary
sampling and pilot studies, and amenability to collection on Tenax. The
subjects were selected to represent a total population of 700,000 residents
of cities in New Jersey, North Carolina, North Dakota, and California. Each
participant carried a personal air sampler throughout a normal 24-hour day,
collecting a 12-hour daytime sample and a 12-hour overnight sample.
Identical samplers were set up near some participants' homes to measure
the ambient air. Each participant also collected two drinking water samples.
At the end of the 24 hours, each participant contributed a sample of exhaled
breath. All air, water, and breath samples were analyzed for 20 target
chemicals.
Phase II of the TEAM Study was conducted during three seasons (summer,
fall, and winter) in New Jersey and also in two comparison areas in North
Carolina and North Dakota. Phase III was conducted in two target areas
in California—an area in southwest Los Angeles County and the communities
of Antioch, Pittsburg, and West Pittsburg, northeast of Oakland. The Phase
II questionnaires were revised and received OMB approval. Also, nine
chemicals were added to, and three dropped from, the list of target
compounds. Otherwise, the Phase III study used the same general procedures
as the Phase II study. The Los Angeles area was monitored during two seasons
(winter and spring) and the Antioch/Pittsburg area was monitored during
the spring season.
Special Studies
A series of special studies were undertaken as part of the TEAM Study.
They include:
1. Nursing Mothers Study. Air, water, breath, blood, urine, and mothers'
milk samples were collected from 17 nursing mothers in Bayonne and
Elizabeth, NJ to determine whether the target chemicals were
accumulating in mothers' milk and the relationships between exposure
and body burden. Several target chemicals were more highly
concentrated in mothers' milk; therefore, it may be an important
contributor to babies' exposure.
2. Dry Cleaners Study. Eight employees in three dry cleaners collected
personal, workplace, ambient, and home air samples on one work day
and one weekend day to investigate their exposures to tetrachloroeth-
-------
ylene, 1,1,1-trichloroethane, and aromatic solvents. Exposures and
breath levels ranged up to 1000 times typical nonoccupational levels.
3. Swimming Pool Study. Lifeguards at three pools were investigated
for possible elevated chloroform exposures.
4. Indoor Air Study. Four public buildings were investigated to determine
levels of volatile organics, pesticides, PCBs, respirable particulates,
metals, and formaldehyde in indoor air. Several hundred VOCs were
identified, including about two dozen mutagens and carcinogens.
All TEAM studies are summarized in Table 1.
In this four-volume Final Report, Volume I is an overview of the TEAM
Study. Volume II deals with Phase II (NJ, NC, ND) and Volume III with Phase
III (CA). Volume IV is a compilation of Standard Operating Procedures
developed for the TEAM Study and applicable to similar studies of human
exposure to volatile organic compounds.
-------
CO
•S
•5
a
Table 1. Summary of TEAM St
(6
1
s*;
£
References
•Q
co
s»
o
CD
fc
Name and Description of Study
*-. CM
Wallace 1982a
Zweidinger, 1982
00 0
cr> co
«- 05
.£. *~
P Cb
1 §
5-^
1. Lamar University - UNC Study
.g
V>
c
€-g
£•§ «
.SS|
S a, %
1 ^
:3£
^ Qi S
TB-S S
§ = §
|U
111
S^s
*- J-- to
•i a Q)
5 CO 3
£J!
fig«
>. S CD
1 | 8
•5 S -Q
i-s§
"> 5 $
V ^ %
th O •£
|« Q)
s to -Q
1 S s
*-N
^ S-5
III
5^ 8
6.8 to
$ $ to
^ -^ 5
. CO §
to g ^
•5 to "o
"^-tf
•g^a
fe c o
c5 c
> , 0)
P to K.
^•s s
^ Q S
Eleven college students at Lamar
spirometer for collecting breath sam{
exposure (2-3 orders of magnitude)
chemicals.
•St- K OJ
co 10 to cd o
982
1982a,
982bf
984a
Pellizzari, 1980,
Entz, 1982
Sparacino,
Wallace,
I
1
i.
0)
c (
"
protocols pr
Q)
to
to
§
I
00
5
§
-------
(0
1
><
•S
Table 1. (continued)
Name and Description of
a
o
-*
CO
Q
-C
t
1
^^
&
^2
^^
'>• <2
^ k.
s *
oro, North Carolina; and 25 in L
and also in a number of journal
25 in Greensb
II of this report
>w Jersey;
'n Volume >
£-5
*l
•*-. ^
Q) o
-Q V)
S -£
,N Q
Lu ,<0
"^
T3 ^
C cj
CO 'S
㥤
I ^
350 persons in Bayon
participated in this study.
CO
CN
to OD
CM CM
Pellizzari, 1985b,c
Wallace, 1985
S
CJ)
*^
QJ
|
i
^
CD
CO
£
i
V)
^
§
st
s~.
O
^
CD
c\
i
,c
-Q
CD
^
<
CJ
CO
•8
.CO
•i
1
Si
T3
a
CO
CO
i=
.c
T3
£
CO
•S-
.0
C:
California par,
' Pittsburg,
0
C
CO
CJ
_Q
•C
C
•q;
I
S,
w)
c
^
o a
| SjS^.^ § |
^!|£je
Is ° ^ § "5 ^
•S J2 P S a 8
5 15 c ° to o -c
^ c |« ?! c
-Q -~ > S .c § 5
•> «*« gfc^S
8 .y -o j; .0 J2 *•
o c c 1; -Q .co jo
u co .. c
co g) 0 -e — £ .9
.0 o 5j co £ g %
~S CD co S? -9- E '?
•C3 -S •< S C p
"> ^ O) iw r- ^
O +_, -^ wj >j £. -^
Q. (0 t m n 0
Four commercial and
of indoor air levels of vol
newly finished, one mon
1,1,1-trichloroethane, wen
increased on the last twt.
chamber study of comm,
target toxic chemicals in
o>
2
^ " CD 0 §
S> § S fe .fe -
The TEAM methodok
tionships between air, bl
1,1,1-trichloroethane, the
hydrocarbon mixture)) w,
of 20-25 mg/m3 tetrachi
indicated by the fact tha
-------
85
•w r»
O <
<5 q>
>, to to ^
lUi
***,!
S 85 o
e'S . a
v> S3 g-
S-S ^^S 8
°-&:5 S *~
^ 5 ^ 5 -S
5 * * ^-S
Mll.8
"*• flJ ^ -5 J5
o^uj^ 5
-a r: -o „• ^
4) ^3 C 0) is
« § ""t 8
•« -9
-------
Section 2.
Recommendations
The major finding of this study is the observation that personal and indoor
exposures to these toxic and carcinogenic chemicals are nearly always
greater—often much greater—than outdoor concentrations. We are led to
the conclusion that indoor air in the home and at work far outweighs outdoor
air as a route of exposure to these chemicals.
Since federal and state environmental regulators and directors of research
have until now focused most of their attention on sources affecting outdoor
concentrations, it is important to verify this finding and, if true, incorporate
it into future research and regulatory strategies.
An appropriate next step would be to investigate the sources of these
exposures more systematically than was possible in the TEAM Study. The
relative contribution of building materials, furnishings, personal activities,
and consumer products to personal exposures should be determined by
intensive studies in a number of homes, office buildings, schools, and other
structures where people spend much of their time. In particular, the following
specific recommendations are made:
1. Extend studies of human exposure to other cities and rural areas. The
studies in Greensboro, NC and Devils Lake, ND were too small to provide
much stability to their estimates of human exposure. Thus, additional
studies of medium-sized cities and rural areas are needed. Also, the
larger studies in Elizabeth, Bayonne, Los Angeles, Antioch and Pittsburg
all took place in areas of intensive chemical manufacturing and
petroleum refining. Future studies should include large cities without
such sources to determine the applicability of TEAM findings to the
types of locations in which most people in the U.S. live.
2. Follow up previous studies to determine the reasons for elevated
exposures. By using the persons (or homes) already measured, high-
exposure persons (homes) that represent known numbers of other
persons (homes) can be selected without an expensive screening
process.
3. Perform special studies to determine the strength of hypothesized
sources. These may include experimental studies in occupied houses
or emission studies in chambers.
4. Develop emission inventories of major sources of indoor and personal
exposure. These should emphasize consumer products, building
materials, and personal activities such as smoking, filling gas tanks,
showering, visiting dry cleaners, etc.
5. Develop models capable of combining emissions from indoor sources,
personal activity patterns, outdoor concentrations, and air exchange
rates to predict exposures for large populations.
The second major finding has been the great utility of breath sampling
to estimate levels in the body due to normal daily exposure to toxic chemicals.
-------
Breath sampling is noninvasive and is much more sensitive and less costly
and difficult than blood sampling. In this study, breath sampling alone was
effective in distinguishing between populations exposed to specific sources
and those not so exposed. The technique should be investigated for possible
use in the following situations:
6. Estimate dosages of persons exposed to chemical spills or releases.
7. Survey healthy persons to establish normal baselines and ranges of
biological variability
8. Study diseased persons to establish possible early diagnostic
procedures.
9. Study acute health effects associated with organic emissions ("sick
building syndrome") to determine the extent of the loss of productivity
of U.S. workers due to degraded indoor air quality in the workplace.
A third finding has been the demonstration of the utility of this personal
monitoring approach not only in estimating the exposure of entire urban
area populations, but also in gaining an understanding of the sources of
exposure. The general methodology appears applicable for determining
exposures to many other pollutants (e.g., pesticides and metals) provided
adequate sampling and analysis protocols for individual meals can be
developed. With the development of better instruments, it should also be
possible to carry out large-scale studies of exposure to inhalable particulates
and NOa in the near future.
Control of Toxic Emissions
Reduction of exposure to the toxic chemicals measured in the TEAM Study
may come about through two types of action: individual and organizational.
Individual Actions. Several of the sources identified in the TEAM Study
may be dealt with by simple means. For example, unused paint cans, aerosol
sprays, cleansers, solvents, etc. may be disposed of or stored in a detached
garage or tool shed. Charcoal filters attached to the kitchen and bathroom
taps can remove chloroform and other trihalomethanes from water supplies.
(However, some filters are relatively ineffective: an EPA study and a Consumer
Reports article have identified effective and ineffective brands.) Discontinuing
use of room air fresheners or switching to brands that do not contain p-
dichlorobenzene will reduce exposure to that chemical. Discontinuing
smoking, smoking only outdoors or in well-ventilated rooms, or installing
air cleaners can reduce involuntary smoking by children or spouses. Dry-
cleaned clothes could be aired out for a few hours on a balcony or porch
before hanging them in a closet.
Organizational Actions. As in the case of formaldehyde, manufacturers
may reduce toxic emissions from their products, either by modifying
manufacturing processes or substituting less toxic chemicals. Voluntary
building standards may be adopted, limiting emissions for building materials.
Local, state, or federal governments could adopt a variety of legislative
solutions, such as the various laws restricting smoking in public buildings.
The American Society of Heating, Refrigeration, and Air Conditioning
Engineers (ASHRAE) has for many years set voluntary guidelines for
ventilation of buildings.
Associations such as the Air Pollution Control Association, the American
Lung Association, the Association for Standards and Testing of Materials,
8
-------
the Consumer Federation of America, the National Institute for Building
Sciences, the American Institute of Architects, and others have in recent
years recognized the importance of indoor air pollution and have programs
designed to encourage research, communicate research results, establish
standards, and/or develop control techniques.
-------
Section 3.
Summary and Conclusions
The major findings of the TEAM Study may be summarized as follows:
1. Measurement of personal exposures using the Tenax personal monitors
was shown to be a feasible approach, acceptable to essentially all
subjects (ages 7 to 85), and capable of detecting exposures to most
of the target compounds at normal environmental concentrations.
2. Measurement of exhaled breath proved to be a sensitive and
noninvasive way to determine the presence of the target chemicals
in the blood.
3. Mean personal air exposures to essentially every one of the eleven
prevalent target chemicals were greater than mean outdoor concen-
trations at 7 of 8 locations/monitoring periods. (The one exception
was Los Angeles in February, where strong overnight inversions led
to elevated outdoor concentrations.) The upper 10% of personal
exposures always exceeded the upper 10% of outdoor concentrations
for all sites and time periods.
4. A major reason for these higher personal exposures appears to be
elevated indoor air levels at work and at home.
5. The elevated indoor air levels appear to be due to a variety of sources,
including consumer products, building materials, and personal
activities.
6. The breath levels correlated significantly with personal air exposures
to nearly all chemicals but did not correlate with outdoor air levels.
This is further corroboration of the relative importance of indoor air
compared to outdoor air.
7. A number of specific sources of exposure were identified, including:
a. Smoking (benzene, xylenes, ethylbenzene, styrene in breath)
b. Passive smoking (same chemicals in indoor air)
c. Visiting dry cleaners (tetrachloroethylene in breath)
d. Pumping gas or being exposed to auto exhaust (benzene in breath)
e. Various occupations, including: chemicals, plastics, wood
processing, scientific laboratories, garage or repair work, metal
work, printing, etc. (mostly aromatic chemicals in daytime personal
air)
10
-------
8. Other sources were hypothesized, including:
a. Use of hot water (showers, washing clothes) in the home
(chloroform in indoor air)
b. Room air fresheners, toilet bowl deodorizers, or moth crystals
(p-dichlorobenzene in indoor air)
9. In most cases, these sources far outweighed the impact of traditional
"major" point sources (chemical plants, petroleum refineries,
petrochemical plants) and area sources (dry cleaners and service
stations) on personal exposure.
10. For all chemicals except the trihalomethanes, the air route provided
>99% of the exposure. Drinking water provided nearly all of the
exposure to the three brominated trihalomethanes, and a substantial
fraction of most personal exposures to chloroform.
11
-------
Section 4.
Overview
The TEAM Study was designed by the USEPA to develop and demonstrate
methods to measure human exposure to toxic substances in air and drinking
water. All field operations were carried out by the Research Triangle Institute
(RTI) as the prime contractor. Precursor studies were undertaken in 1980
at Lamar University in Beaumont, Texas and the University of North Carolina
at Chapel Hill. A field test of the methodology (Phase I) was carried out
between July and December 1980 in Bayonne and Elizabeth, New Jersey
and Research Triangle Park, North Carolina. The objective of Phase II, carried
out between September 1981 and February 1983, was to estimate the
distribution of exposures to target substances for a target population in an
industrial/chemical manufacturing area (Bayonne and Elizabeth, New Jersey)
and to compare these estimated exposures to those estimated for populations
in non-chemical manufacturing areas (Greensboro, NC and Devils Lake, ND).
Phase III, carried out between February and June 1984, involved the
application of the methodology refined during Phase II to target populations
in California.
Selection of Target Chemicals
Several criteria were used to select target chemicals for the TEAM Study.
These included:
1. Toxicity, carcinogenicity, mutagenicity
2. Production volume
3. Presence in ambient air or drinking water at the field sites, as
determined by initial ambient sampling prior to each TEAM study
4. Existence of NBS permeation standards
5. Amenability to collection on Tenax
Each of these criteria will be discussed in turn. Toxic, carcinogenic, and
mutagenic chemicals received high priority because of their possible human
health effects. Thus, benzene (a human carcinogen); and four animal
carcinogens (chloroform, carbon tetrachloride, trichloroethylene, and
tetrachloroethylene) were selected. Mutagenic compounds such as styrene,
1,1,1-trichloroethane, 1,2-dichloroethane, various brominated compounds,
and the dichlorobenzene isomers were also selected. (Recently, a National
Toxicology Program (NTP) test of p-dichlorobenzene has shown it to be an
animal carcinogen.) Certain compounds considered neither carcinogenic nor
mutagenic at the time, but known to be toxic at high concentrations were
also selected: xylenes and ethylbenzene. Many common nontoxic compounds
such as hexane, heptane, cyclohexane, and trimethylbenzenes were omitted.
All the trihalomethanesfbromoform, chloroform, bromodichloromethane, and
dibromochloromethane) were included because of their prevalence in drinking
water. In the California portion of the study, several straight-chain
12
-------
hydrocarbons, some of which are promoters, or co-carcinogens, were added:
octane, decane, undecane, and dodecane, Another mutagen (a-pinene) was
also added at this time. Three compounds, vinylidene chloride, toluene, and
1,2-dichloropropane, were dropped because of low breakthrough volume,
contamination of blanks, and nondetectable environmental concentrations,
respectively.
Production volume was also considered in selecting chemicals. High-
volume chemicals such as benzene, toluene, xylenes, ethylbenzene, styrene,
and others were favored over low-volume chemicals, based on the probability
of finding them in the general populations.'
Presence in ambient air and drinking water was determined in both New
Jersey and California by preliminary sampling trips, during which sites were
established near known point sources (chemical plants, petroleum refineries,
etc.) and a series of 2-hr integrated samples collected on Tenax and
qualitatively analyzed to identify all chemicals collected via GC-MS analysis
and comparison with a library of spectra. These visits resulted in verifying
the presence of most of the initially selected target chemicals.
Existence of NBS permeation standards was the most stringent criterion:
at the time of planning the study, only 30-40 such standards existed. Without
such standards, only semiquantitative estimates could be made.
Amenability to collection on Tenax ruled out several chemicals of interest.
High-volatility chemicals such as vinyl chloride, methylene chloride, and
vinylidene chloride have breakthrough volumes on Tenax well below the
planned sampling volume of 20 L Reactive chemicals such as formaldehyde
cannot be collected on Tenax. Benzaldehyde, acetophenone, and phenols
are known artifacts of Tenax sampling and therefore could not be included.
Toluene was originally a target chemical, but was found .in such high and
variable amounts on the blank cartridges prepared by the principal laboratory
that it could not be included. (This may not be a fundamental problem of
Tenax, but rather a problem related to high levels of toluene in the primary
laboratory.)
Despite the above exclusions, the final target list of 20 compounds in
New Jersey (Table 2) and 26 in California (Table 3) included many of the
most prevalent toxic and carcinogenic volatile organic chemicals in outdoor
and indoor air and drinking water.
Table 2. Target Compounds Selected for Monitoring in Environmental
Media3: New Jersey
Vinylidene chloride Dibromochloropropane
Chloroform m-Dichlorobenzene
1,2-Dichloroethane o-Dichlorobenzene
1,1,1-Trichloroethane p-Dichlorobenzene
Carbon tetrachloride Benzene
Trichloroethylene Styrene
Bromodichloromethane Ethylbenzene
Dibromodichloromethane o-Xylene
Tetrachloroethylene m-Xylene
Chlorobenzene p-Xylene
Bromoform
"All compounds monitored in personal air, fixed-site air, breath and water.
13
-------
Table 3. Target Compounds Selected for Monitoring in Environmental
Media: California
Matrix: Personal and Fixed-Site Air
Chloroform
1,1,1-Trichloroethane
Benzene
Carbon tetrachloride
Trichloroethylene
Tetrachloroethylene
n-Decane
Dodecane
1,4-Dioxane
1,1,1,2-Tetrachloroethane
a-Pinene
Matrix: Drinking Water
Chloroform
Trichloroethylene
Dibromochloromethane
Chlorobenzene
Matrix: Breath
Bromodichloromethane
Dibromochloromethane
Chloroform
1,1,1-Trichloroethane
Benzene
Carbon tetrachloride
Tetrachloroethylene
n-Decane
Dodecane
1,4-Dioxane
1,1,.1,2-Tetrachloroethane
Bromoform
Chlorobenzene
Styrene
o.m.p-Dichlorobemenes
Ethylbenzene
o,m,p-Xylenes
1,2-Dibromoethane
Undecane
n-Octane
1,2-Dichloroethane
1,1,2,2-Tetrachloroethane
1,1,1-Trichloroethane
Bromodichloromethane
Tetrachloroethylene
Bromoform
Chlorobenzene
Styrene
o, m,p-Dich/orobenzene
Ethylbenzene
o,m,p-Xylenes
Trichloroethylene
1,2-Dibromoethane
n- Octane
Undecane
1,2-Dichloroethane
1,1,2,2-Tetrachloroethane
a-Pinene
Study Design
Phase II: New Jersey, North Carolina, and North Dakota
An initial probability sample of 5500 households located in 108 areas
in the two New Jersey cities was used to collect stratification data (age,
socio-economic status, occupation, proximity to major point sources) on over
14
-------
10,000 residents of these cities. A stratified probability sample of these
individuals yielded 355 participants for the Phase II study. Each eligible person
selected for monitoring had a "weight" equal to the inverse of his selection
probability. (For example, a person selected with a probability of 1 in 1000
had a "weight" of 1000—he represented 1000 persons.) These weights were
then adjusted for nonresponse—if only half the eligible persons in one stratum
responded, all the weights in that stratum were multiplied by two.
Occupationally-exposed persons were overrepresented. The probability-
based survey design provides a basis for robust inferences to the
approximately 128,000 members of the target population—individuals who
were residents of the target cities and over six years of age when the Phase
II study was conducted in the fall of 1981.
Each of the 355 participants carried a personal sampler during normal
daily activities for two consecutive 12-hour periods. An identical sampler
operated in the backyard of one participant in each of the 108 clusters of
homes for the same two 12-hour periods. Two drinking water samples were
also collected for each participant. At the end of the 24-hour sampling period,
a sample of exhaled breath, which was analyzed for the same compounds,
was collected for each participant. All participants also completed
questionnaires about their personal and household characteristics and
activities during the sampling period.
A return visit was made to 157 of the original participants in the summer
of 1982, and a final visit was made to 49 of these 157 persons in January-
February of 1983. The individuals contacted on each return visit were a
probability sample of the participants from the previous visit.
A small comparison study was undertaken in Greensboro, North Carolina
in May 1982. Greensboro was selected because its population is similar
in size to the Bayonne-Elizabeth area and it has similar small industries,
but no chemical manufacturing or petroleum refining operations. The target
sample size was set at 25 for a three-stage sample survey design to represent
approximately 131,000 Greensboro residents. Monitoring methods were
identical to those employed in New Jersey.
A second comparison site was selected to investigate whether the
population of a small, rural, agricultural town far from any industry would
exhibit personal exposures clearly different from those of the Northern New
Jersey population. Once again, the target sample size was set at 25 subjects
to represent approximately 7000 residents of Devils Lake, North Dakota.
Both comparison studies were meant to provide only a rough indication
of the range of likely exposures. Assuming a normal or log-normal distribution,
the median value for a sample of 25 would be expected to lie between the
30th and 70th percentiles of the true distribution with 95% confidence.
Phase III: California
This final phase of the TEAM Study was designed to replicate the New
Jersey study (using streamlined questionnaires and other improvements) in
areas of different meteorological conditions and complex chemical
manufacturing and petroleum refining industries.
Between February 3 and March 2, 1984, 117 residents selected from the
South Bay section of Los Angeles (Torrance, Carson, Hermosa Beach,
Redondo Beach, Manhattan Beach, El Segundo, Lomita, West Carson, six
Census tracts in Los Angeles, and seven adjoining Census tracts, with a
total population of 360,000) participated in the study. As in New Jersey,
they collected two consecutive 12-hour personal air samples and gave a
breath sample at the end of the 24-hour monitoring period (usually between
6 pm and 9 pm). The technicians collected a tap water sample on their
15
-------
final two visits to each home. These were analyzed separately and averaged.
Participants also filled out the household questionnaire and a 24-hour activity
recall diary.
The second Los Angeles trip (May 21 - June 2, 1984) included 52
participants, all of whom had participated in the first season study. The final
trip (June 3-25, 1984) included 71 residents of Antioch and Pittsburg,
California. These cities northeast of Oakland have extensive petrochemical
facilities and a combined population of 91,000.
Table 4 summarizes the locations and seasons of the TEAM Study, as
well as the numbers of participants for each location/season, and the target
populations at each site.
Response Rates
In New Jersey, 4426 of the 5578 households contacted agreed to fill out
the questionnaire, providing information on 11,414 people. The response
rates to the household screening stage ranged from 85% in New Jersey
to 95% in North Carolina and 96% in North Dakota. The response rates
of those asked to participate in the full study ranged from a low of 53%
in New Jersey (first visit) to 67% in North Dakota and 80% in North Carolina
(Table 5). The return visits to the New Jersey respondents showed
successively higher response rates of 79% and 91%.
The overall response rate is a product of the rates at each stage. Thus,
the New Jersey rate (first visit) is 85% x 51% = 44%. The North Carolina
overall response rate is 76% and the North Dakota rate is 64%.
In California, 1864 homes were screened (1260 in Los Angeles, 604 in
Contra Costa County) with an 88% completion rate. From the information
collected on more than 5000 residents of these homes, a total of 311 were
selected to participate, of which 293 were eligible, with 188 (64%) completing
the study. Thus, the overall response rate was 56% (88% x 64%).
Respondent Characteristics
Characteristics of the participants are listed in Table 6. Half are males
and 10-15% Black or Hispanic. Median ages were 30-35. About 60% were
employed. Smokers ranged from 31 % (Contra Costa) to 46% (New Jersey).
Table 4. Sites Visited in the Main TEAM Study
Site
Visit
Code Location
NJ1 Bayonne and Elizabeth, NJ
NJ2 Bayonne and Elizabeth, NJ
NJ3 Bayonne and Elizabeth, NJ
NC Greensboro, NC
ND Devils Lake, ND
LA1 Los Angeles, CA
LA2 Los Angeles, CA
CC Antioch and Pittsburg, CA
(Contra Costa County)
TOTAL 7 cities
Number of
Time of Visit Respondents
Sept-Nov 1981
July-Aug 1982
Jan-Feb 1983
May 1982
October 1982
February 1984
May 1984
June 1984
355
157a
49b
24
24
117
52C
71
591
Population
Represented
128,000
109,000
94,000
131,000
7,000
360,000
333,000
91,000
717,000
a Subset of NJ1 respondents.
b Subset of NJ2 respondents.
c Subset of LAI respondents.
16
-------
Table 5. Response Rates: All TEAM Sites
Households screened
Eligible households
Screening completed
Persons
Selected
Eligible
Completed study
Overall response rate3
New
Jersey
5578
5208
4426
(85%)
852
693
355
(51%)
44%
North
Carolina
307
295
280
(95%)
33
30
24
(80%)
76%
North
Dakota
104
91
87
(96%)
45
36
24
(67%)
64%
Los Angeles
1260
1219
1063
(87%)
190
182
117
(64%)
56%
Antioch/
Pittsburg
604
561
502
(89%)
121
111
71
(64%)
57%
3 Overall response rate = Screening rate x completion rate.
Table 6. Respondent Characteristics
Category
Sex
Race
Age
Heating Fuel
Stove Type
Employed
Potential Occupational
Exposure
Smoking Status
Smoke During Moni-
toring Period
Close Contact With
Smokers
Male
Female
White
Black
Hispanic
Asian
5-17
18-29
30-39
40-49
50-59
60-69
70-89
Gas
Oil
Gas
Electric
Yes
No
Yes
No
Current
Ex
Never
Yes
No
Yes
No
NJ1
183
179
249
60
44
1
54
WO
76
40
44
36
11
161
182
342
28
203
159
120
230
168
60
134
161
199
215
144
LA1
62
55
77
19
7
11
17
38
24
14
11
8
5
97
0
85
62
78
39
38
87
39
20
58
36
81
45
71
CC
36
34
51
5
5
7
14
22
15
13
4
2
0
68
0
10
59
44
26
22
52
22
12
36
22
49
31
39
17
-------
Measurement Methods
A complete description of the sampling and analytical protocols and
Standard Operating Procedures employed in this study may be found in
Volumes II, III, and IV of this report (refs. 22-24, Table 52). The following
is a brief description.
Personal and outdoor air samplers employed a glass cartridge containing
the solid granular sorbent Tenax-GC. A small Du Pont pump drew air at
—30 mL/min through the cartridge for —12 hrs to collect a target volume
of 20 L. A sampling vest was designed to hold the pump and the cartridge
close to breathing level (Figure 1) while leaving the participant's hands free
for normal activities.
Breath samples were collected using a specially-designed spirometer
(Figure 2) mounted in a van (Figure 3). The subject provided the breath sample
at his home in the evening (6-9 pm) at the end of the 24-hour sampling
period.
Water samples were collected from the tap at each participant's home
after a 20-second flushing period. Samples were collected in 2-oz glass jars
containing sodium thiosulfate to quench residual chlorine reactions.
Air and breath samples were analyzed by capillary gas chromatography
mass spectrometry (GC-MS) techniques followed by a combination of manual
and automated analyses of spectra. Water samples were analyzed by a purge
and trap GC method utilizing a Hall Electroconductivity detector for
halogenated compounds and a flame mnization detector for aromatics.
Depending on temperature, the sampling volume of —20 L sometimes
exceeded the breakthrough volumes for two of the target compounds:
chloroform and 1,1,1-trichloroethane. For these samples, concentrations
were calculated by dividing by the breakthrough volume rather than the
sampling volume. Thus in hot weather the concentrations of these two
chemicals reflect the final portion of the sampling period only
Two sampling protocol refinement studies were performed as a result of
difficulties encountered during sample collection and analysis. The first study
addressed sources of contamination associated with breath collection and
resulted in a greatly improved spirometer design. The second study evaluated
various approaches to preparation of clean Tenax cartridges and reduction
of contamination during storage, transport, and sampling. Improvements to
the sampling and analysis protocols resulting from these modifications were
implemented in subsequent sampling trips.
A total of nearly 5000 air, breath, and drinking water samples were collected
for 400 respondents (600 person-days) in the New Jersey, North Carolina,
and North Dakota sites. This represented about 95% of all samples originally
scheduled. During the California phase, about 1800 air, breath, and drinking
water samples were collected from 188 respondents (240 person-days). This
represented about 98% of all samples originally scheduled (Table 7).
Quality of the Data
An extensive quality assurance (QA) program was carried out. About 30%
of all samples were either blanks, spikes, or duplicates. Analysis of each
medium (air, water, breath) was repeated for 10% of samples in external
QA laboratories (NT Research Institute and the University of Miami Medical
School). Audits of all laboratory activities were undertaken by EPA's
Environmental Monitoring Systems Laboratory at Research Triangle Park,
North Carolina (EMSL-RTP) and spiked samples were supplied by EMSL-
RTP (air) and EPA's Environmental Monitoring and Support Laboratory in
Cincinnati (water). A separate QA report (included in its entirety in the
18
-------
Appendix to Volume II of this four-volume report (ref. 22 in Table 52)) was
written by an independent laboratory (Northrop Services, Inc.) concluding
that no significant analytical differences could be found among the three
air monitoring laboratories (Research Triangle Institute, NT Research Institute,
and EMSL-RTP).
Figure 1. Personal monitor and vest, showing glass cartridge containing Tenax-
GC sorbent. Vel-Cro flap to protect cartridge, and Dupont pump (in
pocket).
19
-------
Figure 2. Schematic of breath sampling apparatus.
Ultrapure
Air Tank
Tenax GC
Cartridges
Douglas
Valve and
Mouthpiece
Figure 3. Breath sampling system inside van with subject giving exhaled air.
20
-------
Table 7. Samples Collected at All TEAM Sites
NJ NC ND CA Total
Personal air
Drinking water
Breath
Outdoor air
QA/QCa
Total
1114
1130
559'
341
1282
4426
48
48
24
12
108
240
47
48
24
10
108
237
480
486
238
118
512
1834
1689
1712
845
481
2010
6737
a Includes blanks, controls, and duplicates.
Results
Quality Control/Quality Assurance
Recovery Efficiencies and Blank Values. In New Jersey, 155 field and
laboratory blanks analyzed during the first trip (Fall 1 981) showed generally
low background levels (<10 ng/cartridge, the equivalent of 0.5 /ug/m3) for
all target compounds except benzene, 1,1,1 -trichloroethane, chloroform, and
m,p-xylene (Table 8). Recovery efficiencies ranged from 80-110%.
In California, 40 blank cartridges for air and breath samples normally
contained less than 10 ng of all chemicals except benzene (15-36 ng),
chloroform (2-58 ng), and 1,1,1 -trichloroethane (6-36 ng) (Table 9).
Recoveries on 41 control cartridges ranged between 70-130% for most
chemicals, with the exception of the four trihalomethanes (42-200%).
Cartridges loaded with deuterated benzene, deuterated chlorobenzene, and
deuterated ethylbenzene gave recoveries ranging between 70-100%,
indicating acceptable operating losses.
Blanks for the water samples were very clean (Table 10); however,
recoveries were generally low: 50-90%.
After completion of the second visit to New Jersey in July-August 1982,
analysis of field blanks revealed very high background levels for a significant
portion of the Tenax cartridges. An investigation determined that renovations
had occurred at the hotel before the sampling team arrived. Although the
field cartridges were stored in paint cans, contamination apparently occurred.
The effect of the high blank levels can be seen in the increased coefficients
of variance (CVs) for the duplicate samples. Comparison of the variance of
the observed values with the variance of the duplicate samples indicates
that, except for benzene, the high blank values did not invalidate the results;
however, the possibility of a systematic bias due to over- and under-correction
for the blank values cannot be ruled out. Also, the correction factors that
should be applied to the observed frequency distributions are larger in the
second season than in the other seasons. In short, the precision of the second
season results is worse than the other seasons, and the residual bias could
be larger and of unknown direction.
Following this incident, all Tenax cartridges in the field were placed under
a constant helium bath during temporary storage in the field headquarters
site.
Because of the very large number of samples collected, some were not
analyzed until 2-3 months after they were collected. However, blanks and
controls stored with the field cartridges for the same length of time showed
acceptable contamination levels and recovery efficiency.
21
-------
Table 8. Blank Values and Recovery Efficiencies for Air and Breath
Samples: New Jersey
Compound
Vinylidene chloride
Chloroform
1, 2-Dichloroe thane
1, 1, 1-Trichloroethane
Benzene
Carbon tetrachloride
Trichloroethylene
Bromodichloromethane
Dibromochloromethane
Tetrachloroethylene
Chlorobenzene
Bromoform
Dibromochloropropane
Styrene
p-Dichlorobenzene
Ethylbenzene
o-Xylene
p-Xy/ene
o-Dichlorobenzene
Blanks
(ng/cartridge ± SDj
Field Lab
(N= 76J
-------
Table 9. Ranges of Mean Recoveries and Backgrounds for Field
Controls and Blanks—Air and Breath Samples: California
Range of Mean
Recoveries3
(%)
Personal Air
Compound fN-34)
1 ,2-Dichloroethane
7,7,7- Trichloroethane
Benzene
Carbon tetrachloride
Bromodichloromethane
Trichloroethylene
p-D/oxa/7e
Chlorodibromoethane
1 ,2-Dibromoethane
n-Octane
Tetrach/oroethy/ene
Chlorobenzene
Ethylbenzene
p-Xylene
Styrene
o-Xylene
a-Pinene
p-Dichlorobenzene
Bromoform
o-Dichlorobenzene
n-Decane
r\-Undecane
n-Dodecane
Chloroform
1 , 1 , 1 ,2-Tetrachloroethane
1 , 1 ,2,2-Tetrachloroethane
100-150
93-140
90-120
68-110
NAC
100-130
69-120
NA
74-120
97-120
78-120
85-110
91-100
90-110
85-100
96-120
79-110
81-110
NA
85-110
91-110
86-110
80-110
80-140
82-100
82-140
Range of Mean
Backgrounds*
(ng/cartridge)
Breath Personal Air
(N=16) (N=33)
87-100
71-106
77-117
56-90
48-74
97-120
64-100
42-92
74-130
88-100
91-100
84-100
93-98
89-95
69-97
95-99
80-90
81-98
46-52
88-100
71-94
88-98
92-100
45-200
92-110
100-110
NDh
6-8
17-31
ND
NA
ND-1
ND-5
NA
ND
ND-2
ND-3
ND-1
ND-5
2-5
2-13
2-3
ND
2-8
NA
3-6
ND-3
4-5
ND-2
2-58
ND-5
ND-9
Breath
(N=16)
ND
8-36
15-36
ND
ND
1-4
ND-5
ND
ND
ND-7
2-10
ND
3-4
6-9
6-11
3-6
ND-2
2-6
ND
ND-4
3-9
4-13
ND-7
8-29
ND-3
ND-8
"Each mean value calculated for a separate batch of Tenax—5 batches used
for personal sampling; 3 batches for breath sampling.
bNot detected.
cNot analyzed.
used Tedlar bags with GC/ECD ana lysis for halogens andGC/PID for benzene.
The TEAM samples were collected as part of the main study in the normal
fashion: two consecutive 1 2-hour samples using Tenax with GC/MS analysis.
The results indicate close agreement for five compounds above the
detection limits (Table 14); an additional six compounds were below the
detection limits of each system. The CARB results for trichloroethylene
23
-------
Table 10. Recoveries and Backgrounds for Field Controls and
Blanks—Water Samples
Recoveries
(Percent ± S.D.)
LA1a
(N= 12)
Chloroform
Bromodichloro-
methane
Chlorodibromo-
methane
Bromoform
1, 1, 1-Trichloroethane
Trichloroethylene
Tetrachloroethylene
Chlorobenzene
86
72
46
+
+
+
NA
88
86
78
65
+
+
+
+
16
27
56
e
18
14
16
15
LA2b
(N=6)
72 ±
58 ±
26 ±
28 ±
71 +
68 ±
67 ±
56 ±
10
10
12
13
13
11
11
12
CC°
(N=7)
62
50
47
47
53
55
51
53
±
±
±
±
±
±
±
±
16
19
25
16
26
21
22
11
LAI
(N= 12)
1.0
ND
ND
NA
ND
ND
ND
ND
Blanks
(ng/ml)
LA 2
(N=6)
NDd
ND
ND
ND
0.10
0.06
ND
ND
CC
(N=3)
ND
ND
ND
ND
0.06
ND
ND
ND
aLos Angeles—First trip—February 1984.
bLos Angeles—Second trip—May 1984.
°Contra Costa (Antioch/Pittsburg)—June 1984.
dNot detected.
eNot analyzed.
Table 11. Coefficients of Variation (%) for Duplicate Air and Breath
Samples in New Jersey—Season I
Compound Persona/3
Chloroform
1, 1, 1-Trichloroethane
Benzene
Carbon tetrachloride
Trichloroethylene
Tetrachloroethylene
Styrene
p-Dichlorobenzene
Ethylbenzene
o-Xylene
m,p-Xylene
2O
27
36
24
14
21
18
23
20
19
24
Median
75th Percent He
Outdoor11 Breath0
24
23
47
15
25
20
18
22
27
21
24
36
46
41
42
28
18
22
16
30
15
23
Personal
35
45
69
37
31
37
38
40
42
41
50
Outdoor
70
67
67
32
37
31
37
27
35
43
48
Breath
63
56
73
59
48
41
41
43
66
56
58
3N = 134.
bN = 34.
CN = 35.
exceeded the TEAM values in every case, while the reverse was true for
1,1,1-trichloroethane.
Performance Audits. EPA spiked the Tenax cartridges (provided by RTI)
with nine target compounds. These performance audit samples were
submitted blind to the RTI analyst. The samples from the third New Jersey
visit exhibited the lowest bias over all sites, reflecting improvement in the
24
-------
Table 12. Duplicate Air and Breath Samples—Median Relative Standard
Deviations (%): California
Personal Air
Target Chemicals
No. of Samples
Chloroform
1,2-Dichloroethane
1, 1, 1-Trichloroethane
Benzene
Carbon tetrachloride
Trichloroethylene
Tetrachloroethylene
Styrene
p-Dichlorobenzene
Ethylbenzene
o-Xylene
p-Xy/ene
n-Decane
n-Dodecane
1,4-Dioxane
r\-0ctane
n-Undecane
a-Pinene
Chlorobenzene
o-Dichlorobenzene
LAI3
24
28
12
8
13
15
12
12
28
30
13
13
15
12
13
20
11
15
14
72
12
LA2b
10
30
21
22
13
9
47
14
40
16
10
11
10
11
20
-
19
17
20
—
-
CC°
14
26
-
19
46
12
54
13
41
21
15
11
9
9
11
-
19
7
10
—
24
LA1a
12
25
-
11
20
7
10
15
24
11
26
14
20
26
19
-
4
51
11
—
115
Breath
LA2b
5
_d
-
13
52
14
24
28
-
45
-
-
33
-
—
-
—
—
25
—
-
Outdoor Air
CC°
7
—
—
43
18
0
-
18
17
17
25
5
14
15
8
-
3
7
19
—
-
LA1a
6
34
9
18
6
28
9
21
20
9
12
17
18
24
37
8
20
24
10
—
77
LA2b
6
40
-
25
10
17
-
17
27
21
27
45
18
14
6
-
20
26
22
—
-
CC?
2
111
—
46
11
43
-
-
-
-
29
35
20
17
—
—
—
—
—
—
-
aLos Angeles—First trip—January 1984.
bLos Angeles —Second trip —May 1984.
cContra Costa County (Antioch/Pittsburg)—June 1984.
dNo measurable values.
field procedures over time. The observed biases associated with most of
the target chemicals during all other trips were less than 30%, except for
the North Dakota samples, which exhibited the highest bias, apparently due
to a substandard batch of Tenax.
The performance audit water samples were provided by EMSL/EPA in
Cincinnati. In general, recoveries ranged from 80-90%. Bromine-containing
targets were recovered less completely (40-75%).
Percent Measurable
All measurements were classified into three groups: nondetectable, trace,
and measurable. Nondetectable values were those falling below the Limit
of Detection (LOD). Trace values exceeded the LOD, but fell below the
Quantifiable Limit (QL), generally chosen to be 4 times the LOD. Measurable
values exceeded the QL.
25
-------
Table 13. Duplicate Water Samples—Median Relative Standard
Deviations (%)
Chemical
Chloroform
Bromodichloromethane
Chlorodibromomethane
Bromoform
1, 1, 1-Trichloroe thane
Tetrachloroethylene
Trichloroethylene
LA1a
(N=24)
6
4
4
13
17
2
NAe
LA2b
(N= 10)
1
3
3
8
4
7
7
CC0
(N= 14)
9
3
6
5
57d
3
3
aLos Angeles—First trip—January 1984.
bLos Angeles—Second trip—May 1984.
c Contra Costa County (Antioch/Pittsburg)—June 1984.
dOnly one sample measurable.
eNot analyzed.
Table 14. Comparison of TEAM and CARB Co-located Sampling
Results (in ppb)
Chemical
Benzene
Carbon tetrachloride
Chloroform
1,2-Dichloroethane
1, 1, 1-Trichloroethane
Trichloroethylene
Tetrachloroethylene
Manhattan Beach
(Feb 19-20)
T> C"
8.0
0.11
0.24
0.08
5.9
0.08
1.2
10
0.08
0.15
0.22
1.4
0.47
1.2
Manhattan Beach
(Feb 20-21)
T C
6.2
0.14
0.62
0.06
6.5
0.18
2.0
6.2
0.11
0.63
<0.10
1.7
0.30
2.1
Carson
(Feb 21-22)
T C
5.2
0.10
0.068
0.03
4.7
0.09
1.2
4.7
0.09
0.065
<0.10
1.5
0.19
1.3
a TEAM results: average of two consecutive 12-hour outdoor air samples
collected on Tenax and analyzed by GC/MS.
b CARB results: one 24-hour Tedlar bag sample analyzed by GC/ECD
(halocarbons) and GC/PID (benzene).
Because of unavoidable losses of sampled materials on sorbents, values
below ~1 //g/m3 of most substances could not be reliably quantitated. Thus,
a classification of Not Detected cannot be construed to mean the chemical
was not present. In fact, most of the target chemicals have nonzero global
backgrounds.
For New Jersey, the target chemicals may be sorted into several categories
based on the percent of samples exceeding the QL (Table 15).
The first class, ubiquitous chemicals that were found in 33-100% of all
air and breath samples, includes two common solvents (1,1,1-trichloroethane
and tetrachloroethylene); several aromatic components of gasoline, paints,
and other petrochemical products (benzene, the xylene isomers, and
ethylbenzene); and two isomers of dichlorobenzene, used in moth crystals
and deodorizers.
The second class, compounds often but not always found in all sample
types, includes one additional solvent (trichloroethylene); a compound mainly
found in drinking water (chloroform); and a common component of consumer
26
-------
Table 15. Target Compounds Sorted by Percent Measurable in Breath
and Air Samples — All Three Seasons
Category and Compound Range of % Measurable
Ubiquitous Compounds
Benzene 55 - 100
Tetrachloroethylene 66 - 100
Ethylbenzene 62 - 100
o-Xylene 58 - 100
m,p-Xylene 68 - 100
m,p-Dichlorobenzene 44 - 100
1,1,1-Trichloroethane 33 - 99
Often Found
Chloroform 4 - 92
Trich/oroethy/ene 33 - 79
Styrene 46-91
Occasionally Found
Vinylidene chloride 0 - 95
1,2-Dich/oroethane 0 - 22
Carbon tetrachloride 0 - 53
Chlorobenzene 2 - 40
o-Dichlorobenzene 1 - 34
Bromodichloromethane 0 - 24
Dibromoch/oromethane 0 - 1
Bromoform 0 - 1
Dibromochloropropane 0 - 1
products (styrene, used in insulation and plastics). The sources of styrene
and the dichlorobenzenes may have been in the home based on the much
greater frequencies of measurable amounts in personal air samples (70-
80%) compared to outdoor air samples (20-40%).
The third class of substances were only occasionally found (<10%
measurable in most sample types). This class includes ethylene dichloride,
vinylidene chloride, carbon tetrachloride, bromodichloromethane, chloro-
benzene, and o-dichlorobenzene.
Finally, three brominated substances were almost never found in air or
breath: bromoform, dibromochloromethane, and dibromochloropropane.
Fewer target chemicals were found in drinking water in New Jersey (Table
16), and only the three trihalomethanes were ubiquitous. A second group
of three solvents appeared at low levels in nearly all tap water samples
collected in Elizabeth but in hardly any of the Bayonne samples.
For the personal air and breath samples collected at the two comparison
sites in Greensboro, North Carolina and Devils Lake, North Dakota, most
of the prevalent chemicals in New Jersey air and breath samples were again
found (Table 17). Only carbon tetrachloride appeared considerably less often
27
-------
Table 16. Target Compounds Sorted by Percent Measurable in Water
Samples — NJ — All Three Seasons
Category and Compound Range of % Measurable
Ubiquitous Compounds
Chloroform 99 - WO
Bromodichloromethane 99 - 100
Dibromochloromethane 93 - 100
Often Found
1,1,1-Trichloroethane 46 - 50
Trichloroethylene 44-51
Tetrachloroethylene 43 - 53
Occasionally Found
Vinylidene chloride 26 - 43
1,2-Dichloroethane 1
Benzene 1-25
Carbon tetrach/oride 6-18
Bromoform 2-6
Chlorobenzene 0 - 1
Dichlorobenzene isomers 0 - 3
Never Found
Ethylbenzene 0
Styrene O
Xylene isomers 0
than in New Jersey. In water samples, the same chemicals (trihalomethanes)
were detected as in New Jersey (Table 18).
In California, all 26 target chemicals were found in at least a few air
or water samples. Many were present in nearly every air or breath sample
(Table 19). The 11 prevalent airborne chemicals in New Jersey were also
prevalent in California; in addition, six of the ten new target chemicals were
also present much of the time.
In drinking water (Table 20) bromoform appeared in 70-90% of the samples,
compared to almost none of the New Jersey water samples. Once again, *
the common solvents (trichloroethylene, tetrachloroethylene, and 1,1,1-
trichloroethane) were present but at very low levels.
Concentrations
New Jersey (Fall 1981), Weighted frequency distributions for the
combined Bayonne-Elizabeth target population of 128,000 persons are shown
for all personal air, outdoor air, and breath samples of the eleven most
prevalent chemicals (Figures 4-14). Notable are the great range of exposures
(< 1 /jg/m3 to > 100 fjg/m3); the greater personal exposures than outdoor
28
-------
Table 17. Target Compounds Sorted by Percent Measurable in Air
Breath Samples — NC and ND
Category and Compound
Ubiquitous Compounds
1, 1, 1-Trichloroethane
Tetrachloroethylene
m , p -Dichlorobenzene
Ethylbenzene
o-Xylene
m,p-Xylene
Benzene
Often Found
Chloroform
Trich/oroethy/ene
Styrene
Occasionally Found
1,2-Dichloroe thane
Carbon tetrachloride
Bromodichloromethane
Chlorobenzene
o-Dichlorobenzene
Bromoform
Never Found
Dibromochloromethane
Dibromochloropropane
Range of %
NC
72 - 76
50 - WO
71 - 80
90 - 100
90 - 100
85 - 100
a
47 - 68
8- 68
41 - 64
4 - 14
4 - 6
0
0 - 16
0 - 2
0- 4
0
0
Measurable
ND
80-91
73 - 95
56 - 89
60 - 80
66 - 91
80 - 97
a
22 - 65
33- 52
59
5 - 17
8 - 14
14
7 - 44
0 - 10
0
0
0
"Benzene was ubiquitous, but high background contamination prevented
quantifying the results.
29
-------
Table 18. Target Compounds Sorted by Percent Measurable in Drinking
Water Samples — NC and ND"
Category and Compound
Ubiquitous Compounds
Chloroform
Bromodichloromethane
Often Found
Dibromochlorome thane
1, 1, 1-Trichloroethane
Occasionally Found
Tetrachloroethylene
Vinyl/dene chloride
Carbon tetrachloride
Trichloroethylene
Toluene
1, 2-Dichloroe thane
Chlorobenzene
Bromoform
Dichlorobenzene isomers
Never Found
Benzene
Styrene
Ethylbenzene
Xylene isomers
Range of °/
NC
93
93
93
24
74
10
3
5
NMb
0
0
0
0
NM
NM
NM
NM
6 Measurable
ND
100
73
18
42
0
0
0
5
30
2
2
8
2
0
0
0
0
aNC = North Carolina, ND = North Dakota.
bNot measured.
concentrations; and the greater breath concentrations than outdoor
concentrations in many cases.
As these figures illustrate, personal exposures were usually greater than
outdoor concentrations for all 11 prevalent target chemicals. The arithmetic
means of the daytime and overnight (i.e., indoor) personal air exposures
are several times the outdoor mean concentrations (Figures 15 and 16).
Because the distributions were more nearly log-normal than normal, the
geometric means are also compared (Figures 17 and 18).
Average 24-hour exposures were calculated from the two consecutive 12-
hour values for each subject, and weighted estimates of the population
frequency distributions were determined for the five aromatic compounds
(Figure 19) and the six halocarbons (Figure 20). Similarly, average 48-hour
exposures were calculated for the 157 persons who had both fall and summer
measurements. The 48-hour frequency distributions display similar
characteristics to the 12-hour distributions (Figure 21), with only a slight
decrease in the geometric standard deviation.
New Jersey (all three seasons). Estimates of 24-hour arithmetic mean
personal air exposures, breath concentrations, and outdoor air concentrations
during all three seasons in New Jersey are summarized in Table 21. Since
the overnight (6 pm - 6 am) personal air exposures were essentially measures
of indoor air (85% of persons did not go outside during the 12-hour monitoring
period) it is possible to compare indoor air concentrations directly with outdoor
air values just outside the residence. In 28 of 30 cases, the mean overnight
30
-------
Table 19. Target Compounds Sorted by Percent Measurable in Air and
Breath Samples
Range of Percent Measurable
Los Angeles, CA Antioch/Pittsburg, CA
Category and Compound
Ubiquitous Compounds
7, /, 1-Trichloroethane
Benzene
Tetrachloroethylene
Ethylbenzene
o-Xy/ene
m,p-Xylene
Often Found
n-Octane
n-Decane
m,p-Dichlorobenzene
Styrene
Carbon tetrachloride
a-Pinene
Chloroform
Occasionally Found
Trichloroethylene
n-Undecane
n-Dodecane
1,2-Dichloroethane
o-Dichlorobenzene
1,4-Dioxane
Chlorobenzene
1,2-Dibromoethane
1, 1, 1,2-Tetrachloroethane
1,1,2,2-Tetrachloroethane
1st Season
99- TOO
95-100
97-100
82-100
91-100
100
81-99
53-96
79-100
47-100
12-100
62-98
36-99
50-97
56-99
30-96
4-68
13-59
8-70
1-12
0-4
0-3
0-10
2nd Season
89- 100
79-100
99-100
70-100
57-100
100
59-94
25-81
61-87
37-94
11-100
47-92
31-80
4-66
48-74
17-45
0-23
0-19
3-21
0-8
0-13
0-12
0-18
49- 100
82-100
58-100
64-100
58-100
84-100
29-96
48-100
0-75
56-91
14-96
0-85
12-79
0-72
8-88
0-77
0-30
0-19
5-25
0-18
0-2
0-18
0-18
31
-------
Table 20. Target Compounds Sorted by Weighted Percent Measurable
in Drinking Water Samples
Range of Percent Measurable
Los Angeles
Category and Compound Jan-Feb 1984 May 1984
Ubiquitous
Chloroform
Bromodichloromethane
Dibromochloromethane
Often Found
Bromoform
94
93
89
69
86
96
85
90
Antioch/Pittsburg
94
96
85
69
Occasionally Found
1, 1, 1-Trichloroethane
Tetrachloroethylene
Trichloroethylene
Chlorobenzene
48
22
8
13
14
19
12
5
10
94
66
6
Figure 4, Benzene: Estimated frequency distributions of personal air exposures,
outdoor air concentrations, and exhaled breath values for the
combined Elizabeth-Bayonne target population (128,000). All air
values are 12-hour integrated samples. The breath value was taken
following the daytime air sample (6:00 am-6:00 pm). All outdoor
air samples were taken in the vicinity of the participants' homes.
Population Exceeding Concentration Shown
115,200 64,000 12,800 1,280
90% 50% 10% 1%
400
WO
I
Night
Benzene
- Legend
10
Personal Air
(N-344J
_ _ Breath
(N-320)
, _. Outdoor Air
IN-86)
400
100
10
10% 50% 90% 99%
12,800 64.000 115.200 127,000
Population eelow Concentration Shown
32
-------
to
to
to
CD
CO
Ui "to
tjO
CO
M
I
to —v
0) 05
CQv_
'~ 5?
^C-
•^
^5:
•£ -^
155
gj ^*
OQ -
'^ C
^v_
g
•^
^v_
•£ -^
co O
to O
00 v_
^i
^v
y
CO
ii
/, 7, 1 -Trichloroethane
m , p -Dichlorobenzene
^:
05
CO
O
o
* —
CO
o
05
^
CM
f)
to
-CO
a
E
^,
**,
CM
rj
CM
Q
CM
CO
T^.
CO
*"*"
O
Id
10
^
Tetrachloroethylene
0
*
^
y
Cj
5
0
5
6
^
05
^
05
00
CM
Benzene
CM
00
CO
CM
IV.
^
CM
CO
CM
10
^
q
05
*~~
Ethylbenzene
10 10
^ d
io ^f
co d
10
CO T*
•^ 05
IO US
10 00
co K
CO
CM cd
»^
\J- 00
co *-:
o c\
^ CNi
10 CO
**• **•
o-Xylene
Trichloroethylene
CO
d
CO
d
o
*
CO
id
co
CO
r->
CO
^
^
O
OQ
Chloroform
IV.
d
^
Q
^.
CM
CO
^
^
d
CM
CM
»-.'
0)
d
01
00
Styrene
9
^
^
1
^
d
o
0
CO
v^
"^
CO
o>
Carbon tetrachloride
^
CO
10
CM
10
CM
CO
10
05
10
Q
CM
8
00
00
00
CO
Total (11 compounds)
i
*o
-$
to
to
»»
o
CM
^ -1
-C *"*
! 1
to C
X °
to °
o "O
0 3 oo
-------
Figure 5.
Chloroform: Estimated frequency distributions of personal air
exposures, outdoor air concentrations, and exhaled breath values for
the combined Elizabeth-Bayonne target population (128,000). All
air values are 12-hour integrated samples. The breath value was taken
following the daytime air sample (6:00 am-6:00 pm). All outdoor
air samples were taken in the vicinity of the participants' homes.
Population Exceeding Concentration Shown
115,200
90%
64,000
50%
12,800
10%
1.280
1%
300
100
10
n—i—r—r
Chloroform
i i i—r~r
Day
Legend
Night
Breath
Personal Air
(N-344)
Breath
(N-320)
Outdoor Air
(N~86)
300
100
O)
10
10% 50% 90% 99%
12.800 64,000 115,200 127,000
Population Below Concentration Shown
personal air exposures exceeded the overnight outdoor air concentrations,
usually by factors of 2-10 (Table 22). The most extreme example was the
combined m- and p-dichlorobenzene isomers, with arithmetic means indoors
of about 50 fjg/m3 compared to outdoor values of less than 2 //g/m3. The
maximum personal air values for all chemicals were consistently in the
hundreds or thousands of /yg/m3, while maximum outdoor concentrations
were usually less than 100 //g/m3 (Table 23). Even breath maximum values
normally exceeded the outdoor air maxima. Finally, the comparison of drinking
water values across the three seasons (Table 24) shows that only the three
trihalomethanes had nonnegligible concentrations in the tap water samples.
Also clear is the sharp decline in the winter levels of trihalomethanes in
drinking water.
The observation of higher indoor than outdoor values in the fall of 1981
was corroborated in the summer and winter seasons. Figures 22 and 23
show an increase in the indoor/outdoor ratios of the median and 90th
34
-------
Figure 6.
1,1,1- Trichloroethsne: Estimated frequency distributions of personal
air exposures, outdoor air concentrations, and exhaled breath values
for the combined Elizabeth-Bayonne target population (128,000). All
air values are 12-hour integrated samples. The breath value was taken
following the daytime air sample (6:00 am-6:00 pm). All outdoor
air samples were taken in the vicinity of the participants' homes.
Population Exceeding Concentration Shown
115,200
90%
64,000
50%
12,800
10%
1,280
1%
5.000
1,1,1 - Trichloroethane
Day
1,000
Legend
100
I
Personal Air
(N-344)
Breath
IN-320)
Outdoor Air
(N~86)
10
Night
Breath
Day
Night
5.000
1,000
100
10
10% 50% 90% 99%
12,800 64.000 115.200 127.000
Population Below Concentration Shown
35
-------
Figure 7.
Tetrachloroethylene: Estimated frequency distributions of personal
air exposures, outdoor air concentrations, and exhaled breath values
for the combined Elizabeth-Bayonne target population (128.000). All
air values are 12-hour integrated samples. The breath value was taken
following the daytime air sample (6:00 am-6:00 pm). All outdoor
air samples were taken in the vicinity of the participants' homes.
Population Exceeding Concentration Shown
115,200 64.000 12,800 1,280
2,000
1,000
100
1
10
90%
—I—
50%
I l i
-\—r
10%
—I—I
1%
Tetrachloroethylene
Legend
' Personal Air
(N-344)
Breath
(N-320)
Outdoor Air
(N~86)
Night
2,000
1,000
100
1
10
10% 50% 90% 99%
12,800 64,000 115,200 127,000
Population Below Concentration Shown
percentile values for most chemicals from summer to fall to winter. The
wintertime increase appears to be due in some cases to somewhat reduced
outdoor concentrations rather than to increased indoor concentrations.
However, three chemicals (1,1,1-trichloroethane, tetrachloroethylene, and
/r?,p-dichlorobenzene) showed absolute increases in their indoor-outdoor
differences, consistent with either increased source activity or reduced air
exchange. An example of the increased indoor air concentrations in winter
compared to stable outdoor air concentrations is shown forp-dichlorobenzene
(Figure 24). (See Appendix F for other chemicals.)
36
-------
in
^
.0
tM
£
c
u
§
<3
•S:
V.
0
^
S
a
s
1
§•
o
G
?
5
•^
§
1
&
LU
1
s
2
Q*
^^
•S>
i^
O "x.
c ~^-
*OMLOCT)tj'^»-.OM'^-tOQ
«~:^o6^<:ooooddd$
DO o o c\ u
^-.^CTjOoO^-CTjOo'^ONJQ
oo LO CM *-• ^ *-- ^
to ^
c\i LO CM CM CM CM d d oo *-
00
^ O LO 00 00 to Q
OT--:*..^Ooo'^ixCMd»-:
v- *. $ *-
O^ 00 O 00 to O ^
*••» O) O) CT) >J l^s GO ^J* XJ1 C^ T--^
C\ \J~ *•» ^*
QixLofoo^oo^f-o^rxoofV]
CM oo ^
^•LO ^-coOoO^-fNO^c^i
LT)*--*-.OoC600^CNJ*-.Q»^
*^"
oo rs rx
O^oLo^O^^of^oofN^.
^--loio^-c^o*-*- i-..
Q)
g S g 1
1 S 1 g 5
B Q) >> oj >
0 •§ ^ <» ^ S
5 P ~-^JcoCj
co
O
OM
rx
LO
OM
to
LO
to
OM
IN
00
^t"
00
Total (11 compounds)
^
"5
§:
to
?
a
to
c
•^
to
tu
!§
-5
^ S
.O to
^ .c
id Bayonne
res contami
ties).
tB "Q ^"
_. '5 5
3 Population of Elizabetl
b Not calculated — cart
0 Not detected (most st
37
-------
Table 23. Maximum Concentrations fag/m3) of Organic Compounds in
Air and Breath of 350 NJ Residents
Personal Air3
Chemical
Chloroform
1, 1, 1-Trich/oroethane
Benzene
Carbon tetrachloride
Trichloroethylene
Tetrachloroethylene
Styrene
m, p -Dichlorobenzene
Ethylbenzene
o-Xylene
m,p-Xylene
Night
210
8300
510
1100
350
250
76
1600
380
750
3100
Day
140
330,000
270
900
1,400
12,000
6,500
2,600
1,500
1,800
10,000
Outdoor Air*
Night
130
51
91
14
61
27
11
13
28
31
70
Day
230
470
44
7.1
100
95
6.3
57
39
19
47
Breath0
29
520
200
250
30
280
31
160
290
220
350
3 Number of samples: 540 during three seasons.
bNumber of samples: 150 during three seasons.
c Number of samples: 500 during three seasons.
Table 24. Arithmetic Means and Maxima l^g/L) of Organic Compounds
in New Jersey Drinking Water
Fall 1981
(128,000)a
Chemical
Chloroform
Bromodichloromethane
Dibromochloromethane
1, 1, 1-Trichloroethane
Trichloroethylene
Tetrachloroethylene
Toluene
Vinylidene chloride
Benzene
Mean
70
14
2.4
0.6
0.6
0.4
0.4
0.2
—
Max
170
23
8.4
5.3
4.2
3.3
2.7
2.4
—
Summer 1982
(109,000)b
Mean
61
14
2.1
0.2
0.4
0.4
-
0.1
O.7
Max
130
54
7.2
2.6
8.3
9.3
-
2.5
4.8
Winter 1983
(94,000)°
Mean
17
5.4
1.4
0.2
0.4
0.4
—
0.2
—
Max
33
16
3
1.6
3.4
5.0
-
0.9
—
a,b,cpopulation of Bayonne and Elizabeth to which estimates apply.
38
-------
Figure 8.
Trichloroethylene: Estimated frequency distributions of personal air
exposures, outdoor air concentrations, and exhaled breath values for
the combined Elizabeth-Bayonne target population (128.000). All air
values are 12-hour integrated samples. The breath value was taken
following the daytime air sample (6:00 am-6:00 pm). All outdoor
air samples were taken in the vicinity of the participants' homes.
Population Exceeding Concentration Shown
115,200
90%
64,000
50%
12,800
10%
1,280
1%
500
Trichloroethylene
Day
Night
100
Legend
10
—• Personal Air
f/V-3441
—• Breath
(N-320)
•-• Outdoor Air
(N~86)
500
100
10
10% 50% 90% 99%
12,800 64.000 115,200 127,000
Population Below Concentration Shown
39
-------
Figure 9. Carbon tetrachloride: Estimated frequency distributions of personal
air exposures, outdoor air concentrations, and exhaled breath values
for the combined Elizabeth-Bayonne target population (128.000). All
air values are 12-hour integrated samples. The breath value was taken
following the daytime air sample (6:00 am-6:00 pmj. AH outdoor
air samples were taken in the vicinity of the participants' homes.
Population Exceeding Concentration Shown
115.200 64,000 J 2,800 1,280
90% 50% 10% 1%
200
100
i—r i i i i i i
Carbon Tetrachloride
Night
Legend
10
Day
Breath
-» Personal Air
(N-344)
-« Breath
(N-320)
-• Outdoor Air
(N-861
J L
200
700
10
10% 50% 90% 99%
12,800 64.000 115,200 127,000
Population Below Concentration Shown
40
-------
Figure 10. m,p-Dichlorobemene: Estimated frequency distributions of personal
air exposures, outdoor air concentrations, and exhaled breath values
for the combined Elizabeth-Bayonne target population (128.000).
All air values are 12-hour integrated samples. The breath value was
taken following the daytime air sample (6:00 am-6:00 pm). All
outdoor air samples were taken in the vicinity of the participants'
homes.
Population Exceeding Concentration Shown
115.200 64,000 12,800 1,280
90% 50% 10% 1%
2,000
1,000
m ,p-Dichlorobenzene
100
Legend
> Personal Air
(N-344)
' Breath
(N~320>
• Outdoor Air
(N~86)
10
Night
Day
Breath
Day
Night
2,000
1.000
100
10
10% 50% 90% 99%
12,800 64.000 115.200 127.000
Population Below Concentration Shown
41
-------
Figure 11. Styrene: Estimated frequency distributions of personal air
exposures, outdoor air concentrations, and exhaled breath values
for the combined Elizabeth-Bayonne target population (128.000).
All air values are 12-hour integrated samples. The breath value
was taken following the daytime air sample (6:00 am-6:00 pm).
All outdoor air samples were taken in the vicinity of the participants'
homes.
Population Exceeding Concentration Shown
115.200 64.000 12.800 1.280
90% 50% 10% 1%
200
Day
100
Styrene
Legend
10
• Persona/ Air
(N-344)
• Breath
IN-320)
, Outdoor Air
fN-86)
200
100
\
10
10% 50% 90% 99%
12,800 64,000 115,200 127,000
Population Below Concentration Shown
42
-------
Figure 12.
Ethylbenzene: Estimated frequency distributions of personal air
exposures, outdoor air concentrations, and exhaled breath values
for the combined Elizabeth-Bayonne target population (128,000).
All air values are 12-hour integrated samples. The breath value was
taken following the daytime air sample (6:00 am-6:00 pm). All
outdoor air samples were taken in the vicinity of the participants'
homes.
Population Exceeding Concentration Shown
115,200 64,000 12.800 1,280
90%
50%
10%
1%
500
Ethylbenzene
100
Legend
10
• 500
Persona/ Air
(N-344)
Breath
(N-320)
Outdoor Air
IN-86)
100
10
10%
12,800
50%
64.000
90% 99%
115,200 127,000
Population Below Concentration Shown
43
-------
Figure 13. m.p-Xylene: Estimated frequency distributions of personal air
exposures, outdoor air concentrations, and exhaled breath values
for the combined Elizabeth-Bayonne target population (128,000).
AH air values are 12-hour integrated samples. The breath value was
taken following the daytime air sample (6:00 am-6:00 pm). All
outdoor air samples were taken in the vicinity of the participants'
homes.
Population Exceeding Concentration Shown
115,200 64,000 12,800 1,280
90% 50% 10% 1%
1,000
100
Legend
i Personal Air
(N-344)
i Breath
(N-320)
> Outdoor Air
IN-86)
Breath
10
I i i
7,000
;oo
10
10%
12,800
50%
64.000
90% 99%
115,200 127.000
Population Below Concentration Shown
44
-------
Figure 14. o-Xylene: Estimated frequency distributions of personal air
exposures, outdoor air concentrations, and exhaled breath values
for the combined Clizabeth-Bayonne target population (128,000).
Alt air values are 1 2-hour integrated samples. The breath value was
taken following the daytime air sample (6:00 am-6:00 pm). All
outdoor air samples were taken in the vicinity of the participants'
homes.
Population Exceeding Concentration Shown
115,200
90%
64,000
50%
12,800
10%
1,280
1%
400
•Ill I
o-Xylene
100
Legend
10
Personal Air
(N-344)
Breath
(N-320)
Outdoor Air
(N~86)
400
100
10
10% 50% 90% 99%
12,800 64,000 115,200 127.000
Population Below Concentration Shown
-------
Figure 15. Estimated arithmetic means of 11 toxic compounds in daytime (6:00
am - 6:00 pm) air samples for the target population 1128.000) of
Elizabeth and Bayonne, New Jersey, between September and
November 1981. Personal air estimates based on 340 samples:
outdoor air estimates based on 88 samples.
6
i
c
to
I
.o
9)
•g
0)
I
I
100
10
0.1
Actual Personal
Value is 820
Legend:
Personal
Outdoor
Figure 16. Estimated arithmetic means of 11 toxic compounds in overnight
(6:OO pm - 6:OOam) air samples for the target population (128.0OO)
of Elizabeth and Bayonne. New Jersey, between September and
November 1981. Personal air (i.e.. indoor) estimates based on 347
samples; outdoor air estimates based on 84 samples.
— Legend:
Indoor
Outdoor
-------
Figure 17. Estimated geometric means of 11 toxic compounds in daytime (6:00
am - 6:00 pm) air samples for the target population (128,000) of
Elizabeth and Bayonne, New Jersey, between September and
November 1981. Personal air estimates based on 340 samples;
outdoor air estimates based on 88 samples.
Legend:
• Personal
123 Outdoor
Figure 18. Estimated geometric means of 11 toxic compounds in overnight
(6:00 pm -6:00 am) air samples for the target population (128.000)
of Elizabeth and Bayonne, New Jersey, between September and
November 1981. Personal air estimates based on 340 samples;
outdoor air estimates based on 84 samples.
Legend:
Persona/
Outdoor
47
-------
Figure 19. Weighted frequency distributions for 24-hour exposures of 355 New
Jersey residents to aromatic compounds (Fall 1981).
Population Exceeding Concentration Shown
64000 32000 12800 2500
WOO,
1000
500 -
200-
a
p-Xylene
o-Xylene
Ethylbenzene
Benzene
Styrene
100 -
10
50
75 90 95
Cumulative Frequency, Percent
99
48
-------
Figure 20. Weighted frequency distributions for 24-hour exposures of 355 New
Jersey residents to six chlorinated compounds (Fall 1981).
Population Exceeding Concentration Shown
64000 32000 12800 2500
WOO
500
o 7,1.1 - Trichloroethane
u p.Dichlorobenzene
A Tetrachloroethylene
• Carbon Tetrachloride
• Trichloroethylene
A Chloroform
1000
500
200
100
c
91
U
C
O
50
20
10
75 90
Cumulative Frequency. Percent
95
98 99
49
-------
Figure 21. Weighted frequency distributions of day and night 12-hour personal
air exposures compared to the 48-hour average for 160 New Jersey
residents (Fall-Summer 1981-82).
WOO
500
300
200
700
50
I 30
Population Exceeding Concentration Shown (x 103J
64 32 13 6.4 1.3
5
I
a
20
10-
Tetrachloroethylene
48 h
average
50 70 9095 99
Cumulative Frequency, Percent
Two chemicals (chloroform and trichloroethylene) had elevated outdoor
concentrations in summer.
Although the two smaller studies in Greensboro, North Carolina and Devils
Lake, North Dakota were carried out in different seasons, a limited comparison
indicates that the same chemicals with few exceptions were prevalent in
air, breath, and water samples in the two cities. Personal air and breath
levels were also similar in both cities.
Greensboro. A total of 242 samples were collected, of which 110 were
quality control or quality assurance samples. Blank values were very high
for 1,1,1 -trichloroethane and benzene; thus, data for these chemicals should
be viewed with caution. Precision was very good for air duplicates and
acceptable for breath duplicates.
Personal air exposures were again greater than outdoor air exposures for
most of the target chemicals (Table 25), although the small number of outdoor
air samples makes this only a tentative conclusion. A large range in personal
air exposures and breath concentrations was again evident, although mean
daytime personal air values were somewhat below those observed in the
winter season in New Jersey. Correlations between breath and daytime
personal air exposures were significant for only three of eight prevalent
chemicals.
Devils Lake. A total of 237 air, water, and breath samples were collected,
of which 108 were QA/QC samples. As with the Greensboro Tenax samples,
-------
Figure 22. Ratios of median 12-hour indoor air concentrations to simultaneous
12-hour outdoor air concentrations for New Jersey homes (N-8S
in Fall 1981; N=70 in Summer 1981; N=10 in Winter 1983).
Effect of Seasons on Indoor-Outdoor Ratios of
Median 12 -Hour Integrated Overnight Concentrations:
Matched Homes in Bayonne-Elizabeth, NJ
Legend
( i Summer
Fall
Winter
Figure 23. Ratios of 90th-percentHe 12-hour indoor air concentrations to
simultaneous outdoor air concentrations in New Jersey homes.
Effect of Seasons on Indoor-Outdoor Ratios of
90th Percentile 12-Hour Integrated Overnight Concentrations:
Matched Homes in Bayonne-Elizabeth. NJ
—Actual Ratio-
Value is 63.2
Legend
\ 1 Summer
BB Fall
Mi Winter
51
-------
Figure 24.
WOO
700 -
Weighted cumulative frequency distributions of overnight personal
air exposures and outdoor air concentrations of m,p-
dichlorobenzene isomers in New Jersey. Sample sizes are 350 (Fall
1981); 160 (Summer 1982); and 50 (Winter 1983) for the personal
air exposures; and 85 (Fall 1981); 70 (Summer 1982); and 10
(Winter 1983) for the outdoor air concentrations.
10
20 3040506070 80 90 95 98 9999.5 99.9
Cumulative Frequency, Percent
52
-------
unacceptably high and variable background concentrations of benzene and
1,1,1-trichloroethane occurred. Median coefficients of variance of duplicate
samples were in the usual ranges of 10-30% for air, but very high levels
of 30-70% for breath samples. Thus the Devils Lake breath data may be
less trustworthy than other breath values.
Personal air exposures again exceeded outdoor air concentrations for all
target compounds, although caution is indicated since the number of outdoor
air samples was extremely small (Table 26). Most chemicals were not
measurable in outdoor air, but indoor levels remained comparable to those
observed in Greensboro. Drinking water concentrations of chloroform were
exceedingly low (< 1 fjg/L).
Los Angeles (February 1984) The 117 participants represented a total
of 360,000 residents of the South Bay section of Los Angeles. The highest
weighted 24-hour personal air exposures (Table 27} were to 1,1,1-
trichloroethane (Figure 25), m,p-xylene, m,p-dichlorobenzene (Figure 26),
benzene (Figure 27), and tetrachloroethylene (Figure 28). Outdoor
concentrations, particularly at night, were unusually high, exceeding daytime
outdoor levels by 50% or more. Breath means ranged from 10-30% of personal
exposures for most chemicals except tetrachloroethylene (75%) and benzene
(45%).The four straight-chain hydrocarbons added for the California study
maintained consistent relationships among themselves in both outdoor and
indoor air, with octane and undecane the highest, dodecane the lowest (Figure
29).
Figure 25. 1,1,1-Trichloroethane: Estimated frequency distributions of
personal air exposures, outdoor air concentrations, and exhaled
breath values for the target population of 360.000 persons in the
South Bay section of Los Angeles. All air values are 10-14 hr
integrated samples. The breath values were taken following the
daytime air sample (6:00 am-6:00 pm). All outdoor air samples
were taken in the vicinity of the participants' homes. (Feb. 1984)
Population (000) Exceeding Concentration Shown
300 WO JO
1,000
1 10°
i
10
"'' '' \
Personal Air (N= 110)
Breath (N= 110)
Outdoor Air (N=25)
1 5 20 40 60 80 95 99
Cumulative Frequency, percent
53
-------
Table 25. Indoor/Outdoor Ratios in Greensboro, NC
Chemical
Chloroform
1, 1, 1-Trichloroethane
Benzene
Carbon tetrachloride
Trichloroethylene
Tetrachloroethylene
Styrene
m, p -Dichlorobenzene
Ethylbenzene
o-Xylene
m,p-Xylene
Median
Indoor3
2.3°
26
11
1.3
1.0
2.8
0.8
3.4
2.2
3.7
6.4
Values
Outdoor11
0.1 4C
60
0.4
0.1
0.2
0.7
0.1
0.4
0.3
0.6
1.5
Ratio
(I/O)
15
0.5
20
10
5
4
8
8
7
6
4
Maximum Values
Indoor
5.5C
110
43
3.6
8.7
57
3.1
72
20
26
62
Outdoor
1.3°
275.0
82.0
0.45
2.4
1.7
0.31
1.7
3.3
3.8
11.0
Ratio
(I/O)
4
0.4
0.5
8
3
30
10
40
6
7
6
aN = 24 (overnight personal air samples).
bN = 6.
Table 26. Indoor/ Outdoor Radios in Devils Lake, ND
Median Values
Chemical
Chloroform
1,1, 1-Trichloroethane
Benzene
Carbon tetrachloride
Trichloroethylene
Tetra chloroeth ylene
Styrene
m ,p-Dichlorobenzene
Ethylbenzene
o-Xylene
m,p-Xylene
Indoor'
0.14C
37
e
0.8
0.7
4.4
—
1.7
2.8
3.5
8.4
Outdoor*
0.05c'd
0.05"
—
0.46"
0.08"
0.69
—
0.07d
0.03d
0.05d
0.05"
Ratio
(I/O)
3
70
—
2
9
6
—
25
90
70
170
Maximum Values
Indoor
2.8C
1100
—
10
32
45
—
230
11
19
40
Outdoor
0.78C
5.0
—
0.84
1.1
3.4
—
2.0
1.8
1.0
2.2
Ratio
(I/O)
3
200
—
12
30
13
—
110
6
19
18
"/V = 23 (overnight personal air samples).
6 Not detectable - value equals 1/2 the limit of detection.
'Data uncertain based on Quality assurance results.
54
-------
*a
wl f
.sii
>S °
11
*
SI
i|!
« '5 S
I:
0
O
CM
QQ
|
rr^S
t
-------
^
Qk
g
•C
O
,JJ
is'
M
I.
"5
-^.
2
(J
U
"2.
1
^
CM
<
•£
CO -^
Qj IX
^^
< >- _
Sl^l
"fQ —,
i! t"- Co
v, <. V^ >U
j; k. *• J
o-§^^l
"CD ^
•i. C < O
# 9. S= 10
•-W vj x, -«^
1
<
_J
•g
CO r\
gS
*is v««
OQ ti
•*~ o , S'
O^^S
§
"7S *^"
, ^ a
>i c ^ H
3> 0 13 >
0, U) ^. ^
Chemical
CN
d
^
d
03
*^.
c\
d
IX
d
CD
d
d
d
-S
Q.
1
to
^
O
-C
53
1
c
-o
c
to
x
-§
S^
o
1
2
5
.0
4^
.
O
2
o
s
s
nj
3
§
^>
-C
1
tl]
<§
0
-------
Figure 26. p-Dichlorobenzene: Estimated frequency distributions of personal
air exposures, outdoor air concentrations, and exhaled breath values
for the target population of 360.000 persons in the South Bay
section of Los Angeles. All air values are 10-14 hr integrated
samples. The breath values were taken following the daytime air
sample (6:00 am-6:00 pm). All outdoor air samples were taken in
the vicinity of the participants' homes. (Feb. 1984)
Population (000) Exceeding Concentration Shown
i,ooot
700
Night
.§
a l°t
— Personal Air (N= 110)
- Breath (N=110)
-• Outdoor Air (N=25)
1 5 20 406080 95 99
Cumulative Frequency, percent
Los Angeles (May 1984). The second trip to 50 of the original participants
resulted in estimates of exposures for 330,000 Los Angeles residents (Table
27). Concentrations, both personal and outdoor, were considerably reduced
for 18 of the 19 prevalent chemicals. However, the same chemicals appeared
in roughly the same order. Outdoor overnight values no longer exceeded
daytime levels, and personal exposures nearly always exceeded outdoor
concentrations. Benzene (Figure 30) and /r?,p-dichlorobenzene (Figure 31)
concentrations in air and breath are presented as examples.
Contra Costa (June 1984). Seventy-one residents of Antioch and
Pittsburg, California represented a target population of 91,000 persons.
Weighted air and breath exposures were lower than in Los Angeles (Table
27), but again the same five chemicals were responsible for the highest
exposures. Air and breath concentrations of benzene (Figure 32) and m.p-
dichlorobenzene (Figure 33) are again presented for comparison. The relative
concentrations of the straight-chain hydrocarbons were different in Contra
Costa, with decane highest outdoors (Figure 34).
Concentrations in Drinking Water. Table 28 gives the levels of chemicals
measured in drinking water. Chloroform was the predominant trihalome-
thane. Brominated trihalomethanes were very evident also, especially
bromoform during the May 1984 period in Los Angeles, where the arithmetic
mean was 8 /vg/L.
57
-------
Figure 27. Benzene: Estimated frequency distributions of personal air
exposures, outdoor air concentrations, and exhaled breath values
for the target population of 360,000 persons in the South Bay
section of Los Angeles. All air values are 10-14 hr integrated
samples. The breath values were taken following the daytime air
sample (6:00 am-6:00 pm). All outdoor air samples were taken
in the vicinity of the participants' homes. (Feb. 1984)
Population (000) Exceeding Concentration Shown
1,000
300
100
10 1
100
c
5)
u
I 10
Day
•—• Personal Air (N=110)
- -« Breath (N=110)
•—-• Outdoor Air fN=25)
Table 28.
1 5 20 40 60 80 95 99
Cumulative Frequency, percent
Estimates of Drinking Water Concentrations for
California Residents
Los Angeles
(N=117)
Chemical
Chloroform
Bromodichloromethane
Dibromochloromethane
Bromoform
1, 1, 1 -Trichloroethane
Trichloroethylene
Tetrachloroethylene
Feb.
Arith.
Mean
14a
11
9.4
0.8
0.15
0.08
0.07
1984
SE
1.41a
0.84
0.91
0.14
0.04
0.01
0.01
Los Angeles
(N=52)
May
Arith.
Mean
29*
20
28
8
0.08
0.07
0.04
1984
SE
3.4"
2.3
3.1
2.4
0.02
0.02
0.02
Contra Costa
(N= 71)
June
Arith.
Mean
42a
21
8
0.8
0.09
0.06
0.10
1984
SE
3.1"
1.4
0.56
0.09
0.04
0.01
0.09
58
-------
Figure 28. Tetrachloroethylene: Estimated frequency distributions of personal
air exposures, outdoor air concentrations, and exhaled breath values
for the target population of 360.000 persons in the South Bay
section of Los Angeles. All air values are 10-14 hr integrated
samples. The breath values were taken following the daytime air
sample (6:00 am-6:00 pm). All outdoor air samples were taken in
the vicinity of the participants' homes. (Feb. 1984)
Population (000) Exceeding Concentration Shown
300 100 10 1
1.000
1/00
§
a
10
'—• Personal Air (N=110)
- - Breath (N= 110)
•—-• Outdoor Air f/v=25)
1 5 20 406080 95 99
Cumulative Frequency, percent
Indoor-Outdoor Comparisons
Since most participants remained in their homes during the overnight
sampling period (6 pm - 6 am), these personal air samples may be considered
indoor air samples and may be compared with the outdoor air samples
collected concurrently in the backyards of the homes. Most chemicals were
higher indoors than outdoors at all locations: many were significantly higher
(Table 29).
Correlations
Breath versus Personal Air. Spearman rank correlation coefficients were
calculated for the breath measurements and the preceding 12-hour personal
air exposures. The Spearman nonparametric statistics were employed to avoid
the problems of parametric statistics in dealing with highly skewed
distributions. Ten of the eleven prevalent chemicals in the breath of the
355 New Jersey residents were significantly correlated (most at probabilities
p < .0001) with the previous 12-hour average air exposures (Table 30). (The
11th chemical, chloroform, showed a significant correlation between breath
and drinking water concentrations.) Since many of these chemicals are
metabolized, excreted through other pathways than breath, and stored in
different body compartments for different characteristic residence times, and
since their concentration in breath depends partially on the previous blood
concentration at the beginning of the monitoring period and also on the
59
-------
Figure 29. Octane. Decane, Undecane, and Dodecane: Estimated frequency
distributions of overnight concentrations in participants' homes
compared to overnight outdoor air concentrations. (L.A., Feb. 1984)
Population (000) Exceeding Concentration Shown
300 100 JO 1
.§
01
u
c
100
10
0.1
Octane
•—• Personal Air (N= 110) ' L Undecane
•- --Outdoor Air (N=25) //// Decane
,«,Vx .
Octane «•'/
Undecane ' l[
Decane
Dodecane r
Dodecane
1 5 20 40 60 80 95 99
Cumulative Frequency, percent
Figure 30. Benzene: Estimated frequency distributions of personal air
exposures, outdoor air concentrations, and breath values for the
target population of 330,000 residents in the South Bay section
of Los Angeles. All air values are 10-14 hr integrated samples.
The breath values were taken following the daytime air sample (6:00
am-6:00 pm). All outdoor air samples were taken in the vicinity
of the participants' homes. (May 1984)
Population (000) Exceeding Concentration Shown
100
300
100
10
1
i
§ io\-
c
01
8
a
.--•Night
5 20 40 60 80 95 99
Cumulative Frequency, percent
Personal Air (N=50)
Breath (N=50)
Outdoor Air (N=25)
60
-------
VI
I
0)
o
I
!
2
O
1
9)
•C
I
o
o
i
2
00
.3
LO
CSJ CSJ
-S
(1)
c
8
-c
o
C C
61
-------
^5
Q)
.C
**
o
^
0>
CN
1
5
^
Q>
O)
c
^
to
5
o
II
>
^
XJ-
00
2?
03
1
ss
o
•^
-------
Figure 31. p-Dichlorobenzene: Estimated frequency distributions of personal
air exposures, outdoor air concentrations, and exhaled breath values
for the target population of 330,000 residents in the South Bay
section of Los Angeles. All air values are 10-14 hr integrated
samples. The breath values were taken following the daytime air
sample (6:00 am-6:00 pm). All outdoor air samples were taken in
the vicinity of the participants' homes. (May 1984)
Population (000) Exceeding Concentration Shown
300 100 W J
1,000
c
•
c
10r
<3
1 5 20406080 95 99
Cumulative Frequency, percent
Personal Air IN=50)
Breath (N=50)
Outdoor Air (N=25)
time history of air concentrations over the 12-hour monitoring period, high
correlation coefficients related to a single 12-hour integrated concentration
were not expected. However, the fact that significant correlations of breath
values with previous exposures in air or water were observed for every one
of the eleven prevalent chemicals in the first and largest of the field trips
suggests that breath measurements may be capable of providing rough
estimates of preceding exposures.
These correlations continued to be significant for some chemicals (xylenes,
trichloroethylene, 1,1,1-trichloroethane, tetrachloroethylene, and p-
dichlorobenzene) at most or all TEAM Study sites.
In California, correlations between breath concentrations and preceding
personal air exposures were again significant for many chemicals (Table
31) although the magnitudes were not large. Correlations with outdoor air
concentrations were almost never significant. In drinking water only
chloroform showed occasional significant correlations with breath
concentrations.
Intramedium Correlations. Spearman rank correlations were calculated
for all possible pairs of the prevalent target chemicals for the New Jersey
personal air, outdoor air, and breath samples. Correlations were high for
certain chemicals in all media. For example, the xylene isomers and
63
-------
Figure 32. Benzene: Estimated frequency distributions of personal air
exposures, outdoor air concentrations, and breath values for 91.000
residents of Antioch and Pittsburg. California. All air values are
10-14 hr integrated samples. Breath values were taken following
the daytime air sample (6:00 am-6:00 pm). All outdoor air samples
were taken in the vicinity of the participants' homes. (June 1984)
Population (000) Exceeding Concentration Shown
50 20105 2 1
100 r
10
c
Q>
U
c
o
o
0.1
1 5 20 4060 80 95 99
Cumulative Frequency, percent
'—• Personal Air (N=70)
, • Breath (N=67)
<•---•• Outdoor Air (N=10)
ethylbenzene had correlation coefficients exceeding 0.9 in virtually all cases
(Table 32). On the other hand, chloroform and p-dichlorobenzene showed
little correlation with any of the other chemicals or with each other.
Statistical Analysis of Questionnaire Data
Two questionnaires were administered to each participant. The household
questionnaire included questions on age, sex, occupations, household
characteristics, and customary activities of the participant and also of other
members of the household. The 24-hour recall questionnaire, administered
immediately following the end of the 24-hour monitoring period, included
questions on the participant's activities. Information on more than 100 items
were collected for each person. Of these, about 60 items were selected
for statistical analysis (Table 33). Two approaches were adopted: pairwide
comparisons (t-tests) followed by stepwise regressions. The logarithms of
the chemical concentrations were used in both approaches because of the
approximately log-normal distributions observed for all chemicals in air and
breath.
Pairwise Comparisons (t-tests). The 60 questionnaire items were
examined for possible associations with increased exposure to each of 1 2
chemicals in New Jersey and 16 in California. All three measures of personal
exposures (daytime air, overnight air, and breath) were examined in each
of the three New Jersey and three California visits. For example, in the
64
-------
Figure 33.
p-Dichorobenzene: Estimated frequency distributions of personal
air exposures, outdoor air concentrations, and breath values for
91.000 residents of Antioch and Pittsburg. California. All air values
are 10-14-hour integrated samples. Breath values were taken
following the daytime air sample (6:00 am - 6:00 pm). All outdoor
air samples were taken in the vicinity of the participants' home.
Population (000) Exceeding Concentration Shown
50 20/0527
10
I
o
O
0.1
Night
Day
•—• Personal Air (N = 70) /
- - Breath (N - 67)
••—• Outdoor Air (N= 1
-* Night
Day
1 5 20 4060 80 95 99
Cumulative Frequency, percent
Figure 34.
Octane, Decane, Undecane, and Dodecane: Estimated frequency
distributions of overnight concentrations in participants' homes
compared to overnight outdoor air concentrations. (June 19884)
Population (000) Exceeding Concentration Shown
50 2010521
1OO c , ,—,-T—i—i
10
c
o
1 1
c
4)
0.1
Decane
Personal Air (N = 70)
> Outdoor Air (N = 10)
Decane .
Octane
Undecane
Dodecane
Decane
Undecane
Dodecane
Octane
Octane
','Undecane
.-'/' Dodecane -
1 5 204060 80 95 99
Cumulative Frequency, percent
65
-------
Table 30. Spearman Correlations Between Breath Concentrations and
Preceding Daytime 12-Hour Personal Exposures to Eleven
Compounds in New Jersey, North Carolina, and
North Dakota
Compound
NJ1a NJ2b NJ3° NDd
(N=330) (N=130> (N=47) (N=23) (N=23)
Chloroform
1, 1, 1-Trichloroethane
Benzene
Carbon tetrachloride
Trichloroethylene
Tetrachloroethylene
Styrene
m,p-Dichlorobenzene
Ethylbenzene
o-Xylene
m,p-Xy/ene
.07
.28*
.21*
.24*
.38*
.46*
.19*
.54*
.33*
.26*
.32*
-.11
.28*
_f
-.01
.10
.23*
.20*
.38*
.22*
.22*
.27*
-.03
.32*
—
—
.35*
.37*
.19
.61*
.44*
.45*
.48*
-.01
.71*
—
-.23
.26
.53*
-
.63*
.12
.21
.19
.45*
-
.22
-.53*
.38
.58*
.32
.68*
-.01
.28
.08
a Fall 1981.
b Summer 1982.
c Winter 1983.
d Fall 1982.
e Spring 1982.
f Data uncertain based on quality assurance results.
* Significant at p < .05 level.
first New Jersey visit (Fall 1981) 11 questionnaire items had one or more
associations significant at p < 0.0001, and an additional six variables had
one or more associations significant at p < 0.001 (Table 34). These 17
variables accounted for a total of 47 t-tests significant at p < 0.001, compared
to only two expected to occur by chance at that level. Chemicals appearing
most often were three aromatic compounds: ethylbenzene (12 times), m.p-
xylene (9), and o-xylene (8). Chemicals never appearing were chloroform
and carbon tetrachloride.
Of the 60-70 questionnaire variables, about half appeared to have
considerable influence on personal exposure to one or more of the target
chemicals. These are ranked in order of the number of significant associations
observed during the six visits to New Jersey and California (Table 35). As
can be seen, variables related to smoking, occupation, home characteristics,
activities, and automobile travel were the most important determinants of
exposure.
Exposure to Active Smokers. The breath concentrations of all prevalent
chemicals were compared for smokers and nonsmokers (Table 36). Since
the distributions were skewed to the right, significance tests were performed
using the logarithms of the concentrations. Five aromatic chemicals (and
also octane, measured only in California) were significantly higher in the
breath of persons who had smoked tobacco the day they were monitored;
Six chlorinated compounds and three other straight-chain hydrocarbons
66
-------
Table 31. Spearman Correlations Between Breath and Preceding Air
Concentrations (Measurable Amounts Only)
Breath vs. Daytime
Personal Air
Compound
Trichloroethylene
rr\,p-Dichlorobenzene
Tetrachloroethylene
1, 1, 1-Trichloroethane
Ethylbenzene
o-Xylene
m,p-Xylene
Benzene
Styrene
n- Octane
n-Decane
n-Undecane
n-Dodecane
Chloroform
Carbon tetrachloride
a-Pinene
LA1a
(N=11-
112)
0.74*
0.71*
0.32*
0.57*
0.31*
0.39*
0.42*
0.25*
0.31*
0.31*
0.22
0.10
0.23
-0.06
-0.32
0.21*
LA2b
(13-49)
0.84*
0.40*
0.36*
0.62*
0.45*
0.51*
0.44*
0.25
0.12
0.38*
0.63*
0.34
0.66
0.17
NC
0.10
cc°
(10-58)
0.72*
0.46*
0.44*
0.11
0.13
0.03
0.16
0.07
0.06
0.25
0.01
0.09
NC
NC
0.05
0.10
Breath vs. Daytime
Outdoor Air
LA1a
(N=8-
24)
-0.05-
0.54*
0.11
-0.17
-0.12
0.14
0.02
-0.04
0.08
-0.07
-0.09
0.22
0.33
NC
NC
-0.15
LA2b
(7-24)
NCd
0.60*
-0.09
0.19
0.29
-0.22
0.14
0.11
0.23
0.53
NC
0.56
NC
NC
NC
-0.15
CC0
(7)
NC
NC
NC
NC
NC
NC
-0.29
NC
NC
NC
NC
NC
NC
NC
NC
NC
a Los Angeles—First trip—February 1984.
b Los Angeles-Second trip-May 1984.
c Contra Costa (Antioch/Pittsburg) — June 1984.
d Not calculated—N < 5.
*Significant at p < 0.05.
showed no consistent differences. The magnitude of the increase was
considerable—smokers had 2-10 times higher geometric mean concentra-
tions of benzene, styrene, ethylbenzene, and xylenes in their breath than
nonsmokers. Benzene concentrations in the breath of smokers and
nonsmokers in the fall visit to New Jersey are compared in Figure 35.
Exposure in Homes of Smokers. Overnight indoor air concentrations in
homes with smokers were compared to concentrations in homes with no
smokers for all six visits (Table 37). The fall 1981 visit to New Jersey (Figure
36) and the winter 1984 visit to Los Angeles showed significant increases
ranging from 50-100% for all five aromatics in the indoor air of homes with
smokers; however, the spring and early summer visits to Los Angeles and
Antioch/Pittsburg, California and the summer and winter visits to New Jersey
showed no difference. It was not possible to determine from the questionnaire
whether the homes with resident smokers actually experienced smoking
during the 12-hour overnight period (which included the sleeping period).
Such homes would be misclassified as smoking homes, which would tend
to obscure differences. Similarly, homes classified as nonsmoking may have
had a smoking guest on the day of monitoring. Therefore, the true increases
67
-------
^
«
£•
"»
g
^
^
•^
^
t/5
^
S
2
K.
^
W
5
•o
^
.c
•8
C
§
o
1
o
u
c
1
d
cS
« *~
Is
« c
OQ ^^
S
"> •>
Q P^
/\\ °"
Correlations
Summer 1
»..*
C 00
^
r
CN o to
co to oo
II II II
^^^
CN r-~ — -
05 6 10
CN CM <~
ii n n
^ $. ^
ij >-; I;
ij S S
CO
^ °^
(-O
U-05
to
CO Co
-> CN
> 05
CN
U. 05
^-.
CO 05
£*
u-c?
CM
CO 05
$05
U-05
^^
CO 05
to
^ 05
**.
U-05
„ O
CO 05
C
-$
6
4)
,
CL
E
^
00
^4.
0)
to
05
05
co
^
05
to
05
00
to
05
00
05
05
00
05
to
00
CM
05
00
05
05
00
0)
C
-S
O
0}
1
"^
f
k.
CM
r^
05
00 IX
05 10
O
05
rx ix
05 00
CM 05
05 tO
^J- ^^
05 00
CO O
05 to
to o
05 to
^ to
05 to
to to
05 to
CM to
05 to
to co
CO to
to ^f
05 rx
05 10
rx CM
05 ix
,p-Xylene
'hy/benzene
E .£
Q)
c-
N
1 1
^- 2
-c £
Uj CO
00 SJ-
rx
05
•^~
to
CN
rx
to
to
rx
to
0
to
CN
to
O
to
,^
^
05
to
^
to
c
~S>
o
C
2?
:x
CO
to
CM
10
CO
CO
t to
00
rx
00
10
CO 0
to to
to
to
00
to
to rx
to to
CO
10
CO
to
CM
to
o
rx
g
Q) 0)
^< -S
E 6
m
c 1
2j N
CO OQ
to ix
to rx
to to
co io
to to
10 to
10 to
O CM
to to
CO 00
io to
,p-Xy/ene
fhy/benzene
E tjj
0) 0)
c c
0) 0)
N N
C C
D Q)
OQ QQ
00 05
05
to
Q
to
CM
CO
to oo
io to
*~.
IX
CO
10
CM
10
to
tyrene
fhy/benzene
CO LU
0)
S
^
-c
01
. i
c -c
0} O
N to
£ 0)
OQ K
O *•
*-^ v^.
to to
05 CM
to io
05 05
Co rx
O O
to to
05 CM
to to
to rx
to to
o
to
0)
Q) 0}
E o
d) Q)
C C
0) .£>
v^ ^
-c -c
0> 0)
S S
o o
•c -c
0 0
fc t:
£ ^
CM o^
M~
to
to CM
to to
Q
to
to
to
<* to
>o to
to
c
at
tyrene
ichloroethyh
CO ft
0) 01
S §
^s ^s
-c -c
Q) Q)
1 1
•S -5
m TO
•fc i;
is !$
•^ to
68
-------
10
o
I
O
CM o
IX Co
CO CO
0
is
Q
°? *T
CN 6
c
o
z
£
°? *~ ^f
CNI o CQ
CQ Lf) C^
Compou
CO CT)
O) 6
CN CN
(S
-------
Table 33. Variables Included in Statistical Analysis
Variable Name
Personal Characteristics
MALE
NONWHITE
CHILD
YOUTH
ADULT
OLD
WEIGHTY
Occupation
EMPLOYED
JPAINTER
JGARAGE
JHOSPIT
JMETAL
JDRIVER
HPAINTER
HCHEMIC
HGARAGE
HMETAL
HDRIVER
Activities
HOBFURN
HOBPAIN
HOBMOD
HOBGAR
HOBFURNO
HOBPAINO
HOBMODO
HOBGARO
PEST
NUM_PEST
PUMPGAS
PEST24
SMOKED24
HOMESMOK
Na
183
107
23
60
148
40
33
203
9
12
6
7
15
10
14
9
21
11
15
51
7
68
37
15
7
68
47
162
9
20
161
258
Description
Hispanic (44) + Black (60) + Other (3)
Age 5-17
Age 18-39
Age 40-65
Age 65-85
Weighs over 200 Ibs
Occupation: painter
Occupation: garage/service station
Occupation: hospital worker
Occupation: metal worker
Occupation: taxi/bus/truck driver
Painter in household
Chemical worker in household
Garage/service station worker in household
Metal worker in household
Taxi/bus/truck driver in household
Hobby: furniture refinishing
Hobby: painting
Hobby: scale models
Hobby: gardening
Other household members' hobby: furniture
refinishing
OHMH: painting
OHMH: scale models
OHMH: gardening
Often use pesticides
House is treated regularly for pesticides
Pumped gas on day of monitoring
Exposed to pesticides that day
Smoked that day
Smoker in household
Household Characteristics
OLDHOUSE 121
CENT_J\_C 14
WIND_A_C 267
FAN_OUT 113
CIRCFAN 72
ELECSTOV 17
GASHEAT 161
Occupation-Related
XPAINT24 27
XDRYCI24 13
XCHEM24 21
XPETRO24 9
XGARAG24 67
XFURN24 7
House older than 10 years
Central air conditioning
Window air conditioner
Window fan or ceiling exhaust fan
Circulating fan
Electric stove
Gas furnace
Worked at or in on day of monitoring:
Paint store
Dry cleaners
Chemical plant
Petroleum plant
Garage/service station
Furniture refinishing shop
70
-------
Table 33. (continued)
Variable Name Na Description
XPLAS24
XTEXT24
XWOOD24
XPRINT24
XLAB24
XDYE24
XHPSP24
XMETAL24
XNONE24
11
5
6
9
14
4
13
17
124
Plastics plant
Textiles plant
Wood processing plant
Printing shop
Scientific laboratory
Dye plant
Hospital
Metal work
None of the above
Activity/Occupation-Related Exposed to on day of monitoring:
XSOLV24 37 Solvents
XODOR24 83 Odorous chemicals
XPEST24 27 Pesticides
XDUST24 63 High dust levels
XEXHAU24 62 Auto/truck exhaust
XCLEAN24 94 Household cleaners
XGREAS24 19 Degreasing chemicals
XOTHER24 19 Other chemicals or mixtures
a Number of persons in category during first New Jersey visit (Total
number of respondents: 362)
Table 34. Questionnaire Items Associated with Significantly Increased
Exposures (p< 0.001): New Jersey, Fall 1981
Geometric Means fag/m3)
Breath Personal Air
Questionnaire Item/ Day Night
Chemical Yes No Yes No Yes No
Employed (N= 188-194)
1,1,1-Trichloroethane 6.7 3.5 31 12
Tetrachloroethylene 9.7 5.5
Ethylbenzene 3.4 1.9 12 6.1
o-Xylene 2.6 1.6 8.9 5.3
m,p-Xylene 7.1 4.5 29 15
Smoked (N= 144-154)
Benzene 21 5.3 18 11
Styrene 1.3 0.6
Ethylbenzene 3.9 2.0
m.p-Xylene 7.9 4.5
Smoker in Home (N= 223-245)
Benzene 13 4.9 14 8.1
Styrene 1.0 0.5 2.1 1.1
71
-------
Table 34. (continued)
Geometric Means
Breath Personal Air
Day Night
Yes No Yes No Yes No
Ethylbenzene 7.7 5.2
rr\,p-Xylene 19 12
High Potential Exposure*
(N=238)
Tetrachloroethylene 8.8 5.7
Ethylbenzene 10.7 5.9
o-Xylene 8.7 4.9
m,p-Xylene 26 15
Male (N= 172-173)
Ethylbenzene 11 6.9
o-Xy/ene 8.9 5.7
m,p-Xylene 29 17
Service Station/Garage
Worker (N=11)
Ethylbenzene 55 8.3
m,p-Xylene 132 21
Hospital Worker
-------
Table 34. (continued)
Geometric Means
Breath Personal Air
Day Night
Yes No Yes No Yes No
Dye Plant (N=4)
Ethylbenzene 42 8.6
Solvents (N=33)
Ethylbenzene 6.2 2.4
o-Xylene 4.2 2.0
m,p-Xylene 12.1 5.2
Odorous Chemicals IN =78)
Ethylbenzene 15 7.5
o-Xylene 12 6.1
High Dust/Paniculate Exp.
(N=56)
Ethylbenzene 18 7.6
m,p-Xylene 44 19
* All those who were employed in or exposed to at least one of the 14
listed occupations/activities on the day of monitoring; the inverse of the
XNONE24 variable (See previous table/.
in homes that experienced smoking on the day of monitoring may exceed
the values in the table.
Occupational Exposure. About 85 of the 350 participants were classified
as having potential occupational exposures to some of the target compounds
Certain occupations showed significant (p < .05, Mann-Whitney nonpara-
metric test) increases in breath concentrations or personal air exposures
to some chemicals, whereas other occupations showed no increased
exposures. Figures 37 to 41 compare unweighted median breath values for
workers in several occupations (chemicals, paint, plastics, petroleum, and
printing) to persons not engaged in those occupations. In these pairwise
comparisons, no attempt is made to control for confounding factors; however,
stepwise regressions (see below) confirmed most of the pairwise results.
Effects of Activities and Potential Sources on Exposures
All participants were asked if they had been exposed to potential sources
of target chemicals on the day they were monitored or within the previous
week. Sources included industrial plants, auto exhaust, and paint. For ten
of the twelve sources, at least one (and as many as six) of the eleven most
prevalent chemicals appeared at significantly higher levels in the breath of
persons exposed during the day or week compared to those not exposed
to the source. In most cases, the chemicals that were elevated were those
expected to be associated with a given source, such as tetrachloroethylene
with dry cleaners (Figure 42) and benzene with service stations (Figure 43)
73
-------
Table 35. Variables Ranked by Number of Pairwise Associations with
Significantly Increased or Decreased Exposures (p < 0.05)
(All New Jersey and California Visits)
Variable
1. Employed
2. Adult (21-65)
3. Student
4. Smoke free
5. Smoked
6. Smoker in home
7. Never smoked
8. High potential exposure (24-hr)
9. Exposed to so/vents (24-hr)
10. Hispanic
1 1. Exposed to dust/particles (24-hr)
12, Circulating fan
13. Hazardous job worker in home
14. Hazardous job
15. Old O 65)
16. Fan in window/ceiling exhaust
1 7. Child « 12)
18. Exposed to solvents (wk)
19. High potential exposure (wk)
20. Visited garage/service station (wk)
21. Exposed to degreasers
22. Exposed to tobacco smoke
23. Pumped gasoline
24. Gardened
25. Exposed to auto exhaust
26. Exposed to odorous chemicals
27. Youth (12-20)
28. Gas heat
29. Auto exhaust (wk)
30. Visited garage/service station (24-hr)
3 1 . Prin ting shop
32. Pesticide Exposure
General
Category
Occupation
Age/Occup.
Age/Occup.
Smoking
Smoking
Smoking
Smoking
Occupation
Occupation
Race
Occupation
Home
Occupa tion/home
Occupation
Age
Home
Age
Occupation
Occupation
Auto
Occupation
Smoking
Auto
Hobby
Auto
Occupation
Age
Home
Auto
Auto
Occupation
Activity
Na
(p < 0.05)
76
65
62
62
61
58
58
40
40
40
38
38
38
37
36
36
35
32
30
29
29
28
27
26
24
24
20
20
17
17
12
12
a The maximum possible number of significant associations is 243:
11 chemicals x 3 media x 3 New Jersey trips + 16 chemicals x 3 media
x 3 California trips
74
-------
I
I
!
* CO
*Q> ^
CO
U
0
Ei
*QJ CO
c <
5
CO
CO
I
0)
CQ CO
^ IN CM
1 ci ci ci
O Oq to
I ^^ *** Q
Oo 05 to IN
10
O ^ Ci Ci O
CM to 10 to o
CM
o 'o *- to 10 co
CM o *~ ci c! ci
oi o co M- CM o
co CM CM CM co CM
to co CM CM *^ c\
ci ci ci ci ci o
co O to oo O)»~
CO tt Qc ^ <0 00
10 5: ^ csj »-: o
CM
ct "*> "
> *^
DO (O Co IO O
oo IN CM oo ^j-
O g) Cb
CO
O
O
CO
Q) '.
"S,
CO
01 (
Cb
C \
U) .
N
I" CO
g*
I
% CO
CO
00 CO
CO
Q)
1
\ \ \^*-*
IN 00 CO
CM
*- CO *- IN *— [N
t^. t^ C\ T^ t^
QO ic CM i
'•O Is* IN ^ Q O
0) 'Sh IT) •
co Ci O
03
cr ct
Oo Oj
- CM co CD co
- ^- ID CM co
CM »~ IN Q.
U "3 £
II
(U o w)
t ^ *
O Ct) U
Q. Q) C
S S S
t. *. ,9)
P ti *:
75
-------
Figure 35. Unweighted cumulative frequency distributions of benzene
concentrations in the breath of current smokers vs. non-smokers
(New Jersey. Fall 1981).
Percent Exceeding Concentration Shown
90 50 JO 1
1 1
200
100
I
c
o
50
§
o
O
g 20
!
01
03
I
5-
99
—r
T
o Smokers (N - 150)
Q Non-Smokers (N= 151)
200
- 100
!
.i
c
01
u
c
o
o
-------
Figure 36. Unweighted cumulative frequency distributions of benzene
concentrations in the air in homes with at least one smoker vs.
homes with no smokers (New Jersey, Fall 1981).
Percent Exceeding Concentration Shown
75 50 25 10 5 2 7 0.5
500
500
200
§> 700
§
c
o
O
0)
c
1
I
I
50
20
10
o Smokers (N = 248)
o Non-Smokers (N = 94)
50 75 90 95 98 99 99.5
Cumulative Frequency, Percent
exposed to that source are listed in Table 39. A complete set of comparisons
of breath and personal air concentrations is presented in Appendix Y of Vol
II.
Since many chemicals have multiple sources, some members of the so-
called "unexposed" groups in the above analyses may have been exposed
to the same chemical through a different source, thus blurring the distinction
between exposed and unexposed groups. Therefore, the breath and personal
air levels of groups exposed to each source were compared to the group
of persons who responded that they were not exposed to any source. As
could be expected, the number of chemicals showing significant differences
increased considerably. The number showing simultaneously elevated air
and breath values doubled (Table 40).
Caution in interpreting these results is indicated because of the small
numbers of persons in some of the exposed groups and the possibility of
confounding variables (such as smoking, which may be more prevalent in
77
-------
Figure 37. Median breath concentrations of 21 chemical plant workers vs. 330
other participants (NJ. Fall, 1981). Asterisks indicate significant
(p < .051 differences using Mann-Whitney nonparametric test.
Breath Values for Those Who Had Worked At or Been in a
Chemical Plant During the Past 24 Hours Versus
Those Who Had Not Worked at or Been in a Chemical Plant
Asterisk Indicates
Statistically Significant
Difference Below .05
Legend
Exposed
Not Exposed
Figure 38. Median breath values for 28 paint plant workers vs. 320 other
participants (NJ, Fall. 1981). Asterisks indicate significant (p <. OS)
differences using Mann-Whitney nonparametric test.
Breath Values for Those Who Had Worked At or Been in a
Paint Plant During the Past 24 Hours Versus
Those Who Had Not Worked at or Been in a Paint Plant
35
30
25
1
10
5
0
Asterisk Indicates
Statistically Significant
Difference Below .05
•n^H
Legend
m Exposed
W777A Not Exposed
w
78
-------
Figure 39. Median breath values for 11 plastics manufacturing workers vs.
340 other participants (NJ, Fall, 1981). Asterisks indicate significant
(p < .05) differences using Mann-Whitney nonparametric test.
Breath Values for Those Who Had Worked At or Been in a
Plastics Plant During the Past 24 Hours Versus
Those Who Had Not Worked at or Been in a Plastics Plant
Asterisk Indicates
Statistically Significant
Difference Below .05
^^ Legend
•i Exposed
tsssssi Not Exposed
Figure 40. Median breath values for 19 petroleum plant workers vs. 330 other
participants (NJ. Fall. 1981). Asterisks indicate significant (p < .05)
differences using Mann-Whitney nonparametric test.
Breath Values for Those Who Had Worked At or Been in a
Petroleum Plant During the Past 24 Hours Versus
Those Who Had Not Worked at or Been in a Petroleum Plant
Asterisk Indicates
Statistically Significant
Difference Below .05
Legend
m Exposed
123 Not Exposed
-------
Figure 41. Median breath values for 9 printing plant workers vs. 340 other
participants (NJ. Fall, 1981). Asterisks indicate significant (p <. 05)
differences using Mann-Whitney nonparametric test.
Breath Values for Those Who Had Worked At or Been in a
Printing Plant During the Past 24 Hours Versus
Those Who Had Not Worked at or Been in a Printing Plant
Legend
mm Exposed
1=3 Not Exposed
Figure 42. Median breath values for 11 persons visiting dry cleaning shops
on the day they were sampled vs. 340 other participants (NJ, Fall,
1981). Asterisk indicates significantly (p < .05) higher exposure
to tetrachloroethylene (Mann-Whitney test).
Breath Values for Those Who Had Worked At or Been in
Dry Cleaners During the Past 24 Hours Versus
Those Who Had Not Worked at or Been in Dry Cleaners
30
20
f
.11
10 —
f
I
K >
Wf]
F
n
$
xi
i
-11
f g ^
Asterisk Indicates
Statistically Significant
Difference Below .05
„
:/^\ mi^\ mt^\ m/\ mK\ _,., ^m
" # //
-------
Figure 43. Median breath values for 67 persons visiting a service station the
day they were sampled vs. 270 other participants (NJ. Fall. 1981).
Asterisk indicates significantly (p < .05) higher levels of benzene
(Mann-Whitney test).
Breath Values for Those Who Had Worked At or Been in a
Service Station During the Past 24 Hours Versus
Those Who Had Not Worked at or Been in a Service Station
30
25
-
.
w
Asterisk Indicates
Statistically Significant
Difference Below .05
Legend
^ Exposed
<=^ Not Exposed
Figure 44. Median breath values for 62 persons exposed to automobile or truck
exhaust on the day they were sampled vs. —280 other participants
(NJ. Fall, 1981). Asterisk indicates significantly (p < .05) higher
levels of benzene (Mann-Whitney test).
Breath Values for Those Exposed to Exhaust in the Past 24 Hours
Versus Those Not Exposed to Exhaust in the Past 24 Hours
Median (fjg/m3)
-> -« hj N
3 Ol O Oi O I
^
1
"
'2
''/
*
Asterisk Indicates
Statistically Significant
Difference Below .05
t]
fell m^m^fl —
ttllll .m
Legend
K Exposed
a Not Exposed
////////*//
/
$
'
-------
figure 45. Median breath concentrations of 150 smokers compared to 150
nonsmokers (NJ. Fall, 19811. Benzene and other aromatic
compounds were elevated..
Breath Values for Those Using Tobacco in the Past 24 Hours
Versus Those Not Using Tobacco in the Past 24 Hours
Legend
M Exposed
1221 Not Exposed
Figure 46. Median breath concentrations for 20 persons using pesticides vs.
330 other participants (NJ. Fall, 1981). No compounds were
signigicantly different.
Breath Values for Those Using Pesticides in the Past 24 Hours
Versus Those Not Using Pesticides in the Past 24 Hours
30
25
. 20
6
15
10
Legend
•• Exposed
1=1 /Vof Exposed
82
-------
c
S
g
1
«l
(0
C
1
s»
O
1
i
00
•5
•e
]§,
$
io
V
o.
Hicantly \
nitored
C 0
•2*5
»
^i
1 x
o> »
-x «
ft
II
C 1
13
Je/j
<
I
"""'
'•-
-c
Ol
^
0)
-
0
-Q
to
*>
^
00
»J
oi
05
»~:
oo
K
00
to
to
o
r< 06 K io ui ^
* ^
#i^* * ^ #_^
loiofvcMO CMfvco CM OOQCM
CMCMCNCNOO OOCNCN CN CMCMOO
CM oo *^ *••• CM io **• *^ co oj) to to *^ *** *^ CN *^ oo
to
c I *
J •§ $ -S to
^ E 1 i i i i
IB S -c -ft 5- s
•N .8 I^^SS^I
t-t2*-O'^cv^CC^c«:i4^v's.
-C-S'SjS'g 3 * G^ C
O
-------
oo
K
-------
Table 39. Chemicals Showing Significantly (p < .05) Higher
Concentrations in Air and Breath of Persons Recently
Exposed to Potential Sources Compared to Persons
not Exposed to That Source
No. of Persons
Potential Source Exposed
Dry Cleaners 37
Tetrachloroethylene
Paint 28
Styrene
Ethylbenzene
o-Xylene
m,p-Xylene
Auto Exhaust 62
None
Tobacco Smokers 161
Styrene
Chemical Plant 21
Styrene
Ethylbenzene
m,p-Xylene
Pesticides 20
None
Furniture Refinishing 7
None
Printing Shop 9
None
Petroleum Plant 19
None
Ratio of Median
Concentrations
Breath Air
2.8 2.0 (.01 f
2.6 1.6 (.002)
1.8 1.8 (.0009)
1.4 1.9 (.006)
1.8 2.1 (.0003)
1.4 1.4 (.0002)
1.8 1.8 (.02)
2.3 1.5 (.008)
1.8 1.61.01)
Science Laboratory
None
Service Station
Benzene
14
67
1.9
1.2 (.03)
85
-------
Table 39. (continued)
Ratio of Median
Concentrations
No. of Persons
Potential Source Exposed
Plastics Manufacturing 1 1
Styrene
Hospital 13
None
Solvents 37
Styrene
Ethylbenzene
o-Xylene
m,p-Xylene
Odorus Chemicals 83
Tetrachloroethylene
Benzene
Ethylbenzene
o-Xylene
m,p-Xylene
Degreasing Compounds 19
None
Dust 63
m,p-Xylene
Tobacco Smoke
(non-smokers only) 99
None
Breath
2.0
1.7
1.9
1.7
2.0
1.1
1.3
1.2
1.2
1.2
1.1
Air
2.4 1.04)
1.5 1.03)
1.5 (.01)
2.2 (.002)
1.5 (.005)
1.2 (.02)
1.5 (.003)
1.6 (.0001)
1.8 (.0000)
1.5 (.0001)
1.2 (.002)
Cleaning Solutions
None
Toxic Chemicals
None
94
27
"Probability that the ratio is due to chance (Wilcoxon rank-sum test).
86
-------
Table 40. Chemicals with Significantly (p < .05) Higher
Concentrations in Air and Breath of Persons Recently
Exposed to Potential Sources Compared to Persons
Not Exposed to Any Source
Ratio of Median
Concert tra tions:
Exposed vs Unexposed
Groups
Potential Source
Paint
Benzene
Tetrachloroethylene
Styrene
Ethylbenzene
o-Xylene
m,p-Xylene
Chemical Plant
Styrene
Ethylbenzene
o-Xylene
m,p-Xylene
Plastics Manufacturing
Styrene
Ethylbenzene
o-Xylene
m,p-Xy/ene
Dry Cleaning
Tetrachloroethylene
Benzene
Petroleum Plant
None
Service Station
Benzene
Printing
Ethylbenzene
o-Xylene
No. of Persons
Exposed Breath
28
2.3 I.0002)3
2.0 (.0000)
2.8 (.0004)
1.9 (.0004)
7.4 (.009)
1.7 (.002)
21
1.9 (.02)
2.5 (.0008)
1.4 (.05)
1.9 (.004)
11
2.0 (.01)
2.8 (.003)
3.4 (.0006)
2.5 (.001)
37
2.3 (.0000)
2.2 (.02)
19
67
2.2 (.0000)
9
1.8 (.02)
1.3 (.03)
Air
1.3 (.03)
2.7 (.02)
1.8 (.0005)
2.1 (.0001)
2.5 (.0003)
2.5 (.0000)
2.0 (.004)
1.8 (.0006)
2.3 (.0003)
1.9 (.0006)
2.6 (.02)
1.8 (.03)
2.3 (.02)
2.1 (.02)
2.2 (.003)
1.7 (.03)
1.3 (.02)
1.6 (.03)
2.2 (.02)
87
-------
Table 40. (continued)
Ratio of Median
Concentra tions:
Exposed vs Unexposed
Groups
Potential Source
Metal Working
Tetrachloroethylene
Ethylbenzene
o-Xylene
No. of Persons
Exposed
17
Breath
1.4 (.01)
1.8 (.05)
1.8 1.05)
Air
1.8 (.03)
3.7 (.0000)
4.4 (.OOOO)
Science Laboratory
Ethylbenzene
o-Xylene
Furniture Refinishing
Ethylbenzene
o-Xylene
Hospital
None
14
1.7 (.03)
1.4 (.05)
2.8 (.03)
2.5 (.04)
2.2 (.002)
2.7 (.001)
2.2 (.02)
2.4 (.006)
13
aProbability of no difference between exposed and unexposed groups
Wilcoxon Rank-Sum Test.
occupationally exposed groups). To account for such confounding variables,
a set of stepwise regressions were performed.
Stepwise Regressions. Stepwise regressions were performed using the
model:
y = a + Ib,q,
where y = In concentration
q, = questionnaire index variable
Because of the large number of variables on the two questionnaires, an
extensive investigation of collinearity was carried out. The methods of Belsley,
Kuh, and Welsh (1) were employed to identify collinearities. In most cases,
it was possible to reduce collinearities without eliminating questions or
otherwise losing data. The final matrix of variates and eigenvalues seldom
included variables associated with a condition number higher than 20. (The
threshold value for seriously degraded estimates is considered by Belsley
et al. to be about 30.)
The SAS (Statistical Analysis System) STEPWISE procedure (combined
forward and backward selection) was employed with criteria of p < 0.15
for inclusion. The final model included only variables for which p < 0.05.
The results of the stepwise regressions of all six New Jersey and California
sites are presented in Appendix A.
Three major sources of increased exposures were identified. Smoking,
employment, and auto-related activities were all significantly related to
88
-------
increased exposures to many of the 11 prevalent chemicals in New Jersey
and the 1 6 in California.
Smoking was responsible for greatly elevated breath concentrations of
benzene and styrene, and significantly elevated breath concentrations of
ethylbenzene, xylenes, and octane Table 41 summarizes the effects of
smoking on breath concentrations of smokers during all six trips. Benzene
concentrations in smokers' breath increased six-fold, styrene four-fold, and
four other compounds more than doubled compared to nonsmokers' breath
concentrations. Having a smoker in the home resulted in increased overnight
personal air exposures to the same group of hydrocarbons during the fall
season in New Jersey and the winter season in California.
Employment in many occupations was associated with increased exposures
to one or more of the chemicals. Self-reported exposures to solvents, odorous
chemicals, dust and particulates, degreasers, and other mixtures were
repeatedly associated with increased exposures to the target chemicals.
Auto-related activities (driving, pumping gas, visiting service stations) were
associated with increased exposures to many aromatics and straight-chain
hydrocarbons in all California trips.
Other important variables included age, race, and sex. Adults showed
consistently elevated exposures, while children and old people showed
depressed exposures. Occasionally Hispanics showed elevated exposures.
Males often had higher exposures to aromatics, but females sometimes
showed higher exposures to trichloroethylene.
Household characteristics were sometimes associated with increased
indoor air levels. In fall and winter, homes with gas furnaces often were
associated with increased overnight indoor air concentrations of the aromatics
compared to homes with oil furnaces. Ventilation characteristics, however,
showed inconsistent effects. Window air conditioners and circulating fans
were usually associated with increased indoor concentrations, as might be
expected if their use leads to decreased outdoor ventilation, but sometimes
circulating fans were associated with reduced exposures.
Certain variables were associated with increased exposures to one chemical
only One example is visiting a dry cleaners (tetrachloroethylene).
Table 41. Effects of Smoking on Breath Concentrations of Benzene
and Other Hydrocarbons
Compound
Benzene
Styrene
Ethylbenzene
m,p-Xylene
o-Xylene
Octane
New Jersey
Fall Summer Winter
1.38*
0.81
0.56
0.53
NSC
NMd
2.38
1.16
1.46
1.25
NS
NM
1.19
1.20
0.66
0.49
0,58
NM
California
Winter Spring Summer
1.
1.
1.
1.
85
56
37
02
0.82
0.
83
1.67
2.59
1.75
1.27
1.19
1.10
2.
1.
1.
1.
1.
1.
70
14
48
32
03
00
1.
1.
1.
0.
0.
Grand
Mean
86" ±
41 ±
2 ±
98 ±
90 ±
0.98 ±
0.58
0.63
0.49
0.38
0.26
0.14
a Coefficient of SMOKER variable in stepwise regression; thus smokers
had e' 38 ~ 4 times as much benzene in their breath as nonsmokers. All
listed coefficients were significant at p < 0.05.
b Arithmetic mean of all six trips, unweighted; thus on average, smokers
had e' 8B = 6.4 times as much benzene on their breath as nonsmokers.
0 Not significant.
d Not measured.
89
-------
Although the questionnaires were successful in identifying major sources
of exposure for some chemicals, they were unsuccessful for other chemicals.
For example, the sources of the elevated indoor air levels of chloroform,
/77,,0-dichlorobenzene, trichloroethylene, 1,1,1 -trichloroethane, carbon
tetrachloride, decane, undecane, dodecane, and cr-pmene were not
determined by the questionnaire.
Effect of Outdoor Concentrations on Exposures
Stepwise regressions were run to determine the effect of outdoor
concentrations on personal exposures of the New Jersey and California
subjects who had outdoor measurements in their backyards. A reduced set
of approximately 20 independent questionnaire variables was selected for
the New Jersey subjects (85 persons in the fall and 71 in the summer)
based on their frequencies and importance in previous stepwise regressions.
Because of the smaller number of persons in California with outdoor
measurements (25 in Los Angeles each season and 10 in Contra Costa),
only six questionnaire variables in Los Angeles and three in Contra Costa
were included in the regressions.
The model was of the form-
In Cm = a + bln C0ut+Ic,q,
where C,n = indoor concentration (or, for New Jersey only, breath
concentration or daytime personal air concentration)
Gout = outdoor concentration
q, = questionnaire variables (occupation, household character-
istics, etc.) generally indexed to 0 or 1
c, = coefficients of the q,
The natural logarithms of the concentration variables were employed because
their distributions are closer to being log-normal than normal.
The results (displayed in Appendix B) indicated that outdoor concentrations
were sometimes significantly associated with personal exposures to some
chemicals but seldom on a consistent basis. For example, in New Jersey
overnight indoor air levels of carbon tetrachloride and trichloroethylene were
significantly associated with outdoor air levels in both summer and winter;
but eight other chemicals showed no significant association in the fall and
three showed none in the summer. Breath levels were significantly associated
with daytime outdoor air levels of seven chemicals in the fall but none in
the summer. Daytime personal air exposures were significantly related to
daytime outdoor air concentrations of five chemicals. The observed slopes
of the log-log regressions usually lie between 0.2 and 0.4, indicating a weakly
positive relationship (Table 42). Partial R2 values for the significant
associations range from 0.03 to 0.35. Other important determinants of
personal exposure in these subsets were smoking, having a smoker in the
home, certain occupations (particularly those involving paints, solvents, and
odorous chemicals), and activities (particularly auto exhaust exposure and
visiting a dry cleaners or service station).
In California, only overnight personal air exposures were compared to
outdoor levels (Table 43). Only tetrachloroethylene showed a significant (p
< 0.10) dependence on outdoor levels on all three trips. Six of 16 chemicals
never displayed a significant association with outdoor levels in California
Discussion
Comparison of New Jersey and California Results
Quality Control. Considerable improvements were evident in comparing
field blanks collected in the first trip to Los Angeles with those collected
90
-------
Table 42. Effect of Outdoor Air Concentrations on Measures of
Personal Exposure 25%
of samples measurable), and 3 of 4 trihalomethanes were prevalent in drinking
water. In California, 26 chemicals were selected, including 9 that had not
been measured in New Jersey Of these, 19 were prevalent in air and breath,
and all 4 trihalomethanes were prevalent in drinking water. (The 19 prevalent
CA chemicals included all 11 of the prevalent New Jersey chemicals.)
91
-------
Table 43. Effect of Overnight Outdoor Air Concentrations on Indoor Air
Concentrations (CA): Coefficients of Stepwise Regressions
Chemical
Aromatic Hydrocarbons
Benzene
Styrene
Ethylbenzene
o-Xylene
m,p-Xy/ene
Chlorinated Hydrocarbons
Chloroform
1, 1, 1-Trichloroe thane
Carbon tetrachloride
Trichloroethylene
Tetrachloroethylene
m,p-Dichlorobenzene
Aliphatic Hydrocarbons
Decane
Dodecane
Octane
Undecane
a-Pinene
Los Angeles Antioch/Pittsburg
February May June
NSa
NS
0.1 9b
NS
NS
NS
0.38
NS
NS
0.23
0.98
0.42
NS
NS
O.26
NS
NS
NS
NS
NS
0.52
NS
0.71
NS
NS
0.51
NS
NS
NS
0.33
NS
-0.61
NS
NS
NS
NS
NS
NS
NS
NS
NS
0.39
NS
NS
NS
NS
NS
NS
aNot significant fp < 0.05^ in step wise regression.
^Coefficient of In (outdoor concentration). Thus ethylbenzene indoors =
afethy/benzene outdoors}019. (See also footnote to Table 42.)
Concentrations. For the concentrations in air and breath described above,
several observations are evident:
1. Exposures were highly variable. For many chemicals, the range in
personal air exposures exceeded a factor of 1000 or even 10,000. This
was far greater than for typical criteria pollutants such as carbon
monoxide and suspended particulates. The range in breath concen-
trations was almost equally variable, indicating that the higher
exposures may have been producing a higher body burden.
2. All eleven chemicals had higher personal air concentrations than
outdoor air concentrations. This is the case even for overnight
exposures, when participants were normally at home for the entire
12 hours.
3. Breath levels were also often higher than outdoor levels. Since levels
in exhaled breath are often only 20-40% of total intake, the remainder
being metabolized or excreted through other pathways, the breath levels
imply exposures several times greater. This is further indication that
92
-------
Table 44. Control and Blank Data for Tenax Cartridges Used in
New Jersey and California: TEAM Study
Field Controls
Recovery3 {%)
Target Compound
Chloroform
1,2-Dichloroethane
1, 1, 1-Trichloroethane
Benzene
Carbon tetrachloride
Bromodichloromethane
Trichloroethylene
p-Dioxane
Chlorodibromomethane
1,2-Dibromoethane
n-Octane
Tetrachloroethylene
Chlorobenzene
Ethylbenzene
Bromoform
p-Xylene
Styrene
o-Xylene
1,1,2,2-Tetrachloroethane
a-Pinene
p-Dichlorobenzene
n-Decane
o-Dichlorobenzene
r\-Undecane
n-Dodecane
NJ
-------
Table 45. Median Coefficients of Variation (%) for Duplicate
Personal Air Samples in New Jersey and California:
TEAM Study
Target Compound
Chloroform
1, 1, 1-Trichloroethane
Benzene
Carbon tetrachloride
Trichloroethylene
Tetrachloroethylene
Styrene
p-Dichlorobenzene
Ethylbenzene
o-Xylene
m,p-Xylene
n-Decane
n-Dodecane
1,4-Dioxane
n-Octane
r\-Undecane
a-Pinene
o-Dichlorobenzene
NJ>
20
27
36
24
14
21
18
23
20
19
24
NM°
NM
NM
NM
NM
NM
NM
CAb
28
8
13
15
12
12
28
30
13
13
15
12
13
20
11
15
13
12
aBayonne and Elizabeth, NJ, Fall 1981, N = 134.
"Los Angeles. CA, Winter 1984. N = 24.
CNM = not measured.
exposures are higher than would be expected from observed outdoor
concentrations.
The ratio of personal exposures to outdoor levels increased with higher
exposures. This can be illustrated by comparing indoor overnight
exposures (when persons were almost invariably inside their homes)
to outdoor overnight concentrations for the 75th percentile and the
99th percentile of each distribution. The ratios increased from 2-5 at
the 75th percentile up to 10-20 at the 99th percentile for most of
the target chemicals (Figures 47 and 48).
The higher overnight personal exposures appear to implicate the home
or personal activities within the home as the major source of exposure
to these eleven compounds. The daytime personal air exposures were
usually the highest, as expected since this time period included the
commuting and occupational activities. However, the overnight personal
air exposures, when people were normally sleeping, were nearly as
high. In fact, all eleven prevalent chemicals had much higher overnight
indoor concentrations than overnight outdoor concentrations,
sometimes 100 times higher for individual paired observations.
94
-------
Figure 47. Comparison of unweighted 75th percentile concentrations of 11
prevalent chemicals in overnight outdoor and personal air in New
Jersey (Fall 1981) with outdoor air measured in a number of U.S.
cities between 1970-1980 (Brodzinsky 1982).
Seventy-Fifth Percentile Values
Legend
aUS. Outdoor
taN.J. Outdoor
m N J. Personal
Figure 48. Comparison of unweighted 99th percentile concentrations of 11
prevalent chemicals in overnight outdoor air and overnight personal
air in New Jersey (Fall 1981).
Ninety-Ninth Percentile Values
30
! 20
O
10
F?H
^B E^B ^»
\Actual Ratio
Value is 70
Legend
N.J. Outdoor
N.J. Personal
**//
*
-------
6. The presence or absence of a source is a far stronger determinant
of indoor air concentrations than the air exchange rate. Although air
exchange rates were not measured, several studies indicate that the
range of rates is quite small (less than a factor of 10). Yet homes
often differed by a factor of 100 in concentration. The most likely reason
for such high concentrations is the presence of a powerful source.
7. Only chloroform and possibly bromodichloromethane were important
contributors to total exposure from drinking water in the study
areas. The median value for chloroform in drinking water in New
Jersey (Fall 1981) was 67 /ug/L; in air, 3.2 /ug/m3. Assuming 2L of
water intake per day and 20 cubic meters of air intake per day, the
median intake of chloroform in water (134 /L/g) was about twice that
in air (64 /JQ). (However, if the water was boiled for tea or coffee, it
would lose its chloroform—thus the water intake may be overestimated.)
Drinking water also accounted for most exposure to bromodichlorome-
thane, since the chemical was detected in only 3% of the personal
air samples.
8. Breath levels and persona/ air exposures to certain toxic and
carcinogenic chemicals are significantly elevated in persons exposed
to potential sources (consumer products, activities, and workplaces).
Indoor versus Outdoor Air Concentrations
Concentrations in overnight indoor and outdoor air are compared for New
Jersey and California in Tables 46 and 47 For indoor air, no obvious
differences between the two sites appear. However, for outdoor air, the
February overnight concentrations in Los Angeles stand out—six chemicals
(benzene, 1,1,1 -trichloroethane, tetrachloroethylene, p-xylene, o-xylene, and
ethylbenzene) exceeded the highest New Jersey values by a factor of 2 or
more, whether medians or 90th percentile concentrations are compared.
The May Los Angeles results are more comparable with the New Jersey
values. Once again personal air and indoor air concentrations were observed
to be higher than outdoor concentrations for nearly all chemicals. As in
New Jersey, maximum indoor concentrations usually far exceeded maximum
outdoor concentrations measured at the same homes (Table 48).
Personal exposures and concentrations were compared for persons who
were inside their homes for the entire overnight period, and for all but 20
minutes or less of the daytime monitoring period. For each of the six trips
to New Jersey and California, the median and mean indoor-outdoor
differences were calculated. Median differences (Table 49) were normally
positive (i.e., indoor levels were greater than outdoor levels), and usually
less than 5 fjg/m3. Mean differences (Table 50) were larger, often exceeding
10 Aig/m3.
The findings of higher indoor concentrations are paralleled by recent studies
in Europe and the U.S., some using different adsorbents than Tenax. Seven
other studies of volatile organics in ten or more homes have been reported
since 1979. Mtflhave (2) found elevated levels of benzene and toluene in
39 Danish dwellings. Jarke (3) found more complex chromatograms and
increased concentrations of organics in 34 Chicago homes Lebret (4) found
that all 35 organics analyzed displayed mean indoor/outdoor ratios exceeding
unity in 134 Dutch homes, with seven mean indoor/outdoor ratios exceeding
10. Tobacco smoking was correlated with increased levels of ten organics.
Factor analysis identified certain clusters of compounds as petroleum
distillate-based. Seifert (5) reported that 15 homes in Berlin displayed
96
-------
o
CN
<$
I
05
c
.CO
-5
I
to
— v
^
' —
^
LO
• —
CN
£;
05
JN
t
FN
00
^
-ft
Target Compoun
^ co *- LO o ^j- LO »- to to ^
CNCNtoOo'cNOoCNOOtQiNCOOoCN^
CN O5LOCNCO O5 00 *
LOlNLodcN^CNOoCOLOlOOilo'^
*-. CN CN ON CN IN
O *-• O5 IN ^J" 05 *-
Lo'LOO>-^to'cNt6sfrtOO5O06Lo'|v
GO oo CN LO *•*• *•- M"
CO 0 ^ *
CNto>>LC5tOT*O'NCNOO
»—Oo 00 ^ ON ON to
05 O
OoCO>»-^CNO5^]-OoOorNOO 1
*— LO »-»*— tOT%-*^-00
co to
tOCO^J-LOCNtO^J-CNCNLOIN 1
*— IN LO V-CN COCN*--^
OO »- LO "- 00 »-
OCv5\J-INCNOOINLOO5CN'^-tolNv-
O^^l-QO'—OOr--CNtO'---O>-
IN Lf) ^J- ^J- ^J-
tOCNM-tO00050000LoOINO^O
C5lN'!):C5C)»-C)OONCNCdCN»-»-
LO
LO tO*-*OOOotoO5lN 1 — OOf
'—tOLOQJT^COONONlNOiCNO^CN^
CN v- CN
CN , LotooocNOoQ
CN'CO^ l^to'^^Lo'tdO) 1 1
00
CO CoOolNLO^J-CNO5XJ-
CJCN^'-.CN'lO»-Oo';t:LoOo ' '
00 LofN\fCoCoOOO5
OOINLO»~-CN'to'»--OotO
£ '§ C N
2 0 «> -3> C
€ ^ g; £. co
-£ -Q „
2 P ^- ^ ? c
1 1 . i 1 1 . i i . i i
1 1 SI 1 1 H i ! * 1 1 1
ssii*u?i??sil
to
t- IN tO ?)
f «*> LO o
-. o o
f 00 IN 00
"~
- <» ^
C ^ CN ^
"
1 1 1
1 1 1
1
oo oo
- to ^ O
- o * — o
00
g to Co O
- ci *~- o
^
^ oo ^j- CN
g ^ crj d
1 1
i i i
i i >
1 1 1
Cb
> «> m
: c CD 5
! 2 c -9
Jj .c Q
! 1
-------
oo
05
i
I
3
I
I
•s
I
O
o
I
"
^.^
PM
to
"*
05
PM
xj-
*•-
IN
^
LO
0)
c
to
1, 7, 1-Trichloroeth
PM
PO
DO
LO
PM
PO
CJ
y
LO
*"""
IN
*"•
LO
PM
05
^
(J
5
IN
to
Benzene
IN
^
0
to
O5
O
PM
r~;
o
10
PM
_.
PM
PO
0
LO
to
o
LO
to
o
*
^
to
0
^_
CO
0
I
c?
Carbon tetrach/i
•^t-
0
PO
*-
O
0
PM
0)
PO
0
IN
PM
05
PO
PM
O
PO
0
0
PM
IN
Q}
to
0
^
^
'v-
PO
Cb
Trichloroethylen:
^_
to
O
^
^
sj.
PM
PO
PO
»^
*^
0)
to
LO
PM
0
PO
*--
^
IN
PO
*-
00
*^
to
c
-------
Sa
C <5
I
.2
S
I
O
|
3
i
(0 5)
U ^3
a ^
99
-------
5
nia
Ca
00
o
CQ
£S
co1
§
00
co
o
0)
8
10 Co ^
CM CM
tN CJ)
LO CM
10
I I I
«,
•8 | 9
o
-------
\
s
1
o>
•1
3
Q
14
1
§
<
.S
5
g
§
€
Q
^
O
5 "6
if
•i. «.
Q
!l
Table 49. Median i
Moniti
O ^ *
Q D!
CN
^i
en
^ *-.
Q cr?
^
^
5 2
00
^$!o
Q
ci
^
"•* QO
5:
CD
tj ^
^
^ _^
^ tf
II
>
*-~
Chemical
00 00 "- O O
d CN *^ *^ d
i
\}. v~ ^ 00 lO
»- »— d d d
CN o or) ^ oo
oo *-^ d d d
O tx CQ SJ- IO
IX K C») CN CN
LO Co »- \f O
d CN d *-- d
1 1 ! 1
01
^f Oo ^s CO >-~
to 01 O O ^J-
d oo cvj c\i d
•<* to o o o
d LO CN CN »-:
- M- 0 tv to
10 10 CN CN o
CO O IX !O O
\t ^ »~ CN »-^
«
Aromatic Hydrocarbon
Benzene
m,p-Xylene
o-Xylene
Ethylbenzene
Styrene
CN
10
CN
00
CN
10
XJ-
00
CN
CO
00
I
01
00
01
01
oi
O
«-:
*"•
CO
ui
o
M-
^^
Subtotal
V)
c
o
Chlorinated Hydrocarb
^
01
OM
to
^
CN
V—
CO
to
£!
^J-
d
co
^
to
to
»-
*^"
1, 1, 1-Trichloroethane
^
0
^
o
rv
d
CN
^;
to
d
to
^
CN
CN
CN
v:
^
CN
•*
OM
m,p-Dichlorobenzene
00
CN
01
O
Q
d
d
00
00
d
0
to
*^
ON
O
o
o
o
d
i
00
d
o
d
0
d
o>
d
OM
d
CN
d
•*
O
Carbon tetrachloride
CN
rt
LO
^
10
to
to
CO
r—
-
oi
^_
CN
CN
^.
^
^.
oi
to
^ —
to
oi
*•-
Subtotal
^~ to
5: Q
101
-------
s
g
•c
a
1
a
Q
to
S
Q
*
1
^
I
|
"^
(3
1
(A
1
i!
<
•S
^
1
(4
ft\
1
Q
||
H
r\ '5
>5 ^
8 ^
^S *^»
§ J
§
o ^ 5
^
^
Q S!
"^ —
PM
> co
^- »»*
Q O)
S
-^
^s
PO -^
o < 10
^ """
s
d
PM
^ -^
^ §
^-
^
^3 S
^
a CO
||
Chemical
PM co
PM ^
CO CO
»~ PO
10
M- CO
O 10
*- *~
*~ Px
»- O
1
05 tx
»~ PM
10 10
^ ^
io io
PM cd
PM 10
*~ 10
CO
O co
c.
o
Aromatic Hydrocarb'
Benzene
m,p-Xylene
05 IO *-
*- *^ C) Q
1 >~
IO O ^f
O *- *~ oo
fx N, ^t1
*~ •» d ^
»*.
10 OQ 05
05 03 >-^ io
^
O !O ^ PO
d d d d
i
^
^ ^j" oo ^j-
«~ >- IX
^. ^ »~
»- O O 10
O5 OQ Co
PO 10 *- o
PM
05
^- PO PM fx
PM »- O
PO PM
PO o PM io
"
o-Xylene
Ethylbenzene
Styrene
Subtotal
05
O
0
PM
co
*•*.
00
PO
£
IO
PM
PM
*^
10
10
PO
05
00
-------
increased levels of toluene and xylene attributed to printed material. De Bortoli
(6) found that all of 32 organics measured in 15 northern Italian homes
had indoor/outdoor ratios exceeding unity. Gammage (7) detected gasoline
vapors in 40 east Tennessee homes, most with attached garages. Monteith
(8) found increased levels of ten volatile organic compounds in 44 mobile
homes in Texas.
These eight studies of more than 800 homes show remarkable agreement
on the following points'
1 Essentially every one of the 40 or so organics studied has higher indoor
levels than outdoor, often 10 times higher.
2. Sources are numerous, including building materials, furnishings, dry
cleaned clothes, cigarettes, gasoline, cleansers, moth crystals, hot
showers, and printed material.
3. Ranges of concentrations are great, often two or more orders of
magnitude.
It seems clear that many indoor sources of toxic organics exist; however,
few have been unequivocally identified and fewer still have had their source
emission rates estimated (9). Identification of indoor sources from among
thousands of consumer products and building materials is required to allow
a better estimate of possible risks to public health and corrective actions
that can be taken.
Although occupational exposures did not account for most of the observed
differences between personal and outdoor concentrations, they did account
for the very highest exposures. For example, the person with the highest
exposure to vmylidene chloride and 1,1,1-trichloroethane was a painter.
Commuting was also implicated in increased exposures to benzene and
xylenes.
Outdoor Air. Reliance on outdoor monitors to estimate exposure is
contramdicated by this study. Correlations with personal exposures were
poor, even in Los Angeles where outdoor levets were the most nearly
comparable with personal exposures However, outdoor air concentrations
of two chemicals, trichloroethylene and carbon tetrachloride, were
significantly associated with indoor concentrations in New Jersey. These
outdoor levels are similar to those measured by all techniques (Tenax,
cryogenic trapping, evacuated cylinder) in urban and suburban areas
throughout the U.S. between 1970 and 1980 (10).
Drinking Water. Drinking water was a main source of exposure for the
trihalomethanes. In California, groundwater supplies provided increased
levels of bromoform and dibromochloromethane. Assuming 2 L/day water
intake and 20 mVday air intake, the daily intake of chloroform through water
generally exceeded the air intake. However, for the common chlorinated
solvents (trichloroethylene, tetrachloroethylene, 1,1,1 -trichloroethane),
drinking water usually supplied less than 1% of the total daily intake.
Breath Breath is an important mode of intake and excretion for many
volatile compounds (11). The compounds measured in the exhaled breath
of persons breathing pure air have been supplied by the bloodstream as
it passed through the lungs. The advantages of measuring breath rather
than blood are (1) the technique is noninvasive and therefore preferable
for use in studies requiring reasonable response rates from general public
volunteers; and (2) the measurement technique employed (Tenax, GC/MS
analysis) is more sensitive than the corresponding technique for blood
employed in the first phase of the TEAM Study. In fact, scores of compounds
103
-------
were quantified in breath using this technique but only one (chloroform)
was quantified regularly in blood during Phase I.
However, before these breath measurements can be used as an indicator
of exposure, an adequate model relating exposures at environmental
concentrations to body burden must be available.
Simple comparisons of exposure to breath concentrations do not take into
account the dependence of breath levels on pre-existing concentrations in
the body and also on the effective biological residence times of each chemical.
A simple two-parameter time-dependent model has been developed that
accounts for the effect of the initial breath concentration and the effective
residence time m the body (12). The model was tested in the TEAM Pilot
Study for 27 cases in which two breath samples and three intervening 8-
hour air samples were collected; the model predicted an effective half-life
of 21 hours for tetrachloroethylene and 9 hours for 1,1,1-trichloroethane.
A later "washout" study (13) performed over a 10-hour period in a pure
air chamber on an adult male exposed for 1 hour to tetrachloroethylene
vapors in a dry cleaning shop interior resulted in a measured effective half-
life of 21 hours.
Breath concentrations reflected personal exposures more closely than
outdoor concentrations. Spearman correlations between breath and
preceding personal exposure were significant (although low in magnitude)
for 10 of 11 prevalent chemicals in New Jersey, and for about 10 of 19
prevalent chemicals in Los Angeles, but correlations between breath and
preceding outdoor levels were significant for only three chemicals in New
Jersey and one in Los Angeles. A concurrent study of personal exposures
and breath concentrations of halogenated organics for 146 residents of three
other U.S. cities has recently reported similar findings (14, 15). Thus, the
feasibility of using breath measurements to estimate exposure to these
compounds has been demonstrated. This approach may be useful in cases
of spills or releases that have disappeared from the atmosphere before they
could be monitored—immediate breath measurements could determine the
approximate extent of population exposure. Similarly, breath measurements
of persons living near hazardous waste sites could be used to detect current
or recent exposure.
Sources of Exposure
Smoking. Benzene concentrations in air and breath were significantly
different for smokers and nonsmokers Three other aromatics (p-xylene,
ethylbenzene, and styrene) also showed significantly elevated levels in the
breath of smokers compared to nonsmokers during all six visits to New Jersey
and California. (The fifth aromatic, o-xylene, was elevated, but not always
significantly.) Octane, measured only in California, was significantly elevated
m the breath of smokers on all three visits. Two laboratory studies have
identified the five aromatic components in sidestream (16) and mainstream
smoke (17).
Benzene levels in the homes containing smokers were 30-50% higher
than in nonsmoking households. Since about 60% of U.S. children live in
homes with smokers, it appears possible that a large number of children
have increased exposure to benzene, a known leukemogen, during their early
years. A recent study by Sandier (18) comparing lifetime cancer mortality
rates of persons who were exposed or were not exposed as children to parental
smoking showed significant increases in hematopoietic (leukemia,
lymphomas, etc.) mortality rates in the exposed group. The odds ratio
increased from 1.7 with one parent smoking to 4.6 with both parents smoking.
104
-------
A second study by Stjernfeldt (19) in Sweden has also shown increased
leukemia rates in children of smoking mothers Odds ratios were 1.3 for
mothers smoking <10 cigarettes/day, 2.0 for mothers smoking >10
cigarettes/day.
Proximity to Point Sources. In New Jersey, census tracts were classified
as high and low exposure strata depending on whether they were within
1.5 km of suspected point sources or not. Those strata bordering the high
exposure strata and containing major highways as well were classified as
moderate exposure. In general, few differences in percent measurable or
concentrations in air and breath were seen between the high, moderate,
and low proximity strata Wind directions were measured, with some
chemicals displaying increases when the wind was from the east.
Uncertainty of Estimates
The uncertainty in the estimates of personal exposures of the target
population consists of two parts: survey sampling uncertainty and
measurement errors. For a simple random sample size of 350 persons,
assuming a log-normal distribution, standard sampling theory states that
the estimate of the median will be 95% certain to lie between the 44th
and 56th percentiles (20). Since our sample is clustered, the design effect
will broaden these ranges of uncertainty by a small amount. The
corresponding range for the summer group of 160 persons is 41-59%; and
for the winter group of 40 persons, 35-65%.
The second source of uncertainty is measurement error. Analysis of the
duplicate measurements for all three seasons using a method developed
by the author and based on observations in Evans et al. (21) resulted in
estimated frequency distributions of exposures that had geometric standard
deviations that were 5-20% less than the sample geometric standard
deviations. This is explained m detail in Appendix D.
Comparison of Weighted and Unweighted Frequency
Distributions
In an effort to better characterize relatively rare high exposures, the TEAM
Study selected potentially highly exposed persons with higher probabilities
than persons with low potential exposures. The known selection probabilities
of the sample members can be used to compute unbiased estimates of the
population distributions of exposures by weighting each observation inversely
to its selection probability; the observed ("unweighted") frequency
distributions of exposures are not a proper basis for inferences from the
sample to the target population.
If the initial hypotheses as to the main causes of exposures were correct,
the observed values would contain relatively more high exposures than
actually occur in the general population, represented by the weighted curve.
Thus, the unweighted curve should lie above the weighted curve, at least
at the higher exposures, on a log-normal probability graph. If, however, the
unweighted curve lies below the weighted curve at the higher exposures,
unsuspected causes of high exposures may be predominant.
By graphing both frequency distributions on one set of axes, one can gauge
the relative impact of the weighting process. A sample graph is displayed
as Figure 49. It will be noted that the unweighted curve lies below, instead
of above, the weighted curve, indicating that a preponderance of persons
who were expected to have low exposures in fact had high ones. This was
the case for three of the five chemicals compared in this way (see Appendix
105
-------
Figure 49. Weighted vs. unweighted frequency distributions for 1,1.1-
trichloroethane. The straight line is a log-normal curve with the
same geometric mean and geometric standard deviation as the
observed distribution.
1000
T
\—T
I—T
\
500
• Weighted
• Unweighted
200
100
I
g
o
O
«J
50
20
^ 10
J I
12 5 10 20 30 40 50 60 70 80 90 95
Cumulative Frequency. Percent
989999.599.899.9
106
-------
E). The reason for "guessing wrong" about the high exposures may be that
the importance of indoor sources was not well understood when the study
was designed, and such potential sources were therefore not used to stratify
the sample.
Health Effects
Although this study is concerned only with documenting exposures and
identifying possible sources, some discussion of health effects may be
appropriate, since these are the ultimate reasons for our interest in these
compounds. Two broad types of health effects may be distinguished: chronic
and acute.
Chronic Effects. The chronic effect of greatest interest is cancer. One
of the TEAM target compounds (benzene) is generally considered a known
human carcinogen. Five others are considered animal carcinogens and
therefore possible human carcinogens—carbon tetrachloride, chloroform,
trichloroethylene, tetrachloroethylene, and p-dichlorobenzene. Risk
assessments of human exposure to these six compounds have been made
using the TEAM exposure measurements and potency estimates from EPA
and other organizations, with an estimated range of 1000-5000 excess cancer
cases per year nationwide (22). These numbers far exceed the estimates
of 5-27 cases per year that have been used to regulate hazardous air pollutants
(NESHAPS).
Other TEAM target compounds are mutagens and therefore possible
carcinogens. These include styrene, 1,1,1-trichloroethane, and a-pinene.
Still others are promoters (co-carcinogens)—octane, decane, and undecane.
Others are presently being tested for carcinogenicity (xylenes, ethylbenzene).
Risk assessments of these chemicals are at present highly speculative;
however, it is possible that their effects on cancer incidence are not negligible
(23).
A second chronic effect of interest is chemical sensitivity. This is an ill-
defined condition marked by progressively more debilitating severe reactions
to various consumer products such as perfumes, soaps, tobacco smoke,
plastics, etc. The incidence of this syndrome is unknown; however, anecdotal
accounts indicate that it may be increasing sharply. The effects on productivity
of affected persons can be severe.
Acute Effects. A second ill-defined group of symptoms, sometimes known
as "Sick Building Syndrome," affects a number of office workers. The
symptoms include sleepiness, nausea, eye irritation, irritability, forgetfulness,
and a number of other respiratory and central nervous system disorders.
One experiment has determined that the symptoms are unlikely to be related
to mass psychology or otherwise psychosomatic (24). A second experiment
has shown that mixtures of common organic pollutants (mostly xylenes) at
levels similar to those in new buildings can cause both subjective and objective
symptoms in a group of sensitive individuals (25). The lowest experimental
concentration was 5 mg/m3; effects were still apparent, leading the
experimenter to hypothesize that effects may appear at levels as low as
1 mg/m3. Thus, the indoor air levels measured in the TEAM Study, which
exceeded 1 mg/m3(sum of 11 organics) in ~3% of New Jersey homes, may
have some potential of being associated with frank acute health effects,
although no attempt was made to observe such effects.
Standard Operating Procedures
To make the methods developed in the TEAM Study more widely available,
detailed descriptions of all procedures have been compiled. These Standard
107
-------
Operating Procedures (SOPs) are included as Volume IV of this publication.
The list of SOPs is included as Table 51.
TEAM Study Publications
A number of EPA reports and journal articles have been published on
various aspects of the TEAM Study. All of these publications are listed in
Table 52.
Validity of TEAM Data
At present, no standard methods exist for measuring volatile organic
compounds at environmental concentrations. Without such reference
methods, it is not possible to confirm the accuracy of any measurement
methods. The use of blanks, controls, deuterated compounds, duplicates,
external laboratories, and performance audits can serve to protect against
many errors, but not against all. For example, artifact formation during or
after sampling might not be detected by standard QA precautions (26). Side-
by-side sampling using completely different methods would be desirable,
and was performed to a limited extent in the California TEAM Study. In
that comparison Tenax cartridges and Tedlar bags agreed for 11 of 12
compounds in three 24-hour outdoor samples. A recent experiment (27)
compared Tenax cartridges collected at four widely different flow rates to
stainless steel evacuated canisters. Ten experiments were carried out in
an experimental home under controlled conditions. The two methods agreed
very closely for all ten target chemicals.
Although these results are encouraging, the number of samples is small.
In the absence of direct methods for determining accuracy, indirect methods
must be employed. Several different ways to assess the validity of the TEAM
data are discussed below.
7. Agreement with Other Methods.
Although few side-by-side studies comparing Tenax and other methods
have been carried out, results of ambient monitoring in the same city during
the same time period may be an approximate test of agreement between
two monitoring methods, provided that concentrations do not vary widely
between the sampling locations. Since 1983, the California Air Resources
Board (CARB) has operated a four-station ambient monitoring network in
Los Angeles. The method employed is 24-hour bag sampling followed by
gas chromatography analysis with electron capture or flame ionization
detection. Thus no sorbent is employed and artifacts peculiar to Tenax would
not be expected to occur.
The CARB network collected 25 samples during February 1984 and 30
during May 1984 at the four Los Angeles sites; concurrently, the TEAM
Study collected two consecutive 12-hour samples at 24 locations each month.
After averaging the two 12-hour samples, the TEAM concentrations were
compared to the CARB values (Table 53 and Figures 50 and 51).
Both methods found six chemicals to be generally below detectable limits.
Median values of six additional chemicals agreed to within one standard
deviation of each method except for trichloroethylene, which agreed to within
two standard deviations. The TEAM concentrations were higher for three
chemicals; lower for the other three. Both methods agreed in finding a sharp
decrease in concentration between February and May for four chemicals
but little change for the remaining two. Thus, the two methods appear to
agree to within their limits of precision, with no evidence indicating a
consistent bias.
108
-------
Table 51. Approved SOPs for Phase III TEAM Study
RTI/ACS-SOP No.
SOP Title
321-001
322-001
331-001
331-002
332-001
337-001
340-001
350-001
350-002
361-001
367-002
410-001
431-001 (Air)
432-001 (Water)
437-001 (Breath)
461-001 (Air)
432-001 (Water)
467-001 (Breath)
470-001
482-001
481-001 (Air)
487-001 (Breath)
512-001
Tenax Cleanup and Preparation
Cleanup of Water Collection Bottles
Collection of Personal Air Samples
Collection of Fixed Site Air Samples
Collection of Water Samples
Collection of Breath Samples
Shipment of Field Sampling Equipment
Site Workroom Procedures and Rules
Maintenance and Use of the Van
Calibration of Dupont P-125A Constant Flow
Samples
Calibration of Nutech Model 221 Gas Sampler with
a Dry Gas Meter
Using Samp/ing Protocol/Chain-of-Custody Sheet in
the Field
Storage of Samples at the Field Sampling Site
Shipment of Samples from the Field to RTI
Receipt of Air, Breath, and Water Samples at RTI
Storage of Water Samples at RTI
Storage of Tenax Samples at RTI
Analysis of Drinking Water by Purge Trap Gas
533-001
533-002
612-001
630-001
630-002
630-003
630-004
712-001
711-001 (Air)
717-001 (Breath)
790-001
Chroma tography
Analysis of Organic Compounds Collected on
Tenax Using the Finnigan 3300 GC/MS/COMP
System
Analysis of Organic Compounds Collected on
Tenax Using the Finnigan 4021 GC/MS/COMP
System
Preparation of Purge and Trap Calibration Solutions
Preparing Relative Molar Response Tenax Car-
tridges Using a Permeation System
Preparing Relative Molar Response and Cc nn
Performance Evaluation Tenax Cartridges jsing a
Flash Evaporation System
Loading External Standards on Tenax Cartridges Via
Injection Using a Permeation System
Loading Deuterium Standards on Tenax Cartridges
Using a Permeation System
Quantitation of Volatile Organic Compounds in
Water
Quantitation of Volatile Organic Compounds in
Tenax Samples
Preparation and Submission of Data Summary
Sheets to the Center for Computer Application/Data
Entry (CCA/DE)
109
-------
Table 51. (continued)
RJI/ACS-SOP No. SOP Title
810-001 Preparation and Handling of QA Performance Audit
Samples on Tenax for GC/MS Analysis
812-001 Conducting a QA System Audit of Sample and Data
Collection in the Field
860-001 Preparing Quality Control Samples on Tenax
Cartridges
862-001 Preparation of Water Blanks and Controls
861-002 (Air) Shipment of QC Samples to the Field Sampling Site
862-002 (Water)
867-002 (Breath)
861-003 (Air) Exposure of QC Samples
862-003 (Water)
867-003 (Breath)
881-001 (Air) Submission of QA Samples to and Receipt of Data
882-001 (Water) from a QA Laboratory
887-001 (Breath)
2. Confirmation by Other Studies.
The major finding of the TEAM Study was the higher indoor concentrations
of eleven prevalent chemicals. A total of eight studies (2-8, 15) some using
methods quite different from those employed in the TEAM Study, have also
found higher indoor concentrations of these and other chemicals in other
countries and other areas in the United States.
3. Internal Consistency.
If chemical reactions or other random errors were affecting an appreciable
proportion of samples in a major way, correlations between, for example,
breath and air samples would not be expected. In fact, however, ten of eleven
chemicals showed significant correlations between breath concentrations
and the preceding personal air concentrations. At the same time, few of
these chemicals showed correlations between breath and outdoor air
samples. The most natural conclusion from these observations is that exhaled
breath concentrations are closely related to inhaled concentrations and less
closely related to outdoor concentrations. It is difficult to imagine any
explanation attributing such a pattern of correlations to chance.
4. Ability to Predict Measurable Phenomena.
A number of hypotheses have been generated by the TEAM findings, some
of which have now been tested and confirmed to varying degrees. Some
of these hypotheses are listed below.
a. A Main Source of Exposure to Aromatics is Tobacco Smoke
As noted, on all six trips to New Jersey and California, smokers had
significantly elevated breath levels of benzene, styrene, ethylbenzene,
and m,/5-xylene. A recent study of mainstream cigarette smoke has
confirmed that these components are present in significant amounts
110
-------
Table 52. TEAM Study Publications
1. Pellizzari, E.D., Erickson, M.D., Giguere, M.T., Hartwell, T.D., Williams,
S.R., Sparacino, CM., Zelon, H., and Waddell, R.D. (1980) Preliminary
Study on Toxic Chemicals in Environmental and Human Samples:
Work Plan, Vols. I and II (Phase I), U.S. Environmental Protection
Agency, Washington DC.
2. Pellizzari, E.D., Erickson, M.D., Sparacino, C.M., Hartwell, T.D., Zelon,
H., Rosenzweig, M., and Leininger, C. (1981) Total Exposure Assess-
ment Methodology (TEAM) Study: Phase II Work Plan, U.S. En-
vironmental Protection Agency, Washington, DC.
3. Entz, R., Thomas, K., and Diachenko, G. (1982) Residues of volatile
halocarbons in food using headspace gas chromatography, J. Agric.
Food Chem. 30:846-849.
4. Pellizzari, E.D., Hartwell, J., Zelon, H., Leininger, C., Erickson, M.,
Cooper, S., Whitaker, D., and Wallace, L (1982) Total Exposure
Assessment Methodology (TEAM) Prepilot Study—Northern New
Jersey, U.S. Environmental Protection Agency, Washington, DC.
5. Sparacino, C., Pellizzari, E., and Erickson, M. (1982) Quality Assurance
for the Total Exposure Assessment Methodology (TEAM) Prepilot
Study, U.S. Environmental Protection Agency, Washington, DC.
6. Sparacino, C., Leininger, C., Zelon, H., Hartwell, T., Erickson, M., and
Pellizzari, E. (1982) Sampling and Analysis for the Total Exposure
Assessment Methodology (TEAM) Prepilot Study, U.S. Environmental
Protection Agency, Washington, DC.
7. Wallace, L.A., Zweidinger, R., Erickson, M., Cooper, S., Whitaker, D.,
and Pellizzari, E.D. (1982) Monitoring individual exposure:
measurements of volatile organic compounds in breathing-zone air,
drinking water, and exhaled breath. Environment International
8:269-282.
8. Wallace, L.A. (1982) Measuring direct individual exposure to toxic
substance. Toxic Substances Journal 4:174-183.
9. Wallace, L.A. (1982) Direct measurement of individual human
exposures and body burden: research needs, J. Environmental Science
and Health A17:531-54O.
10. Zweidinger, R., Erickson, M., Cooper, S., Whitaker, D., Pellizzari, E.D.,
and Wallace, L.A. (1982) Direct Measurement of Volatile Organic Com-
pounds in Breathing Zone Air, Drinking Water, Breath, Blood, and
Urine, U.S. Environmental Protection Agency, Washington, DC, NTIS
#PB-82-186-545.
11. Pellizzari, E.D., Hartwell, T.D., Leininger, C., Zelon, H., Williams, S.,
Breen, J., and Wallace, L. (1983) Human exposure to vapor-phase
halogenated hydrocarbons: fixed-site vs. personal exposure. Pro-
ceedings: National Symposium on Recent Advances in Pollutant
Monitoring of Ambient Air and Stationary Sources, Environmental
Monitoring Systems Lab., Research Triangle Park, NC, EPA
600/9-83-007, pp. 264-288.
12. Hartwell, T., Perritt, K., Zelon, H., Whitmore, R., Pellizzari, E., and
Wallace, L. (1984) Comparison of indoor and outdoor levels for air
volatiles in New Jersey, in Indoor Air, v. 4, Chemical Characterization
and Personal Exposure, B. Berglund et a/., eds., Swedish Council for
Building Research. Stockholm, pp. 81-86.
13. Pellizzari, E., Sparacino, C., Sheldon, L., Leininger, C., Zelon, H.,
Hartwell, T., and Wallace, L. (1984) Sampling and analysis for volatile
organics in indoor and outdoor air in New Jersey, in Indoor Air, v. 4,
Chemical Characterization and Personal Exposure, B. Berglund et al.,
111
-------
Table 52. (continued)
eds., Swedish Council for Building Research, Stockholm, pp. 221-226,
14. Pellizzari, E, Sheldon, L, Sparacino, C., Bursey, J., Wallace, L, and
Bromberg, S. (1984) Volatile organic levels in indoor air, in Indoor Air,
v. 4, Chemical Characterization and Personal Exposure, B. Berglund et
a/., eds., Swedish Council for Bui/ding Research, Stockholm,
pp. 303-308.
15. Pellizzari, E., Hartwell, T., Sparacino, C., Sheldon, L., Whitmore, Ft.,
Leininger, C., and Zelon, H. (1984) Total Exposure Assessment
Methodology (TEAM) Study: First Season—Northern New Jersey,
Interim Report, Contract #68-02-3679, U.S. Environmental Protection
Agency, Washington, DC.
16. Pellizzari, E.D., Sparacino, CM, Hartwell, T.D., Sheldon, L.S., Whitmore,
R., Leininger, C., and Zelon, H. (1984) Total Exposure Assessment
Methodology (TEAM) Special Study: Dry Cleaners, Final Report,
Contract #68-02-3679, U.S. Environmental Protection Agency,
Washington, DC.
17. Wallace, L, Pellizzari, E., Hartwell, T., Zelon, H., Sparacino, C., and
Whitmore, R. (1984) Analysis of exhaled breath of 355 urban
residents for volatile organic compounds, in Indoor Air, v. 4, Chemical
Characterization and Personal Exposure, B. Berglund et a/., eds.,
Swedish Council for Building Research, Stockholm, pp. 15-20.
18. Wallace, L, Bromberg, S., Pellizzari, E., Hartwell, T., Zelon, H., and
Sheldon, L. (1984) Plan and preliminary results of the U.S. En-
vironmental Protection Agency's indoor air monitoring program, in
Indoor Air, v. 1, Recent Advances in the Health Sciences and
Technology, B. Berglund et al., eds., Swedish Council for Bui/ding
Research, Stockholm, pp. 173-178.
19. Wallace. LA., Pellizzari, E., Hartwell, T., Rosenzweig, M., Erickson, M.,
Sparacino, C., and Zelon, H. (1984) Personal exposure to volatile
organic compounds: I. direct measurement in breathing-zone air, drinking
water, food, and exhaled breath. Environmental Research 35:293-319.
20. Gordon, S.M., Wallace, L, Pellizzari. E., and O'Neill, H.J. (1987) Breath
measurements in a clean-air chamber to determine "wash-out" times for
volatile organic compounds at normal environmental concentrations.
Atmospheric Environment, in press.
21. Handy, R.W., et al. (1985) Total Exposure Assessment Methodology
(TEAM) Study: Standard Operating Procedures, Volume IV, Final
Report, Contract #68-02-3679, U.S. Environmental Protection Agency,
Washington, DC.
22. Pellizzari, E.D., Perritt, K., Hartwell, T.D., Michael, L.C., Whitmore, R.,
Handy, R.W., Smith, D., and Zelon, H. (1985) Total Exposure Assess-
ment Methodology (TEAM) Study: Elizabeth and Bayonne, New
Jersey; Devils Lake, North Dakota; and Greensboro, North Carolina,
Volume II, Final Report, Contract #68-02-3679, U.S. Environmental
Protection Agency, Washington, DC.
23. Pellizzari, ED., Perritt, K., Hartwell, T.D., Michael, L.C., Whitmore, R.,
Handy, R.W., Smith, D., and Zelon, H. (1985) Total Exposure Assess-
ment Methodology (TEAM) Study: Selected Communities in Northern
and Southern California, Volume III, Final Report, Contract
#68-02-3679, U.S. Environmental Protection Agency, Washington, DC.
24. Sheldon, L.S., Handy, R.W., Hartwell, T.D., Whitmore, R.W., Zelon, H.,
and Pellizzari, E.D. (1985) Total Exposure Assessment Methodology
112
-------
Table 52. (continued)
Special Study: Indoor Air, Final Report, Contract #68-02-3679, U.S.
Environmental Protection Agency, Washington, DC.
25. Sheldon, L.S., Handy, R.W., Hartwell, T.D., Leininger, C., and Zelon, H.
(J985) Total Exposure Assessment Methodology (TEAM) Study:
Mother's Milk, Final Report, Contract #68-02-3679, U.S. Environmen-
tal Protection Agency, Washington, DC.
26. Wallace, L, Pellizzari, E., Hartwell, T., Sparacino, C., Sheldon, L, and
Zelon, H. (1985) Personal exposures, indoor-outdoor relationships and
breath levels of toxic air pollutants measured for 355 persons in New
Jersey, Atmospheric Environment 19:1651-1661.
27. Wallace, L, Pellizzari, E., and Gordon, S. (1985) Organic chemicals in
indoor air: a review of human exposure studies and indoor air quality
studies, in Indoor Air and Human Health: Proceedings of the Seventh
ORNL Life Sciences Symposium, Knoxville, TN, October 29-31, 1984,
Lewis Publishers, Chelsea, Ml.
28. Wallace, L., Pellizzari, E., Hartwell, T., Zelon, H., Sparacino, C., and
Whitmore, R. (1985) Concentrations of 20 volatile compounds in the
air and drinking water of 350 residents of New Jersey compared to
concentrations of their exhaled breath, J. Occup. Med 28:603-608.
29. Wallace, L. (1986) Total Exposure Assessment Methodology (TEAM)
Study: Summary and Analysis, Volume I, Final Report, Contract
#68-02-3679, U.S. Environmental Protection Agency, Washington, DC.
30. Wallace, LA., Pellizzari, E.D., Hartwell, T.D., Whitmore, R., Sparacino,
C., and Zelon, H. (1986) Total Exposure Assessment Methodology
(TEAM) study: personal exposures, indoor-outdoor relationships, and
breath levels of volatile organic compounds in New Jersey, Environ-
ment International, 12: #1-4.
31 Wallace, L.A., Pellizzari, £., Sheldon, L, Hartwell, T., Sparacino, C., and
Zelon, H. (1986) The Total Exposure Assessment Methodology (TEAM)
Study: direct measurement of personal exposures through air and water
for 600 residents of several U.S. cities, in Pollutants in a Multimedia
Environment, Plenum Press, New York.
32. Wallace, L A. (1986) Personal exposures, indoor and outdoor concen-
trations, and exhaled breath concentrations of selected volatile organic
compounds measured for 550 residents of New Jersey, North Dakota,
North Carolina, and California, Toxicological and Environmental
Chemistry, 12.215-236.
33. Wallace, L (1986) Estimating risk from measured exposures to six
suspected carcinogens in personal air and drinking water of 600 U.S.
residents, presented at the 79th Annual Meeting of the Air Pollution
Control Association, Minneapolis, MN, June 22-27, 1986.
34. Wallace, L. (1986) Cancer risks from organic chemicals in the home, in
Environmental Risk Management—Is Analysis Useful?, Air Pollution
Control Association, Chicago. Publication #50-55.
35. Wallace, LA and Pellizzari, E.D. (1986) Personal air exposures and breath
concentrations of benzene and other volatile hydrocarbons for smokers
and nonsmokers. Toxicology Letters 35:113.116.
36. Wallace, L, Pellizzari, E., Hartwell, T., Perritt. K, andZiegenfus. R. (1987)
Exposures to benzene and other volatile compounds from active and
passive smoking. Archives of Environmental Health, in press.
113
-------
Table 52. (continued)
37. Wallace, LA., Pellizzari, E.D., Hartwell, T.D., Sparacino. C., Whitmore, R.,
Sheldon, L, Zelon, H., and Perritt, K. (!987j The TEAM Study, persona/
exposures to toxic substances in air, drinking water and breath of 400
residents of New Jersey, North Carolina, and North Dakota,
Environmental Research, in press.
38. Wallace, LA., Pellizzari, E.D.. Hartwell. T.D., Sparacino, C.M., Sheldon,
L.S., and Zelon, H. (1987) Results from the first three seasons of
the TEAM Study: personal exposures, indoor-outdoor relationships,
and breath levels of toxic air pollutants measured for 355 persons in
New Jersey, in Environmental Epidemiology, Lewis Publishers,
Chelsea, Ml.
39. Wallace, LA., Pellizzari, E.D., Hartwell, T.D., Whitmore, R., Zelon, H.,
Perritt, K., and Sheldon, L. (1987) The California TEAM Study: breath
concentrations and personal exposures to 26 volatile compounds in air
and drinking water of 188 residents of Los Angeles, Antioch, and
Pittsburg, California. Atmospheric Environment, in press.
40. Wallace, L A., Pellizzari, E.D., Leaderer, B., Zelon, H., and Sheldon, L.
(1987) Emissions of volatile organic compounds from building materials
and consumer products. Atmospheric Environment 21:385-393.
Table 53. Comparison of Outdoor Measurements of Toxics by TEAM
Study and by California Air Resources Board
Chemical
Benzene
1, 1, 1-Trichloroethane
Tetrachloroethylene
Trichloroethylene
Carbon tetrachloride
Chloroform
February
CARBa TEAM
(N=25)b fN =25)
20d
13
12
2
0.4
0.3
± 6.4e
± 8.1
± 6.8
±0.7
± 0.1
± 0.1
16
33
8
0.
0.
0.
6
6
7
+ 7.8
+ 35
± 8.7
± 1.9
± 0.2
± 1.0
9.
5.
3.
0.
0.
0.
CARB
(N~32)b
3
1
2
6
5
2
+ 3.2
± 2.9
± 1.9
± 0.3
± 0.1
± 0.1
May
3.
5.
1.
0.
0.
0.
TEAM
(N=25)c
6
9
5
2
7
3
± 3.0
± 0.6
± 1.5
± 1.0
± 0.2
± 0.6
aCalifornia Air Resources Board.
bTotal number of 24-hour measurements made at four sites in Los
Angeles.
cMean of two consecutive 12-hour measurements made at 25 sites in
South Bay section of Los Angeles.
^Arithmetic mean l^g/m3).
eStandard deviation f^g/m3).
(17). The elevated benzene levels observed in indoor air are consistent
with a sidestream concentration 5-10 times that in mainstream
smoke. Such an increased sidestream concentration of benzene (250
/ug/cigarette compared to 35 /ug in mainstream smoke) has recently
been observed (28).
b. Use of Hot Water is the Main Source of Airborne Chloroform
in Homes
This hypothesis is based on the fact that the median indoor level
of chloroform was four times the median outdoor level in New Jersey
in the fall of 1981. Assuming a typical air exchange rate of 0.5 ach
114
-------
Figure 50.
Comparison of outdoor air measurements in Los Angeles by the
California Air Resources Board (CARB) and the TEAM Study. The
CARS measurements employed a Tedlar-bag collection with GC-
ECD analysis; the TEAM measurements employed a Tenax collector
with GC-MS analysis. CARB measurements were 24-hour samples
collected in four locations in the Los Angeles basin; TEAM
measurements were two consecutive 12-hour samples collected at
25 homes in the South Bay section of Los Angeles. (The two 12-
hour values at each home were combined into a single 24-hour
value.) A total of 25 measurements were made by both CARB and
TEAM in February; and 30 by CARB and 25 by TEAM in May.
Not shown are six toxic chemicals for which both methods found
no detectable amounts. Units are fjg/m3. Arithmetic means for all
but one case (trichloroethylene in May) were within one standard
deviation associated with each method.
Outdoor Air Concentrations of Toxics in
Los Angeles: 24-Hr Means
100 r-
50
10
0.5
0.1
I
-------
Figure 51. Comparison of median outdoor air concentrations. (See caption for
Figure 50.)
Outdoor Air Concentrations of Toxics in
Los Angeles' 24-Hr Medians
50
40 -
30
20 -
70 -
0.5
7,1,1-Trich lor o-
ethane
Benzene
Carbon
Tet
Chloroform
116
-------
and a typical home volume of 300 m3, a total of 150 /ug of chloroform
would have to be liberated each hour to achieve a steady-state
concentration of 1 ^g/m3 above background (outdoor) levels.
Assuming a concentration of 50 fjg chloroform per liter of water,
at least 3 L/hr or —70 L/day would be required to liberate all contained
chloroform to achieve this indoor concentration. Common large-scale
uses of hot water in most homes include showers, baths, and washing
clothes or dishes. Since this speculation was first made, a study
funded by EPA has reported such liberation of trichloroethylene from
spiked water sources during model showers under controlled
conditions (29). Similar studies using chloroform have also supported
the hypothesis (30).
c. Tetrachloroethylene Levels in Dry Cleaning Shops Sometimes
Exceed 1000 pg/m3
From the fact that persons who reported visiting a dry cleaning shop
showed twice as much tetrachloroethylene in their breath as the
other persons (median values) and assuming a 5-minute exposure
in the shop, one can calculate that the concentration in the shop
must have been about 12h/5m=140 times the typical ambient level
of 10/ug/m3, or more than 1000 pg/m3. A special study of dry cleaning
shops (31) showed that tetrachloroethylene levels up to 10,000 /ug/
m3 were observed.
d. The Effective Half-Life of Tetrachloroethylene in Breath is ~21
Hours
By using the TEAM pilot study measurements of tetrachloroethylene
in breath and personal air of 12 persons and assuming a common
half-life in the body, a value of 21 hours was calculated (12). This
estimate was confirmed by direct measurement of breath values of
a volunteer over a 10-hour period in a clean-air chamber (13).
e. Benzene Exposures While Filling Gas Tanks May Exceed 1000
fjg/m3
Persons who reported filling their tanks with gasoline had twice as
much benzene on their breath as persons who did not. The same
calculation as above for tetrachloroethylene indicates that concen-
trations at the breathing zone may exceed 1000 jug/m3 (100 times
the ambient level). A recent study (32) has measured benzene levels
during refueling of ~1 ppm (3000 /ug/m3).
Other hypotheses regarding indoor sources have also been generated but
not yet tested. These include1
a. Moth Crystals and Room Air Deodorizers are Important Sources
of f>-Dichlorobenzene Exposures in Homes
This hypothesis is suggested by the fact that p-dichlorobenzene was
prevalent in —80% of homes and that its main uses include the two
uses described above. The greatly elevated indoor concentrations of
p-dichlorobenzene are consistent with the main purpose of both uses,
which is to supply a long-lasting continuous source of elevated levels
of p-dichlorobenzene in the home. The observed steep geometric
117
-------
standard deviation may be explained by the fact that homes with
such sources will have high concentrations while homes without
sources will have background levels; thus a great dynamic range
in concentrations and correspondingly large geometric standard
deviations will be achieved.
b. Tetrachloroethylene Exposures are Elevated by Wearing or
Storing Dry-Cleaned Clothes
The evidence for this comes from observed higher levels in the breath
of persons visiting dry cleaning shops; higher exposures of persons
working in textile plants; and higher exposures of persons visiting
dry cleaning shops. Also, one study has measured increased
concentrations in a home for up to one week after placing newly
dry cleaned clothes in a closet (33).
c. Employment Leads to Increased Exposures to Some Toxic
Chemicals
More than 50 significant relationships with increased exposures or
breath concentrations were observed for 19 employment-related
variables.
d. Common Activities Lead to Increased Exposures to Some Toxic
Chemicals
Among the activities identified with increased exposures were:
pumping gasoline, visiting service stations, visiting dry cleaners,
traveling in a car, furniture refinishing, painting, scale model building,
pesticide use, and smoking. More than 20 such activities were
identified (34).
e. Household and Personal Characteristics are Associated with
Significantly Increased or Decreased Exposures to Some Toxic
Chemicals
Age, race, and sex were personal characteristics occasionally
associated with significantly higher or lower exposures. Significant
household variables included age of the house, type of heat,
ventilation, and the presence in the home of hobbyists, smokers,
and persons with certain types of occupations (particularly chemical,
plastics, and paint plant workers).
Several other hypotheses may be generated by these findings. For example,
the higher exposures of females to trichloroethylene may be due to the
chemical's use in cosmetics as a solvent and in opaquing fluids used in
offices.
The occasional finding of increased indoor air concentrations associated
with the presence of a chemical worker in the home suggests that some
transport of pollutants from the workplace may be occurring.
The reduced exposures associated with gardening are consistent with the
greater amount of time likely to be spent outdoors, where concentrations
are nearly always lower.
If in fact indoor concentrations normally exceed outdoor levels, it will be
important to consider these indoor exposures as part of any regulatory process
dealing with traditional sources. For example, if mean indoor levels are
118
-------
normally 2-4 times the outdoor concentrations (as observed in this study)
a 50% decrease in outdoor levels will produce a decrease in human exposure
of only 12-25%. Some attention to reducing indoor concentrations (by
removing sources, substituting innocuous chemicals in products, establishing
standards for building materials, increasing ventilation, etc.) may provide
more cost-effective reductions of human exposure than traditional
environmental regulations of emissions from major point sources
References
1. Belsley, D. A, Kuh, E., and Welsch, R. E. C\ 980) Regression Diagnostics.
Wiley, New York
2. M0lhave, L, and Moller, J. (1979) The atmospheric environment in
modern Danish dwellings: measurements in 39 flats, in Indoor Climate,
pp. 171-186, Danish Building Research Institute, Copenhagen.
3. Jarke, F. H., Gordon, S., and Dravnieks, A. (1981) ASHRAE Report #87,
IITRI, Chicago
4. Lebret, E., Van de Wiel, H. J., Bos, H. P , Noij, D., and Boleij, J. S
M. (1984) Volatile hydrocarbons in Dutch homes, in Indoor Air, v. 4,
pp. 1 69-1 74, Swedish Council for Building Research, Stockholm.
5. Seifert, B., and Abraham, H. J. (1982) Indoor air concentrations of
benzene and some other aromatic hydrocarbons, Ecotoxicol. Environ.
Safety, 6:190-192
6. De Bortoli, M., Knoppel, H., Pecchio, E., Peil, A., Rogora, L.,
Schauenberg, H., Schhtt, H., and Vissers, H. (1984) Integrating 'real
life' measurements of organic pollution in indoor and outdoor air of
homes in northern Italy, in Indoor Air, v. 4, pp. 21 -26, Swedish Council
for Building Research, Stockholm.
7. Gammage, R. B., White, D. A., and Gupta, K. C. (1984) Residential
measurements of high volatility organics and their sources, in Indoor
Air, v. 4, pp. 157-162, Swedish Council for Building Research,
Stockholm.
8. Monteith, K. D., Stock, T. H., and Seifert, W. E., Jr. (1984) Sources
and characterization of organic air contaminants inside manufactured
housing, in Indoor Air, v 4, pp. 285-290, Swedish Council for Building
Research, Stockholm.
9. Girman, J R., Hodgson, A. T., and Newton, A. S. (1 984) Volatile organic
emissions from adhesives with indoor applications, in Indoor Air, v.
4, pp. 271 -276, Swedish Council for Building Research, Stockholm.
10. Brodzinsky, R., and Singh, H. (1982) Volatile organic chemicals in the
atmosphere: an assessment of available data, Environmental Sciences
Research Laboratory, USEPA, Research Triangle Park, NC.
11. Krotoszynski, B. K., Bruneau, G., and O'Neill, H. J. (1979) Measurement
of chemical inhalation exposure in urban population in the presence
of endogenous effluents, J. Anal. Tox., 3:225-234.
12. Wallace, L., Pellizzari, E., Hartwell, T., Sparacino, C., and Zelon, H.
(1983) Personal exposures to volatile organics and other compounds
indoors and outdoors—the TEAM Study, paper #83.912 presented at
119
-------
the 76th Annual Conference of the Air Pollution Control Assoc., Atlanta,
GA, June.
13. Gordon, S., Wallace, L, Pelhzzari, E., and O'Neill, H. J. (1985) Washout
of volatile organic compounds in exhaled breath of four nonoccupa-
tionally exposed subjects in a clean-air chamber. Paper delivered at
Workshop on Human Exposure Assessment: Monitoring and Modeling,
Harvard University, Cambridge, MA, Sept. 30-Oct. 2, 1S85.
14. Hartwell, T., Zelon, H., Leininger, L., Clayton, C., Crowder, J., and
Pellizzari, E. (1984) Comparative statistical analysis for volatile
halocarbons in indoor and outdoor air, in Indoor Air, v. 4, pp. 57-61,
Swedish Council for Building Research, Stockholm.
15. Pellizzari, E. D., Hartwell, T. D., Leininger, C., Zelon, H., Williams, S.,
Breen, J., and Wallace, L. (1983) Human exposure to vapor-phase
halogenated hydrocarbons: fixed-site vs. personal exposure, Proceed-
ings from Symposium on Ambient, Source, and Exposure Monitoring
of Non-Criteria Pollutants, May 1982, sponsored by EMSL, USEPA,
Research Triangle Park, NC.
16. Jermini, C., Weber, A., and Grandjean, E. (1976) Quantitative
determination of various gas-phase components of the sidestream
smoke of cigarettes in room air (in German) Int. Arch. Occup. Env.
Health, 36:169-181.
17. Higgms, C. et al. (1983) Applications of Tenax trapping to cigarette
smoking, J. Assoc. Official Analytical Chemists, 66:1074-1083.
18. Sandier, D. P., Everson, R. B., Wilcox, A. J , and Browder, J. P. (1985)
Cancer risk in adulthood from early life exposure to parents' smoking,
Am. J. Public Health, 75:467.
19. Stjernfeldt, M., Berglund, K., Lindsten, J., and Ludvigsson, J. (1986)
Maternal Smoking During Pregnancy and Risk of Childhood Cancer,
Lancet, June 14, 1986, pp. 1350-1352
20. Conover, W. J. (1 980) Practical nonparametric statistics, 2nd ed.. New
York: John Wiley.
21. Evans, J. S., Cooper, S. W., and Kinney, P. (1984) On the propagation
of error in air pollution measurements, Env. Mon. and Assess., 4:139-
153.
22. Wallace, L. A. (1986) Estimating risk from measured exposures to six
suspected carcinogens in personal air and drinking water of 600 U.S.
residents. Paper #86-66.4 presented at the 79th Annual Meeting of
the Air Pollution Control Association, Minneapolis, MN, June 22-27,
1986.
23. Tancrede, M., Wilson, R., Zeise, L., and Crouch, E. A. C. (1986) The
carcinogenic risk of organic vapors indoors: a survey, Energy and
Environmental Policy Center Discussion Paper Series #E-86-06,
Kennedy School of Government, Harvard University, June 1986.
24. Berglund, B., Berglund.U., Johansson,!., and Lindvall, T. (1984) Mobile
laboratory for sensory air quality studies in non-industrial environments,
in Indoor Air: Sensory and Hyperreactivity Reactions to Sick Buildings,
vol. 3, B. Berglund et al., eds., Swedish Council for Building Research,
Stockholm.
120
-------
25. Mtflhave, L, Bach, B., and Pedersen, 0. F. (1984) Human reactions
during controlled exposures to low concentrations of organic gases
and vapours known as normal indoor air pollutants, in Indoor Air:
Sensory and Hyperreactivity Reactions to Sick Buildings, vol. 3, B.
Berglund et al., eds., Swedish Council for Building Research, Stockholm.
26. Walling, J. F. (1984) The utility of distributed air volume sets when
sampling ambient air using solid adsorbents, Atmospheric Environment
18:855-859.
27. Spicer, C. W. et al. (1986) Intercomparison of sampling technology for
toxic organic compounds in indoor air, \nProceedings, 1986 EPA/APCA
Symposium on Measurement of Toxic Air Pollutants, Raleigh, NC, April
27-30, 1986.
28. Higgins, C. (1986) Personal communication.
29. Andelman, J. B. (1985) Inhalation exposure in the home to volatile
organic contaminants of drfnking water. Science of the Total
Environment 47:443-460.
30. Andelman, J. B. (1986) Personal communication.
31. Pellizzari, E. D., Sparacino, C. M., Hartwell, T. D., Sheldon, L. S.,
Whitmore, R., Leminger, C., and Zelon.H. (1984) Total Exposure
Assessment Methodology (TEAM) Special Study: Dry Cleaners, Final
Report, Contract No. 68-02-3679, USEPA, Washington, DC 20460.
32. Bond, A. E., Thompson, V. L., Ortman, G., Black, F. M., and Sigsby,
J. E., Jr. (1985) Self Service Station Vehicle Refueling Exposure Study,
Internal report, USEPA, Research Triangle Park, NC.
33. Howie, S. J. (1981) Ambient Perchloroethylene Levels Inside Coin-
Operated Laundries with Drycleaning Machines on the Premises,
Contract A/68-02-2722, USEPA, Research Triangle Park, NC.
34. Wallace, L., Pellizzari, E., Hartwell, T., Sparacino, C., Sheldon, L., Zelon,
H., and Perritt, K. (1 987) The TEAM Study: personal exposures to toxic
substances in air, drinking water and breath of 400 residents of New
Jersey, North Carolina, and North Dakota. Environmental Research,
m press.
121
-------
Appendix A
Sources of Exposure to Volatile Organic Chemicals:
An Analysis of Personal Exposures in the TEAM Study
Stepwise regression results for personal air and breath concentrations
of 11 prevalent chemicals in the three New Jersey trips and 16 chemicals
in the three California trips are summarized below.
New Jersey—Fall 1981
A total of 33 stepwise regressions were run for eleven chemicals in three
media: day and night personal air and breath.
The best models for each of the eleven prevalent chemicals are summarized
for daytime personal air (Table A-1): overnight personal air (Table A-2); and
breath (Table A-3). Listed in the tables are the number of persons in each
category; the slope (b) and its associated standard error (SE) of the linear
regression; the F and p-values associated with each relationship; and the
total R2 (or percent of variance) explained by the "best" model. In general,
the variances of the aromatic compounds were best explained by the
questionnaire variables (R2 values as high as 32%), while those of the
chlorinated compounds were not well explained (R2 values normally less
than 10%). As an additional check of the stepwise regression, comparisons
with the t-tests were performed. In more than 80% of the cases, the two
approaches agreed in identifying significant variables. The best models
identified by the stepwise regression contain only eight (out of 108) variables
(identified by asterisks in Tables A-1 - A-3) that were not also significant
by the t-test.
Of 14 specific occupations selected as having potential for exposures, 11
had at least one positive significant relationship with air or breath levels
of the 11 prevalent chemicals. Similarly 1 6 of 29 activities and 9 of 17
personal or household characteristics were identified with significantly
increased (occasionally decreased) exposures.
The results of this first set of stepwise regressions clearly show that
common daily activities such as filling one's gas tank, visiting the dry cleaners,
or smoking can lead to significantly increased breath concentrations of toxic
chemicals. A number of occupations (paint, chemicals, plastics, textiles, metal
work, wood processing, service stations, etc.) were implicated in increased
exposure to some chemicals during the day. Household characteristics were
sometimes selected as significant variables on the overnight air samples—
for example, a smoker or chemical worker in the home was associated with
significantly increased exposures to some aromatics.
The strength of the association can be quantified by calculating the value
of eb: this is the ratio of the geometric mean concentration for persons having
the characteristic compared to those not having the characteristic. Thus,
a value of b = 0.69 indicates a two-fold increase, a value of b - 2 3 indicates
an order of magnitude increase. (For a few variables with multiple categories,
such as travel time or frequency of pesticide treatments, the value of eb
is the ratio of the geometric mean of each category to the next lower category.)
Breath. Smoking was the single strongest predictor for increased breath
levels of four of the five aromatics: benzene, styrene, ethylbenzene, and
122
-------
oooooooo
dddddddd
ooooooo ooooooo
ddddddd ddddddd
Cs] CN| ^-~
uj
I
C
01
123
-------
z^>
q>
1
c
0
,«
•S
1
Q.
Uj
y"
5:
Variable
1
Q>
5
<£> *~ 22^2
o o o CM CM cv> oo 9922
ooooooo o o o o
ddddddd dddd
CM >o 10 CQ CT) "* °>
CM ^ ^ ^
tX ^J- Q C*^ CNj Q C^ If) O} O) N»
ddddddd dddd
QQ *«•* GO 00 XJ" ^O * — *O ^J1
Q)
8- £
* -S3 .§
•S 'c S1 -I x"^
. * ^ ^§ "^ * -£
g "Jtj * $6 g-ol |
Q, S Q. 5! Q *^ ^ Q. S.
"§§at^l^ ^flg
1511^1 ill!
u>
c
0
L-
co
O Q)
O C
i: co
^. "•£
' 1
Q) l(~l -C
iS 6-:i
CM PO
0 O "-
O O O
odd
to CM o
o o o
^S
ei.
*
^o i>
a c .
g"t s>
Q. ^3 ^
•§£-2
g to . .
5 -5 o
s5;oo
rxo
5
o5 g- Q
0 ^ ^
S Qj ^O
•^ Q) ^^^
«1 *- ^ ^
Illl
* * *
."§
c
o
^
1
tS
c3$
o
o
d
10
CM
d
\
CM
l_
•§
|
t
co
CD
|
6
-------
t
§
c
o
•S
«
(2
*--
Q.
U.
Uj
V)
iT
^
Variable
Chemical
O
0
ix
*"~
R
o
to
0)
CM
to
Hospital worker
tu
01
N
0)
•§
m,p-Dichlort
CM oo
0 O
0 0
O O)
to M-
05 St-
^ LO
0 0
<- 00
CM «-
T- »-
I
^ OM
Central air cond.
Garage worker
^
10 10
0 0
0 O
O 0)
0) to
0 0
co CM
00 Oo
0 0
i
CM ^j-
to ix
T~ ^
Home pesticide treat.
Car travel (daytime)
to
O
O
10
00
to
st-
o
K
00
o
1
<0
Taxi/truck driver
o
tj *-*
CO
(A
tu
|
is
to
•u
CU
5
S-,
s
.*;
CO
C33
.O
5
3
CO
C
03
i.
Q
A,
j
O
£
0
i
tt\
QJ
.C
^
o
CU
o S 5
•w
^ .i-
CO
03 0 ~
S E
to to
c
;
c
5_ -C
' 'S
" s
C ^
qj ^
§ 0
•Q C
v_
r*»
'Jo "5
"•*• O
c
to
5
.CO
CO
2.
CU
CO
to
S-s
€ OS
.to
^
1
1
C
" »2
I "^
CU
-C
o
-Q tu "5 5:
C
CO
j
: S
=. to
to
•vf
m tfl O ^ —
is "§ ^
.CO
CO
.s
to
c
.0
•tw
03
3
Cr
tu
to
C?
0>
-g
S
03
1
C?
0"
w
to <9
J
+
c
4
.t
.c
I
c
c
-c
+
3 nj
5 -C
O
i i
a ^
o to
- to
y eu
M
M1
ncentration is regressed
okers are likely to be e°
+ ^ E
CO
ii
0) j-.
H
.. «
s-s or..
.to
1
o
E
is
iS
nj
? ^
OJ -P *
exposure.
b sample siz.
°The logarit
those who
<
fc
° 'Standard 6
g
CD
-.
1
QJ
to
C
to
1
.y
-Q
CQ
co to
2. C
"Q CU
Q) C"
»< t
"O 'r-
C fc
'- D
Ts 5
o
ga
to fc*-
^ CU O
03 c t~
"r> ^ to
on of the two groups.
no difference in geomei
ce explained by the mnt
t comparing geometric
ices under pairwise t-te,
,<0 V)
S "~
§.£
1^
D *5
Cj t-
cu tO
^£ •S
eF- value of
1 'Probability
C to m
.to tu ^
§ ^^
D .0" »_
"^ ^ CO
'o 2 >c
9 Proportion
*Not signifi
show sign*
125
-------
**.
1
=5
•8
Si
•§
^
J!
^
"5
C
•C
1
Results3: 0
Regression
gj
•3
1
«o
c
o
.a
<6
o
o
-a
l"
.2 a>
E N ^
111
0
C)
0)
00
Ci
10
ci
00
10
CM
4>
|
.C
Smoker
0)
c
a>
ct
§O O O **• f\ CN Oo ^J" ^j-
ooooooooo
CiCiCiCiCiCiCiciCiC)
§ § q § 8 §
CiciCiCiCiCiCiCi
ci ci d ci ci ci ci ci d ci ci ci c> ci ci ci ci ci
Ci Ci Ci C) Ci Ci Ci C) Ci Ci *~ Ci Ci Ci O Ci Ci Ci
C\ »^
0)
I
I
'
4)
£
-------
^
;i
.5
o
^j
CM
I
Q.
U.
Uj
o"
~>
'ariable
.*
Chemical
CM CM
0 O
M-
O 0)
1 —
CM
-------
!
1
I
N"
^
i
•5
c^
U.
uj
CO
o"
5:
Variable
"to
.0
«)
5
*^
o
O ^ fNj
o o o
odd
to CM
^ CO IO
CN
K 0} O
K 10 «-
odd
05 LO • CM
oi ^ d
to ^ co
»- 0
^.
Job at hospital
Central air cond.
Soda consumption
<6
S:
I
1>
•Ct
s
0
-5
95>
Q.O
E9l
o
.^^ -^
2 •£
5 «
to
C .<2
Q) ^
^ Q)
T> V^
"%
•a 1
X; co
n m
S <
5 *5
to ^
0 ^
o
^r Qi
2 |
| -g
O to
to ^
< .c
tu -•
Q. -i;
*- to
§ 0
1 ^
CD c
MS
-el^
€.g |
"S § §
C Q) °
I'l §8
C C -C tb
to C *- -k
Ol.O 0
O -C: rn K
*J <2 5
"*"!)•*
!? 3 -Q 0
a Q-/O |
= n ^
to i to -c
C £ ^ • 5
-c S-c
^ fe^ *
.to -^- C c
•< Q- o o
^ Q) -53 -C
QJ •> 9> <0
bx + 2 eft, where th
r concentration, and !
ncentration is regress
much benzene as in
, O O to
+ o u to
^ O n (o
m a* 8
II 0 « P
" ° ^ -^
^^
•• •*- R .
•2J ^ .g ^
C 0^ II
•8 E ^
ll^
*v ^ "^ _
;"; Q\
Q) jo
p^ -^ 1 — j;
CQ .Q
to
CJ)
.c
t3
to
0)
^
JK
tu
0
s*.
A-1 for explanation o
-SJ
-Q
^S
s
to
s
o
£
O
«S
Q)
,®
CO
128
-------
**.
1
s
o
^
J>
1
1
t
€
1
jj
1
1
1
5
Q)
•2
i
t
to
°?
w
1
Q.
-
Uj
«"
*
-SJ
•Q
.<0
<
5
^5
.u
00-- ^J-ro^ CNO
dd dddd odd dd
.Q
^ ts 5 10 10 CN 10 !5 ? SK
— d dddd odd dd
toS toSS2 to^g ^
2!
1 £§„ ^ ^
i 4S V.^^-'^ ^s-t^1^ k.
Q)Q. 0)0.."^ (l)^^* Q>c
|.c | . s 1 c 1 -1 1- |l
c
C"
^L. q\ Qj
•M ^ Q) qj ^>
||| || || ||
i- co
§0
0 O
sa
-0,
0 0
^ to
O O
§£
es.
1^
|l
^c^
— CM
d d
K rx
10 co
d d
^
Q.
1
u>
c
0
-9 Q)
CD C
O IB
0 -C
i_ f-i
?|
•D <
Q)
(o
CO
d>
1
t
^
Q^
1
Q)
C
t
(b
p
-v
§2
u
123
-------
1
|
|
i
d.
u-
ki
00
&
*
Variable
Chemical
8
0
oo
•^
CN
o
^
CM
ci
^t-
CN
CM
Q.
0)
C
£
o
-c
.0)
CD
T3
'C
.O
5
O
5
IS.
C)
^}-
is
Jo
O
IS
Oi
C)
•^
CN
S
O
1
-2
Q.
.5
Chloroform
(0.02)
1
ci
10
00
01
o
en
^
Y-^
u.
*-
Q
0
'io
Central
tu
1
c
Q>
|
<
q.0
^
CO
>^
-Q
di
^g
"o
1
C;
CO
0)
.g
^
a
co
1
.to
-^
to
CO
-S;
•2
§
•1
CO
c
c
.0
+-J
to
ta-
ct)
*-
CD
CO
CD
-£
S
0)
-c
+
CO
II
\\
.to
1
0
S
CD
iS
(0
•zene
c
4)
•Q
-C
U
1
5
2
C;
CO
-^
Q)
C
O
CD
to
CD
S1
.to
o
ncentrati
o
CD
o
c S
•2 -c
Z^ +~*
to
53)
C
•6
CD
0)
-c
-c
o
'o
.0
CO
c
-2
o ^
c^
c 5
iss
concentra^
"The logari
on his bre
See footno.
130
-------
m,p-xylene. Employment was the strongest factor for 1,1,1-tnchloroethane,
tetrachloroethylene, and o-xylene.
Daytime air. The daytime personal air exposures reflect the working and
commuting period. Employment in plastics, wood processing, service station/
garages, painting, textiles, metals, scientific laboratories, dye plants, and
hospitals was associated with significantly increased exposures to nine of
the eleven chemicals. Close contact with smokers was the strongest
explanatory variable for daytime exposures to benzene.
Overnight air. Overnight personal air exposures were essentially indoor
samples. A smoker in the home was the strongest determinant of indoor
concentrations of benzene, styrene, ethylbenzene, and /r?,p-xylene.
New Jersey—Summer 1982
A second trip to New Jersey took place in July-August 1982. A subset
of 160 of the original participants was monitored.
Breath. Again, smoking was the strongest determinant of breath values
of benzene, styrene, ethylbenzene, and /7?,p-xylene (Table A-4). Dry cleaners
exposure led to a ten-fold (e239) increase in geometric mean breath levels
of tetrachloroethylene for those exposed.
Overnight personal air. Homes with window air conditioning and circulating
fans (which tend to reduce outdoor ventilation) had significantly higher levels
of five chemicals, and homes with window fans had lower levels of another
chemical (Table A-5). Self-reported exposure to auto exhaust was also
associated with increased exposure to aromatics.
Daytime personal air Employment-related exposures—particularly
metalwork and chemical plant exposures were associated with sharp (up
to e29 = 18-fold) increases in daytime exposure (Table A-6). Auto-related
activities (travel time, visits to service stations) were associated with increased
exposure to aromatics associated with gasoline.
New Jersey— Winter 1983
A subset of 49 participants were monitored in the third season (February
1983) in New Jersey.
Breath. Smoking was again most strongly correlated with breath levels
of benzene and styrene, and was second in importance to solvent exposure
for the xylenes (Table A-7). Other important variables for aromatics exposure
were race (whites having higher values); exposure to dust and particulates;
living in a house for more than 10 years; and, for styrene only, having a
gas furnace.
Chlorinated compounds again had few variables associated with higher
breath levels. Females showed higher exposure to trichloroethylene.
Overnight Personal Air. Gas furnaces were associated with increased levels
of aromatics in homes (Table A-8). Employment, sex (male), and smoking
were also associated with significantly elevated exposures.
Daytime Personal Air. Occupational exposure was implicated in increased
daytime exposure to four aromatic compounds (Table A-9). Solvent exposure
was specifically identified by the regression. Again, gas furnaces were
associated with increased exposures to aromatics. Persons reporting
exposure to auto exhaust showed higher exposures to benzene.
Los Angeles—February 1984
Three visits to California were carried out in 1984. In the first visit, 117
participants were monitored in Los Angeles. Five new chemicals (octane,
decane, undecane, dodecane, and a-pmene) were observed in greater than
25% of all samples.
131
-------
00
01
t*.
QJ
S
3
(0
^
4)
2
m
0)
5
\
"£*
co
CQ
a
3
CO
£
egression
(C.
^
Q)
.">
i
a
•
co S Uj S o S. E
IN
§§
d d
CN05
d d
^o PM
»- rx
— —
3?
5
CO
Emp/oyei
Paint exf.
^
9.
o
d
co
o
PO
d
CD
rx
d
c
Cb
S
CO
i
o
0)
1
en
c
o
CD
o
o
"U
-r
hlorinated \
'hloroform
1.08)
O vJ -—
^
O
§CN
Q
d d
—
*-. PO
d d
^j- ^j-
^ Ol
d d
¥
-2.
"^
-------
\
I
•3J
-Q
(5
\
u.
Uj
OJ
-Q
-tu
-Q
to
1
Chemical
T- CM
0 0
d d
CT) tO
to 10
T- CM
CTI oo
d d
05 to
oo I"N
CM o
S
o
Q.
X
Q)
gj
c
to
_cu
"G *:
^ "O
Q ^
Tetrachloroethylene
(0.15)
o
d
rs
^j-
tN
d
10
IO
^;
i
g
-c
0
m , p -Dichlorobenzene
(0.03)
v>
to
2
-Q
1
,C
Q>
C
01
(Si
C
Q)
-Q
•t
O
%
|
CO
d
\\
* ">
fN >•
Q) -3
•Q b
•C Q
t3 £
Q) *~
0 <
I =
o
o
^; vi
O
fo .0
aj •£
Q. 0)
ot I" *
o ^
^ ^
CQ O Q)
'C: s °
ik *^ QJ
.§ l|
2 *** "G
1 o § c
^o § .!2 .">
Qj ^% ""*'
a Coefficient of the qu>
persons who did not
b Standard error.
€F-value of the compa
d Probability that there
tN
00
0)
V*H.
1
s
3
8
jj
'
i
j_
"5
c
o
2
<
.Si
5
9>
O
1
egression Re
DC
j/i
5
|_
•£
v>
10
J
ci.
u.
Uj
CO
-Q
Variable
Chemical
^
O
o
d
00
00
CM
d
CM
to
d
^
c
.0
1
o
None
Window air
v>
d
0
.Q
Q
O
l_
^
Aromatic
Benzene
Styrene
CM ^
0 0
d d
to ^
10 <*
CM co
d d
00 IN
IO to
d d
1
Auto exhau.
Toxics expo
^
T-s.
d
\\
10
O
o
d
o
00
10
-------
^N,
w
i
|
•Si
Q.
Lj^
Uj
C/3
-Q
Variable
Chemical
00
OO
do
CMPO
do
00 IX
do
l_
.0
."t;
Window air cone
Gas stove
0)
PO
o PO
o o
d d
O ix
oj M-'
10 O
PM PO
d d
IX to
d d
>-
1
•5
Window air cone
Auto exhaust
•S
io
oo
O IO
o o
d d
r\i »-
to r\i
d d
to •<»•
O to
r\i d
i
£
5
Degreasers expo
Window fan
Chloroform
(0.07)
*- •
1
c
o
-Q
Chlorinated Hydrocar
7, 7, 1-Trichloroethane
PM PO
00
d d
O ix
ix r\i
d d
^j- \j.
d d
i
-Q
Gardening
Over 65 years ol
Carbon Tetrachloride
(0.06)
\ —
O
d
to
to
d
to
d
Employed
Trichloroethylene
(0.04)
**.
O
d
to
to
00
d
PO
IX
d
Circulating fan
None
Tetrachloroethylene
(0.04)
m,p-Dichlorobenzene
Oj
.g
"S
^
•<*
^
~9£
See footnotes to Tab
134
-------
fN
1
53
£
1
»J
si
%
\
1
^
1
o
1
Daytime
!
0)
*
c
<2
8
S1
DC
o>
.a
i
Q.
I
CO
(0
^
1
Q.
U,
Uj
C/j
-Q
& 1 §
g QJ .*5 C
I IS 1 1
g. <0 ^ 0 0
* c o s; .< <
-i>2 &fe i^S55 1 iS '^ -t,
II St i*l sin
CT3 -yfo <00^3 O<5'O-S
1 | |fe i|| | | | 1
os KCO ^^s ^ 5 s uj
01
c
o
-Q
k_
- ^ CM
<(§£ ^$ G$ d>0_
*—
O to
o o *-•
OOO
odd
00 00
10 r< to
Y-.
CO 10 0)
»» co CQ
odd
O *- co
IO O CO
d »- CN
tioner
•5
o
u
"$ '<8 ij
> Jt
2 * !
o -5
o "o to
*- c ^
3 s 4J
^ s§
0)
c
•Si
X
^5?
Q.»-
E9
O
d
q
d
lO
Oo
c\j
CO
CN
§
o
CL
2
o>
in
c.
o
5 <"
II
?
^
O
.u
0)
g
•5
g
135
-------
0
§•
£
-3J
.0
Dust/parti
Tetrachloroethylene
(0.04)
*-.
o
d
t\
<£5
05
C\|
d
fj
rx
d
Smoked
Q>
m,p-D/chlorobenzen
(0.04)
8,
c
13
-------
Table A-7. Stepwise Regression Results: Breath—New Jersey,
(Winter 1983)
Chemical
Aromatic Hydrocarbons
Benzene
(R2 = 0.321
Styrene
(0,42)
Ethy/benzene
(0.52)
o-Xylene
(0.42)
m,p-Xy/ene
(0.48)
Variable
Smoker
White
So/vents
Smoker
Gas furnace
White
Old house
Smoker
White
Dust exposure
So/vents
Solvents
Smoker
Old house
White
Old house
So/vents
Smoker
Dust exposure
Chlorinated Hydrocarbons
1,1,1-Trichloroethane Dust exposure
(0.18)
Carbon tetrachloride
(0.28)
Trichloroethylene
(0.17)
Tetrachloroethylene
(0.14)
Chloroform
m,p-Dich/orobenzene
No occup.
exposure
Dust exposure
Female
Old house
None
None
b
1.19
1.06
1.01
1.20
0.60
0.69
-0.70
0.66
0.76
0.68
0.75
0.96
0.58
-0.57
0.69
-0.65
0.76
0.49
0.59
2.00
0.21
1.24
0.88
-0.81
S.E.
0.33
0.42
0.48
0.23
0.23
0.31
0.21
0.21
0.28
0.28
0.31
0.30
0.20
0.21
0.27
0.20
0.30
0.20
0.27
0.62
0.05
0.48
0.38
0.29
F
13
4.6
4.4
27
6.6
5.1
11
9.7
7.6
5.8
5.7
10
8.4
7.4
6.6
11
6.5
5.8
4.8
10
18
6.6
5.4
7.6
P
0.0007
0.01
0.04
0.0001
0.01
0.03
0.002
0.003
0.009
0.02
0.02
0.002
0.006
0.009
0.01
0.002
0.01
0.02
0.03
0.002
0.0001
0.01
0.02
0.008
See footnotes to Table A-4 for explanation of headings.
137
-------
Table A-8. Stepwise Regression Results: Overnight Personal Air-
New Jersey, (Feb. 1983)
Chemical
Aromatic Hydrocarbons
Benzene
(R2 = 0.281
Styrene
(0.25)
Ethylbenzene
(0.14)
o-Xylene
(0.25)
m,p-Xy/ene
10. 15)
Variable
Gas furnace
Male
Smoker
No occ. exposure
Male
Employed
Gas furnace
Gas furnace
Employed
Gas furnace
b
0.68
0.62
0.58
1.19
0.67
0.66
0.73
0.65
0.59
0.70
S.E.
0.27
0.28
0.27
0.42
0.30
0.30
0.27
0.24
0.25
0.24
F
6.5
4.8
4.6
8.2
5.1
4.7
7.1
7.6
5.8
8.4
P
0.01
0.03
0.04
0.006
0.03
0.04
0.01
0.008
0.02
0.006
Chlorinated Hydrocarbons
/, 7, 1-Trichloroethane
Tetrachloroethylene
Carbon tetrachloride
(0.20)
m,p-Dich/orobenzene
Chloroform
Trichloroethylene
(0.27)
None
None
Employed
Non white
None
None
No occ exposure
Male
See footnotes to Table A-4 for explanation
Table A-9. Stepwise
Regression Results
0.64
0.65
2.14
0.91
0.22
0.27
0.54
0.40
8.7
5.8
16
5.2
0.005
0.02
0.0003
0.03
of headings.
: Daytime Personal Air—
New Jersey, (Feb. 1983)
Chemical
Aromatic Hydrocarbons
Benzene
(R2 = 0.11)
Styrene
(0.12)
Ethylbenzene
(0.51)
o-Xylene
(0.54)
Variable
Auto exhaust
Occup. exp.
Occup. exp.
Solvent exp.
Gas furnace
Occup. exp.
Solvent exp.
Gas furnace
b
0.68
0.79
0.99
1.35
0.69
1.06
1.32
0.65
S.E.
0.30
0.33
0.29
0.41
0.26
0.28
0.40
0.25
F
5.2
5.8
12
11
5.7
15
11
6.9
P
0.03
0.02
0.001
0.002
0.02
0.0004
0.002
0.01
138
-------
Table A-9. (continued/
Chemical Variable b S.E
m.p-Xylene Solvent exp. 1.28 0.39 11 0.002
10.50) Occup. exp. 0.85 0.27 10 0.003
Gas furnace 0.68 0.24 7.7 0.008
Chlorinated Hydrocarbons
/, 7, 1-Trichloroethane
Tetrachloroethylene
Carbon tetrachloride
m,p-Dichlorobenzene
Chloroform
Trichloroe th ylene
(0.12)
None
None
None
None
None
No exposure
1.46 0.59 6.2 0.02
See footnotes to Table A-4 for explanation of headings.
Breath. Auto-related variables (exposure to auto exhaust, pumping gasoline,
and visiting a service station) were associated with significantly increased
breath concentrations of benzene, ethylbenzene, o-xylene, /r?,p-xylene,
decane, and undecane (Table A-10) Smoking continued to be the most
important determinant of breath concentrations of benzene, styrene,
ethylbenzene, o-xylene, and m,p-xy\ene; among the new chemicals monitored
in California, only octane appeared to be strongly related to smoking.
Employment was associated with increased exposures to carbon tetrach-
loride, trichloroethylene, styrene, and ethylbenzene.
Overnight personal air. Once again a smoker in the home was the most
important determinant of indoor air concentrations of benzene, styrene,
ethylbenzene, and m,p-xylene (Table A-11). Undecane and dodecane also
showed smoking-related increases, as did 1,1,1-trichloroethane and a-
pinene. Homes with circulating fans rather than air conditioning or exhaust
fans showed elevated levels of six chemicals. Pesticide exposures were
associated with increased indoor air concentrations of five chemicals. Self-
reported exposures to solvents were associated with increased indoor air
concentrations of three chemicals. Gardening (which requires extended
outdoor activity) was associated with significantly reduced overnight personal
exposures to two chemicals.
Daytime personal air. Employment, smoking, and auto-related activities
continued to be significantly related to increased daytime personal exposures
(Table A-12) Females appeared to have increased exposure to
trichloroethylene.
Los Angeles—May 1984
A subset of 52 of the 117 February participants were monitored in May.
Breath. Smoking-related variables continued to significantly increase
breath concentrations of all five aromatics, and also octane, decane, and
139
-------
undecane (Table A-13) Auto travel continued to be important for several
chemicals, as did employment.
Overnight personal air Age (over 65) and race (Hispanic) appeared a number
of times in connection with decreased and increased overnight exposures,
respectively (Table A-14). Smoking in the home was not related to increased
concentrations of any of the aromatics or other hydrocarbons, perhaps due
to more open homes in May with increased air exchange.
Daytime personal air. Auto-related activities and employment were most
important in increased daytime exposures to eight chemicals (Table A-15).
Age (over 65) and gardening again had "protective" effects
Contra Costa—June 1984
The final TEAM trip recruited 71 residents of Antioch and Pittsburg in
Contra Costa County, California
Breath Smoking was again the strongest determinant of breath
concentrations of the five aromatics and octane (Table A-16) Employment
increased exposures to five chemicals Once again females were associated
with higher exposures to trichloroethylene. Children (under 12) were less
exposed, and males more highly exposed, to several chemicals
Overnight personal air. Night-time auto travel resulted in increased
exposures to benzene, ethylbenzene, and the xylenes (Table A-1 7) Children
under 1 2 and gardeners had reduced exposures
Daytime personal air. Auto-related activities were most prominent in
increased exposures (Table A-18).
Table A-10. Stepw/se Regression Resu/ts: Breath—Los Angeles
(Feb. 1984)
Chemical
Aromatic Hydrocarbons
Benzene
(R2 = 0.52J
Styrene
(0.28)
Ethylbenzene
(0.42)
o-Xylene
(0.24)
rr\,p-Xylene
(0.33)
Variable
Smoked
Auto exhaust
Smoked
Dust/part, ex p.
Occup. exp.
Smoked
Dust/part, exp.
Pumped gas
Paint exp.
Smoked
Garage/serv.
station exp.
Smoked
Garage/serv.
station exp.
Clean, material
Dust/part, exp.
b
1.85
0.63
1.56
1.55
0.92
1.37
0.81
0.88
0.68
0.82
0.50
1.02
0.67
0.53
0.45
S.E.
0.27
0.32
0.34
0.45
0.37
0.21
0.28
0.32
0.29
0.15
0.18
0.16
0.19
0.23
0.22
F
47
3.9
21
12
6.1
43
8.3
7.8
5.4
28
6.8
40
12
5.3
4.3
P
0.0001
0.05
0.0001
0. 0008
0.02
0.0001
0.005
0.006
0.02
0.0001
0.006
0.0001
0.0007
0.02
0.04
Other Hydrocarbons
Octane
(0.10)
Smoked
140
0.83 0.24 12 0.0009
-------
Table A-10. (continued)
Chemical
Decane
(0.07)
Undecane
(0.23)
Dodecane
(0.11)
cc-Pinene
Variable
Clean, material
Garage/serv.
station ex p.
Unexposed
nonsmoker
Dust/part, ex p.
Garage/serv.
station exp.
'Employed
Auto exhaust
Gas stove
Window fan
Dust/part, exp.
None
b
1.09
0.83
-0.89
1.13
0.74
-0.63
-0.72
0.54
0.48
0.43
S.E.
0.44
0.35
0.27
0.35
0.30
0.27
0.31
0.26
0.15
0.18
F
6.2
5.6
11
11
6.2
5.4
5.3
4.3
11
5.8
P
0.01
0.02
0.001
0.001
0.01
0.02
0.02
0.04
0.002
0.02
Chlorinated Hydrocarbons
Chloroform
(0.17)
1, 1, 1-Trichloroethane
(0.20)
Carbon tetrachloride
(0.04)
Trichloroethylene
(0.06)
Tetrachloroethylene
(0.19)
p-Dichlorobenzene
Auto exhaust
Adult
Pesticide exp.
Adult
Smoked
Painter
Employed
Smoked
Gas stove
None
See footnotes to Table A-4 for explanation
Table A-11. Stepwise
1.04
0.86
-1.72
0.77
0.77
0.45
0.91
0.63
-0.53
0.40
0.33
0.69
0.25
0.26
0.22
0.34
0.17
0.17
6.8
6.8
6.3
9.6
9.0
4.1
7.1
14
9.5
0.01
0.01
0.01
0.002
0.003
0.04
0.009
0.0003
0.003
of headings.
Regression Results: Overnight Personal Air—
Los Angeles (Feb. 1984)
Chemical
Aromatic Hydrocarbons
Benzene
(Ft2 = 0.18)
Styrene
(0.17)
Ethylbenzene
(0.21)
Variable
Smoker in home
Exposed/daytime
Circulating fan
Smoker in home
Circulating fan
Hispanic
Smoker in home
Circulating fan
Exposed/daytime
So/vent exp.
141
b
0.41
-0.46
0.31
0.70
0.61
-0.65
0.42
0.42
-0.39
0.43
S.E.
0.12
0.15
0.13
0.19
0.20
0.26
0.12
0.13
0.15
0.20
F
11
10
5.9
13
9.0
6.0
12
10
7.0
4.6
P
0.001
0.002
0.02
0.0004
0.003
0.02
0.0009
0.002
0.01
0.03
-------
Table A- 1 1. (continued)
Chemical
o-Xylene
(0.19)
m,p-Xy/ene
(0.12)
Other Hydrocarbons
Octane
(O.JO)
Decane
(0.05)
Undecane
(0.12)
Dodecane
(0.17)
a-Pinene
(0.10)
Variable
Circulating fan
Smoker in home
Exposed/daytime
Solvent ex p.
Smoker in home
Pesticide ex p.
Circulating fan
Solvent ex p.
Dust/part, exp.
Smoker in home
Gardening
Pesticide exp.
Gardening
Pesticide exp.
Smoked at night
Circulating fan
Hispanic
Smoker in home
b
0.38
0.34
0.35
0.40
0.32
0.65
0.48
0.61
0.99
0.38
-0.44
0.86
-0.78
1.14
0.41
0.47
-0.58
0.39
S.E.
0.12
0.12
0.14
0.18
0.11
0.26
0.17
0.25
0.41
0.16
0.19
0.40
0.20
0.42
0.20
0.20
0.26
0.19
F
9.8
8.6
6.6
4.8
9.1
6.3
8.2
5.9
5.7
5.5
5.1
4.8
15
7.3
4.1
5.7
5.1
4.4
P
0.002
0.004
0.01
0.03
0.003
0.01
0.005
0.02
0.02
0.02
0.03
0.03
0.0002
0.008
0.04
0.02
0.03
0.04
Chlorinated Hydrocarbons
Chloroform Pesticide exp.
(0.05)
0.92 0.39 5.4 0.02
1,1,1-Trichloroethane
(0.08)
Carbon tetrachloride
Trichloroethylene
(0.09)
Tetrachloroethylene
(0.05)
p-Dich/orobenzene
Smoker in home 0.59 0.19 9.5 0.002
None
Pesticide exp.
So/vent exp.
None
2.23 0.69 10 0.002
0.58 0.25 5.2 0.02
See footnotes to Table A-4 for explanation of headings.
Table A-12. Stepwise Regression Results: Daytime Personal Air-
Los Angeles (Feb. 1984)
Chemical
Variable
S.E.
Aromatic Hydrocarbons
Benzene Auto travel 0.0026 0.0012 4.6 0.03
(R2 = 0.04)
142
-------
TableA-12. (continued)
Chemical
Styrene
10.16)
Ethylbenzene
(0.18)
o-Xylene
(0.18)
m,p-Xylene
(0. 15)
Other Hydrocarbons
Octane
Decane
(0.20)
Undecane
(0.13)
Dodecane
ct-Pinene
(0.06)
Variable
Toxics, exp.
Smoked/daytime
Clean, materials
Smoked/daytime
Travel time
Hispanic
Female
Smoked/daytime
Employed
Solvent exp.
Paint exp.
Hispanic
Smoker in home
None
Employed
Paint exp.
Pumped gas
Pumped gas
Paint exp.
None
Gas stove
b
1.16
0.55
0.67
0.42
0.0036
-0.43
0.40
0.45
0.41
0.52
0.53
-0.41
0.28
0.77
1.11
1.03
1.02
0.82
-0.58
S.E.
0.35
0.18
0.23
0.18
0.0015
0.21
0.16
0.19
0.18
0.25
0.19
0.18
0.13
0.27
0.39
0.41
0.33
0.31
0.22
F
11
9.1
8.7
5.8
5.6
4.0
5.8
5.5
5.4
4.4
7.4
5.4
5.0
8.2
8.0
6.4
10
6.9
6.6
P
0.001
0.003
0.004
0.02
0.02
0.05
0.02
0.02
0.02
0.04
0.008
0.02
0.03
0.005
0.006
0.01
0.002
0.01
0.01
Chlorinated Hydrocarbons
Chloroform None
1, 1, 1-Trichloroethane
(0.16)
Carbon Tetrachloride
(0.09)
Trichloroethylene
(0.06)
Tetrachloroethylene
(0.13)
p-Dichlorobenzene
Smoked/daytime
Toxics exp.
Toxics exp.
Female
Hispanic
Toxics exp.
None
0.88
1.59
0.98
0.96
-0.71
1.05
0.26
0.52
0.30
0.36
0.24
0.38
11
9.5
11
7.2
8.9
7.6
0.001
0.003
0.001
0.008
0.004
0.007
See footnotes to Table A-4 for explanation of headings.
143
-------
Table A-13. Step wise Regression Results: Breath—Los Angeles
(May 1984)
Chemical
Aromatic Hydrocarbons
Benzene
(Ft2 = 0.34)
Styrene
(0.40)
Ethylbenzene
(0.50)
o-Xylene
(0.35)
m,p-Xy/ene
(0.46)
Other Hydrocarbons
Octane
(0.28)
Decane
(0.10)
Undecane
(0.21)
Dodecane
(0.14)
a-Pinene
(0.09)
Variable
Employed
Smoked
Smoked
Old
Smoked
Employed
Smoked
Employed
Smoked
Employed
Smoker in home
Male
Smoker in home
Pesticide exp.
Smoker in home
Auto travel
Auto travel
Auto exhaust
b
2.55
1.67
2.59
- 1.27
1.75
0.81
1.19
0.88
1.27
0.69
1.10
0.58
0.72
1.42
0.78
0.007
0.006
-0.91
S.E.
0.62
0.70
0.51
0.60
0.31
0.27
0.34
0.29
0.26
0.22
0.28
0.28
0.30
0.58
0.36
0.003
0.002
0.41
F
17
5.7
26
4.4
31
9.0
12
9.1
24
9.5
16
4.3
5.6
6.1
4.6
4.2
7.8
5.0
P
0.0001
0.02
0.0001
0.04
0.0001
0.004
0.001
0.004
0.0001
0.004
0.0002
0.04
0.02
0.02
0.04
0.05
0.007
0.03
Chlorinated Hydrocarbons
Chloroform
1, 1, 1-Trichloroethane
(0.22)
Carbon Tetrachloride
Trichloroethylene
Tetrachloroethylene
(0.13)
p-D/ch/orobenzene
None
High exp.
Old
None
None
Employed
None
1.59
- 1.46
0.76
0.53
0.74
0.29
9.0
3.9
7.0
0.004
0.05
0.01
See footnotes to Table A-4 for explanation of headings.
144
-------
Table A-14. Stepwise Regression Results: Overnight Personal Air-
Los Angeles (May 1984)
Chemical
Aromatic Hydrocarbons
Benzene
Styrene
(R2 = 0.18)
Ethylbenzene
(0.20)
o-Xylene
{0.32)
m,p-Xy/ene
(0.25)
Other Hydrocarbons
Octane
(0.14)
Decane
(0.17)
Undecane
(0.19)
Dodecane
(0.27)
a-Pinene
Variable
None
Circulating fan
High exp. act.
Hispanic
Old
'Old
Hispanic
Hispanic
Old
Hispanic
Hispanic
Hispanic
Pesticide exp.
Pesticide exp.
Old
None
Chlorinated Hydrocarbons
Chloroform High exp. act.
(0.27) Smoker in home
Employed
1, 1, 1~Trichloroethane
Carbon tetrachloride
(0.12)
Trichloroethylene
Tetrachloroethylene
m,p-Dichlorobenzene
None
Old
None
None
None
b
1.08
-0.99
1.04
- 1.08
-1.85
1.15
0.95
-1.03
0.90
1.42
1.07
1.10
1.31
-0.97
1.46
1.07
-0.89
-0.45
S.E.
0.44
0.43
0.41
0.47
0.50
0.43
0.34
0.39
0.32
0.46
0.46
0.50
0.45
0.48
0.40
0.40
0.43
0.17
F
6.1
5.2
6.4
5.2
14
7.1
8.0
7.0
8.2
9.6
5.5
4.8
8.4
4.2
13
7.1
4.2
7.0
P
0.02
0.03
0.01
0.03
0.0005
0.01
0.007
0.01
0.006
0.003
0.02
0.03
0.006
0.05
0. 0008
0.01
0.05
0.01
See footnotes to Table A-4 for explanation of headings.
145
-------
Table A-15. Stepwise Regression Results: Daytime Personal Air-
Los Angeles (May 1984)
Chemical
Aromatic Hydrocarbons
Benzene
(R2 = 0.45;
Styrene
(0.40)
Ethylbenzene
(0.33)
o-Xylene
(0.25)
rr\,p-Xy/ene
(0.35)
Other Hydrocarbons
Octane
(0.12)
Decane
(0.13)
Undecane
Dodecane
(0.21)
Q-P/nene
(0.08)
Variable
Auto travel
Odor. chem. exp.
Pumped gas
Old
Gardening
Auto travel
Smoked
Gas stove
Pumped gas
Employed
Pumped gas
Auto travel
Pumped gas
Employed
Auto travel
Pumped gas
Old
None
Gardening
Pesticide exp.
Gas stove
b S.E.
0.007 0.002
0.68 0.27
0.72 0.36
-1.72 0.60
1.34 0.50
0.01 0.004
1.24 0.49
-0.92 0.44
1.66 0.51
0.86 0.32
1.96 0.72
0.009 0.004
1.12 0.47
0.69 0.31
0.005 0.003
1.33 0.51
-1.37 0.50
- 1.43 0.42
1.05 0.45
- 1.03 0.49
F
15
6.3
4.0
8.3
7.2
6.8
6.5
4.3
10
7.3
7.4
5.8
5.7
5.0
4.1
6.8
7.6
12
5.3
4.4
P
0.0003
0.02
0.05
0.006
0.01
0.01
0.01
0.04
0.002
0.01
0.009
0.02
0.02
0.03
0.05
0.01
0.008
0.001
0.03
0.04
Chlorinated Hydrocarbons
Chloroform
1, 1, 1-Trichloroethane
(0.08)
Carbon tetrachtoride
(0.11)
Trichloroethylene
Tetrachloroethylene
(0.10)
m,p-Dich/orobenzene
None
Old
Employed
None
Employed
None
- 1.23 0.59
0.42 0.17
0.99 0.41
4.3
5.8
5.7
0.04
0.02
0.02
See footnotes to Table A-4 for explanation of headings.
146
-------
Table A-16. Stepwise Regression Results: Breath—Contra Costa
(June 1984)
Chemical
Aromatic Hydrocarbons
Benzene
(0.53)
Styrene
(0.42)
Ethylbenzene
(0.47)
o-Xylene
(0.43)
m,p-Xylene
(0.46)
Other Hydrocarbons
Octane
(0.16)
Decane
(0.07)
Undecane
(0.09)
Dodecane
a-Pinene
(0.08)
Variable
Smoked
Window fan
Smoked
Male
Child
Gardening
Smoked
Male
Window fan
Male
Smoked
Window fan
Smoked
Male
Window fan
Child
Smoked
Employed
Cleanser exp.
None
Gas stove
Circulating fan
b
2.70
1.44
1.14
0.84
-1.27
-0.74
1.48
1.06
0.69
1.07
1.03
0.76
1.32
1.13
0.73
- 1.08
1.00
0.56
-0.79
-1.10
-O.76
S.E.
0.39
0.37
0.31
0.29
0.47
0.29
0.30
0.27
0.28
0.26
0.28
0.26
0.33
0.31
0.30
0.49
0.29
0.26
0.32
0.43
0.31
F
47
15
13
8.5
7.5
6.7
25
15
6.2
17
14
8.3
16
14
5.9
4.9
12
4.5
6.1
7.3
5.9
P
0.0001
0.0002
0.0005
0.005
0.008
0.01
0.0001
0.0003
0.02
0.0001
0.0005
0.005
0.0001
0.0005
0.02
0.03
0.0009
0.04
0.02
0.009
0.02
Chlorinated Hydrocarbons
Chloroform None
1, 1, 1-Trichloroethane
(0.18)
Carbon tetrachloride
(0.17)
Trich/oroeth y/ene
(0.18)
Tetrachloroethylene
(0.09)
m , p -Dichlorobenzene
(0.08)
Window fan
Solvent exp.
Child
Circulating fan
Solvent exp.
Female
High exp. act.
Painter
1.75
2.46
-0.72
-0.44
1.91
0.65
1.23
4.19
0.61
1.16
0.28
0.59
0.31
0.47
1.54
8.1
4.5
6.8
10
4.4
6.8
7.4
0.006
0.04
0.01
0.002
0.04
0.01
0.008
See footnotes to Table A-4 for explanation of headings.
147
-------
Table A-17. Stepwise Regression Results: Overnight Personal Air-
Contra Costa (June 1984)
Chemical
Aromatic Hydrocarbons
Benzene
(0.53)
Styrene
(0.13)
Ethylbenzene
(0.06)
o-Xylene
(0.06)
m,p-Xy/ene
(0.06)
Other Hydrocarbons
Octane
(0.10)
Decane
(0.15)
Undecane
(0.12)
Dodecane
a-Pinene
(0.09)
Variable
Auto travel
(night)
Hispanic
Smoked/night
Auto travel
(night)
Auto travel
(night)
Auto travel
(night)
Smoker in home
Auto travel
(night)
Pesticide
treatment
Circulating fan
Circulating fan
Hobby: models
None
Smoker in home
b S.E.
0.004 0.002
1.14 0.53
0.63 0.31
0.004 0.002
0.004 0.002
0.003 0.002
-0.48 0.21
0.004 0.002
-0.38 0.16
-0.56 0.25
-0.65 0.25
1.21 0.59
-0.63 0.25
F P
4.3 0.04
4.6 0.04
4.2 0.04
4.0 0.05
4.4 0.04
4.4 0.04
5.2 0.02
4.6 0.04
5.6 0.02
5.3 0.02
7.1 0.01
4.3 0.04
6.5 0.01
Chlorinated Hydrocarbons
Chloroform None
1, 1, 1-Trichloroethane
(0.08)
Carbon tetrachloride
(0.09)
Trichloroethylene
(0.06)
Tetrach/oroethylene
(0.08)
m,p-D/ch/orobenzene
(0.24)
Circulating fan
Gardening
Gardening
Circulating fan
Gas stove
Hispanic
Child
1.75 0.61
-0.57 0.22
-0.69 0.33
-0.73 0.29
2.22 0.58
2. 14 0. 73
-1.25 0.61
8.1 0.006
6.6 0.01
4.3 0.04
6.2 0.02
14 0.0003
8.5 0.005
4.2 0.04
See footnotes to Table A-4 for explanation of headings.
148
-------
Table A-18. Stepwise Regression Results: Daytime Personal Air-
Contra Costa (June 1984)
Chemical
Variable
Aromatic Hydrocarbons
Benzene Window/ceiling fan
(0.24) Auto travel night
Styrene
(0.18)
Ethylbenzene
(0.32)
o-Xylene
(0.29)
m,p-Xylene
(0.17)
Other Hydrocarbons
Octane
(0.19)
Decane
(0.43)
Undecane
(0.30)
Dodecane
(0.15)
a-Pinene
(0.06)
Employed
Auto exhaust exp.
Window/ceiling fan
Service station exp.
Window/ceiling fan
Service station exp.
Service station exp.
Auto exhaust exp.
Hispanic
Service station exp.
Employed
Hispanic
Service station exp.
So I vent exp.
Employed
Soi 'vent exp.
Dust/part, exp.
b
0.56
0.003
0.57
0.59
0.77
0.62
0.69
0.57
0.72
0.57
1.10
0.62
0.94
1.43
0.72
2.04
0.76
2.08
1.02
S.E.
0.16
0.001
0.26
0.29
0.19
0.21
0.18
0.20
0.27
0.28
0.38
0.22
0.25
0.45
0.27
0.52
0.29
0.62
0.50
F
13
6.0
4.8
4.2
16
8.7
14
8.0
6.9
4.1
8.3
7.9
14
10
7.2
15
7.0
11
4.2
P
0.0007
0.02
0.03
0.04
0.0001
0.004
0.0004
0.006
0.01
0.05
0.005
0.006
0. 0004
0.002
0.009
0.0002
0.01
0.001
0.05
Chlorinated Hydrocarbons
Chloroform None
1, 1, 1-Trichloroe thane
(0.17)
Carbon tetrach/oride
Trichloroe thy/ene
(0.12)
Te trachloroe thy/ene
m,p-Dichlorobenzene
(0.35)
Solvent exp.
None
So/vent exp.
None
Painter
Pesticide treat.
Occup. exp.
High dust/part, exp.
2.04
2.24
5.03
-0.73
1.17
1.21
0.55
0.77
1.54
0.24
0.44
0.49
14
8.4
11
8.9
7.1
6.1
0.0005
0.005
0.002
0.004
0.01
0.02
See footnotes to Table A-4 for explanation of headings.
149
-------
Appendix B
Effect of Outdoor Air on Measures of Personal
Exposure in New Jersey and California
As discussed in the text, an analysis was made of the effect of outdoor
air on the exposures of New Jersey and California subjects. Only those
persons who had outdoor measurements made near their homes were
included in the analysis. In New Jersey, the number of persons with outdoor
measurements was 85 in the fall of 1981, 71 in the summer of 1982, and
9 in the winter of 1983. (Because of the small number of outdoor
measurements in the winter season, no analysis was made of those results.)
In Los Angeles, 24 homes had outdoor measurements in both February and
May of 1984. In Antioch and Pittsburg, 10 homes had outdoor measurements
in June 1984.
The method of analysis was stepwise regression, using the model described
in the text1
In Cm = a + b In Cout + Ic,q,
where Cm = indoor concentration (or breath concentration or daytime
personal air concentration)
Cout = outdoor air concentration
Q, = questionnaire variables, generally indexed to 0 or 1
c, = coefficients of the q.
Using the rule that the number of variables should not be more than about
one-quarter of the number of observations in a stepwise regression, the
New Jersey data on 85 and 71 homes allowed about 20 variables to be
included in the regression, while the California data allowed only six variables
to be included for Los Angeles and only 3 for Antioch/Pittsburg. The larger
number of homes in New Jersey made it possible to carry out the regression
on all three measures of personal exposure (breath, daytime personal air,
and overnight personal air); in California, however, only the overnight personal
air in the residences, which is the most likely to be influenced by outdoor
air near the residence, was employed in the regression.
As in other stepwise regressions, entry and retention values were set
at p < 0 1 5. For the New Jersey data, the final model included only variables
for which p < 0.05. For the California data, because of the smaller number
of homes, a cutoff value of 0 10 was used to allow detection of possibly
significant variables. (If the reader desires to use p < 0.05, he can of course
refer to the listed p-values to identify those variables meeting that criterion.)
The fall 1981 results for New Jersey are summarized in Tables B-1 through
B-3. For overnight personal air, which corresponds to indoor air in the
residence for most of the subjects, only three chemicals showed a significant
influence of outdoor air- benzene, carbon tetrachloride, and trichloroethylene
(Table B-1). Five chemicals showed an effect of daytime outdoor air on daytime
150
-------
Table B-1. Stepwise Regression Results For 87 New Jersey Homes
with Outdoor Measurements: Overnight Personal Air-
Fall 1981
Chemical
Variable
Aromatic Hydrocarbons
Benzene Outdoor benzene
(night)
(R2 = 0.19)
Styrene (0.041
Ethylbenzene
(0.10)
o-Xylene
(0.15)
m,p-Xy/ene
(0.12)
Smoker in home
Smoker in home
Smoker in home
Gas furnace
Smoker in home
Chlorinated Hydrocarbons
Chloroform Chem. worker
(0.15) in home
Auto exhaust
1,1,1-Trich/oroethane Male
(0.10)
Carbon tetrach/oride Outdoor carbon
0.44 0.10 19 0.0001
0.59
0.73
0.49
0.44
(0.19)
Trich/oroethy/ene
(0.12)
tetrach/oride (night)
Outdoor trich/oro-
ethylene (night)
Electric stove 0.86
Tetrach/oroethylene Male 0.87
(0.12) Exposed to cleansers 0.90
p-Dich/orobenzene Hobby: painting —1.18
(0.06)
0.27
0.24
0.20
0.19
4.7 0.03
9.1 0.003
6.0 0.02
5.4 0.02
0.70 0.21 11
0.30 0.07 19
0.26 0.08 13
0.29
0.34
9.1
6.9
0.001
2.48 1.00 6.2 0.01
1.08 0.47 5.3 0.02
0.94 0.31 9.3 0.003
0.0001
0.0005
0.38 4.0 0.03
0.003
0.01
0.52 5.2 0.02
^Coefficient of the questionnaire variable or of the logarithm of the outdoor
concentration, e.g.. In (indoor benzene) = a + 0 44 In (outdoor benzene)
Similarly, indoor air in homes of smokers had e°59~7 8 times as much
styrene as homes with no smokers
b Standard error.
c F-value of the comparison of the two groups.
dProbability that there is no difference in geometric means of the two
groups.
personal air exposures (Table B-2). Since many subjects were away from
their homes for much of the daytime period, it is somewhat unexpected
to have more chemicals showing an effect of outdoor air concentrations
on breath levels (Table B-3)
The summer 1982 results for New Jersey are summarized in Tables B-
4 through B-6. By contrast to the fall results, eight chemicals showed an
effect of outdoor air on overnight indoor concentrations (Table B-4) compared
to only three for daytime personal air (Table B-5) and none for breath (Table
151
-------
Table B-2. Stepwise Regression Results For 87 New Jersey Homes
with Outdoor Measurements: Daytime Personal Air-
Fall 1981
Chemical
Variable
S.E.
Aromatic Hydrocarbons
Benzene Employed
(R2 = 0.10)
Styrene (0.08)
1.21 0.39 9.7 0.003
So/vent exposure 1.52 0.57 7.0 0.01
Ethylbenzene
(0.25)
o-Xylene
10.33)
m,p-Xy/ene
(0.39)
So/vent exposure
Smoker in home
So/vent exposure
Smoker in home
Employed
Solvent exposure
Employed
Smoker in home
2.03
0.63
1.81
0.60
0.51
1.77
0.79
0.56
0.43 22
0.28 5.2
0.38 23
0.24 6.5
0.23 4.8
0.36 24
0.22 13
•0.23 6.2
Chlorinated Hydrocarbons
Chloroform Chem. worker
(0.06) in home
0.0001
0.02
0.0001
0.01
0.03
0.0001
0.0006
0.01
2.36 1.01 5.5 0.02
1, 1, 1-Trichloroethane
(0.26)
Carbon tetrachloride
(0.07)
Trichloroethylene
(0.31)
Tetrachloroethylene
(0.43)
p-Dich/orobenzene
(0.14)
Employed
Daytime outdoor
levels
Daytime outdoor
levels
Daytime outdoor
levels
Solvent exposure
Cleansers exposure
Daytime outdoor
levels
Employed
Daytime outdoor
levels
1.69
0.21
0.22
O.41
1.34
0.69
0.58
0.86
0.42
0.37
0.09
0.09
0.08
0.36
0.27
0.08
0.25
0.12
21
4.8
6.3
25
14
6.4
49
12
13
0.0001
0.03
0.01
0.0001
0.0004
0.01
0.0001
0.0009
0.0005
See footnotes to Table B-1 for explanation of headings.
B-6) The results for the air samples are explainable by the argument above
that outdoor air near the residence should have more effect on exposure
in the residence than on exposure elsewhere. The larger number of chemicals
having an effect of outdoor air concentrations on indoor air concentrations
in the summer may be ascribed to increased air exchange in the summer
due to opening windows at night. The lack of observable results on breath
values may be due to the decreased precision of the summer values resulting
from the contamination incident discussed in the text.
The California results are summarized in Tables B-7 through B-9. Six of
16 chemicals showed an effect (p < 0.10) of outside air on indoor air during
152
-------
Table B-3. Stepwise Regression Results For 87 New Jersey Homes
with Outdoor Measurements: Breath—Fall 1981
Chemical
Variable
b
S.E
F
P
Aromatic Hydrocarbons
Benzene
!R2 = 032)
Styrene (0 16)
Ethylbenzene
(032)
o-Xy/ene
(025)
m,p-Xy/ene
(022)
Smoked
Dry cleaner
exposure
Auto exhaust
exposure
Daytime outdoor
levels
Paint exposure
Old O65)
Smoker in home
Paint exposure
Electric stove
Smoker in home
Daytime outdoor
levels
Paint exposure
Odorous chemical exp
Electric stove
Paint exposure
Daytime outdoor
levels
Solvent exposure
Smoker in home
Old house O10 yrs) -
1 31
3 53
1 52
0 21
1.60
0 99
0 73
0 97
1.20
0 66
0 19
1 04
0 74
1 26
1 28
0.28
0 98
O 58
0 47
0.38
1 22
0 55
0.09
0.70
049
0 26
0.44
0.41
0.23
007
0 41
0 29
0 38
0 41
0.10
0.33
O 27
0 21
12
83
7 7
5.6
5.1
4. 1
80
50
8 6
8.0
7 7
6.5
6 3
11
9.8
7.0
89
7 6
4 8
0.0008
0.005
0.007
0.02
0.03
0 05
0.006
0.03
0.004
0.006
0.007
001
0.01
0.001
0.002
0 01
0004
0 O07
0 03
Chlorinated Hydrocarbons
Chloroform
1, 1, 1-Trichloroethane
(026)
Carbon tetrachloride
Trich/oroethy/ene
(017)
Tetrachloroethy/ene
(018)
p-Dichlorobenzene
(026)
None
Daytime outdoor
levels
None
Smoked
Daytime outdoor
levels
Employed
Daytime outdoor
levels
Garage/service
station exposure
Dry cleaners exposure
Daytime outdoor
levels
Smoker in home
0 24
0 68
0 17
0 68
0 21
0 67
1 35
0 40
0 58
0.09
0 22
0 08
0.21
007
0.31
0 67
0.08
0.29
6.7
9 1
4 7
11
10
4 7
4.0
22
4.1
001
0 004
0 03
O.O02
0.002
0 03
005
00001
005
See footnotes to Table B-1 for explanation of headings.
153
-------
Table B-4. Stepwise Regression Results For 71 New Jersey Homes
with Outdoor Measurements: Overnight Personal Air-
Summer 1982
Chemical
Variable
S.E.
Aromatic Hydrocarbons
Benzene None
Styrene
(R2 = 0.77;
Ethylbenzene
(0.29)
Q-Xylene
(0.27>
m,p-Xylene
(0.27)
Outdoor level
Male
Outdoor level
High-exposure
activity
Paint exposure
Male
Outdoor level
High- exposure
activity
Gas furnace
Outdoor level
High-exposure
activity
Gas furnace
Chlorinated Hydrocarbons
Chloroform None
1,1,1-Trichloroethane Service station
(0.08)
Carbon tetrachloride Outdoor level
(0.10)
Trichloroethylene
(0.13)
Outdoor level
Tetrachloroethylene Outdoor level
(0.19)
High-exposure
activity
p-D/ch/orobenzene Outdoor level
(0.19)
Auto exhaust
0.28
0.56
0.26
1.34
-1.34
0.52
0.21
1.29
0.19
1.42
0.08 12 0.001
0.24 5.6 0.02
0.07
0.50
0.50
0.24
13 0.0005
7.2 0.009
7.0 0.01
4. 7 0.03
0.06 12 0.0009
0.47 7.5 0.008
-0.51 0.22 5.4 0.02
0.06
0.49
10 O.OO2
8.2 0.006
-0.57 0.23 6.0 0.02
-1.08 0.45 5.8 0.02
0.28 0.10 7.8 O.OO7
0.26 0.08 10 0.002
0.36 0.10 13 0.0007
1.61 0.80 4.0 O.05
0.82 0.23 13 0.0007
1.36 0.60 5.2 0.03
See footnotes to Table B-1 for explanation of headings.
the winter of 1984 in Los Angeles (Table B-7) compared to five chemicals
in the spring in Los Angeles (Table B-8) Only tetrachloroethylene showed
an effect in Antioch/Pittsburg (Table B-9)
154
-------
Table B-5. Stepwise Regression Results For 71 New Jersey Homes
with Outdoor Measurements: Daytime Personal Air-
Summer 1982
Chemical
Variable
b S.E.
Aromatic Hydrocarbons
Benzene Gas furnace
(R2 = 0.07)
Styrene
(0.12)
Ethylbenzene
(0.31)
o-Xylene
(0.19)
m,p-Xylene
(0.27)
Paint
Outdoor level
Gas furnace
Employed
Auto travel time
Auto travel time
Sol vents
So/vents
Auto travel time
High-exposure
activity
Pesticide exposure
1.16 0.56 4.3 0.04
1.88 0.65 8.4 0.005
0.31 0.09 10 0.002
1.07 0.34 9.6 0.003
0.85 0.38 5.0 0.03
0.33 0.16 4.3 0.04
0.42
1.30
0.16
0.50
7.2
6.7
0.009
0.01
1.85 0.54
0.57 0.17
2. 14 0. 74
12 0.001
11 0.001
8.3 0.005
1.32 0.54 5.9 0.02
Chlorinated Hydrocarbons
Chloroform Outdoor level -0.16 0.05 12 0.001
(0.26) Old house O10 yrs.) - 1.00 0.38 6.8 0.01
Hobby: gardening 0.97 0.46 4.5 0.04
1, 1, 1-Trichloroethane
(0.15)
Carbon tetrach/oride
(0.13)
Trichloroethylene
Tetrachloroethylene
(0.35)
p-Dich/orobenzene
(0.12)
Employed
Gas furnace
Employed
Old house O10
None
Outdoor level
1.
0.
0.
yrs.) - 0.
0.
35
99
41
37
41
Dry cleaners exposure 2. 2 5
Employed
1.
26
0.
0.
0.
0.
0.
0.
0.
51
48
19
18
08
94
43
7.
4.
4.
4.
24
5.
8.
1
2
8
4
7
7
0.01
0.05
0.03
0.04
0. 000 1
0.02
0.004
See footnotes to Table B-1 for explanation of headings.
155
-------
Table B-6. Stepwise Regression Results For 71 New Jersey Homes
with Outdoor Measurements: Breath— Summer 1982
Chemical
Variable
Aromatic Hydrocarbons
Benzene Smoked
(R2 = 0.29) Pesticide treatment
Styrene
(0.09)
Ethylbenzene
(0.2V
o-Xylene
(0.08)
m,p-Xy/ene
(0.23)
Smoked
Smoked
Paint exposure
Employed
Hobby: gardening
Auto travel time
Employed
3.
1.
1.
1.
3.
1.
__ 1
0.
1.
b
04
24
11
95
26
48
72
68
44
S.E.
0.
0.
0.
0.
1.
0.
0.
0.
0.
82
59
45
59
35
66
76
30
69
14
4.
6.
11
5.
5.
5.
5.
4.
F
4
2
8
0
2
0
3
P
0. 0006
0.04
0.02
0.002
0.02
0.03
0.03
0.03
0.04
Chlorinated Hydrocarbons
Chloroform None
1, 1, 1-Trichloroethane
10.14)
Non white
Auto travel time
1.41
0.55
0.59
0.24
5.
5.
6
2
0.02
0.03
Carbon tetrachloride None
Tnchloroethylene None
Tetrachloroethylene Auto travel time 0.49 0.21 5.7 0.02
(0.09)
p-Dichlorobenzene None
See footnotes to Table B-1 for explanation of headings.
156
-------
Table B-7. Stepwise Regression Results For 24 Homes with Outdoor
Measurements: Overnight Personal Air—Los Angeles
(February 1984)
Chemical Variable b S.E. F p
Aromatic Hydrocarbons
Benzene Smoker in home 0.41 0.24 2.9 0.10
(R2 = 0.11)
Styrene None
Ethylbenzene Outdoor concentration 0.19 0.08 4.9 0.04
(0.18)
o-Xylene None
m,p-Xylene Smoker in home 0.41 0.21 3.7 0.07
(0.14)
Chlorinated Hydrocarbons
Chloroform None
1,1,1-Trich/oroethane Outdoor concentration 0.38 0.14 7.1 0.01
(0.24)
Carbon tetrachloride None
Trichloroethy/ene None
Tetrachloroethylene Time in car 0.02 0.01 4.9 0.04
(0.43) Outdoor concentration 0.23 0.11 4.6 0.04
Smoker in home 0.70 0.36 3.8 0.07
m,p-Dich/orobenzene Outdoor concentration 0.98 0.27 13 0.002
(0.37)
Aliphatic Hydrocarbons
Decane Outdoor concentration 0.42
(0.17)
Dodecane
(0.11)
Octane
Undecane
(0.24)
a-Pinene
No potential high ex p. 0.80
None
No potential high ex p. 0. 69
Outdoor concentration 0.26
None
O.19 4.7 O.O4
0.40 3.9 0.06
0.33 4.4 0.05
0.15 3.1 0.09
See footnotes to Table B-1 for explanation of headings.
157
-------
Table B-8. Stepwise Regression Results For 24 Homes with Outdoor
Measurements: Overnight Personal Air—Los Angeles
(May 1984)
Chemical
Variable b
S.E. F p
Aromatic Hydrocarbons
Benzene None
Styrene
IR2 = 0.18)
Ethylbenzene
o-Xylene
m,p-Xyfene
(0.15)
Circulating fan 1.4
None
None
Outdoor concentration 0. 52
0.66 4.5 0.05
0.27 3.6 0.07
Chlorinated Hydrocarbons
Chloroform No high exposure -1.23 0.62 3.9 0.06
(0.16)
1,1,1-Trichloroethane Outdoor concentration 0.71 0.25 7.7 0.01
(0.28)
Carbon tetrachloride None
Trichloroethylene None
Tetrachloroethylene Outdoor concentration 0.51 0.30 2.9 0.10
(0.34)
m,p-DichlorobenzeneNone
Aliphatic Hydrocarbons
Decane
Dodecane
Octane
10.13)
Undecane
a-Pinene
(0.45)
None
None
Outdoor concentration
None
0.33
Circulating fan 1 . 76
Gardening 1.54
Outdoor concentration- 0.61
0.19
0.57
0.58
0.26
3.0
9.6
7.0
5.5
0.10
0.006
0.02
0.03
See footnotes to Table B-1 for explanation of headings.
158
-------
Table B-9. Stepwise Regression Results For 10 Homes with Outdoor
Measurements: Overnight Personal Air—Contra Costa
(June 1984)
Chemical Variable b S.E. F p
Aromatic Hydrocarbons
Benzene None
Styrene
(R2 = 0.39)
Ethylbenzene
o-Xylene
m,p-Xylene
Smoker in home
None
None
None
1.58 0.69 5.2 0.05
Chlorinated Hydrocarbons
Chloroform None
1,1,1-Trich/oroethane None
Carbon tetrachloride None
Trichloroethy/ene Employed 1.53 0.81 3.6 0.10
(0.31)
Tetrach/oroethylene Outdoor concentration 0.39 0.19 4.1 0.08
(0.31)
m, p -Dichlorobenzene None
Aliphatic Hydrocarbons
Decane
Dodecane
(0.32)
Octane
Undecane
(0.56)
a-Pinene
None
Smoker in home
None
Smoker in home
Employed
None
1.15
1.49
1.19
0.60
0.59
0.63
3.7
6.3
3.5
0.09
0.04
0.10
See footnotes to Table B-1 for explanation of headings.
159
-------
Appendix C
Analysis of Measurement Errors
What are the sources of the measurement errors in the TEAM Study?
Two categories of errors can be distinguished: those that affect all chemicals
in one sample equally and those that affect specific chemicals. The first
category includes errors in measuring flow rates and errors in injecting
external standards—in both cases, the error affects every chemical equally.
The second category includes background contamination and chemical
reactions—in both cases, specific chemicals are affected differently from
other chemicals on the same sample.
Errors Affecting All Chemicals Equally
1. Flow rate during sample collection Flow rate is measured at the
beginning and end of the 12-hour sampling period The two
measurements are averaged and multiplied by the sampling time to
estimate the volume sampled. Two types of error are involved with
this procedure: errors in individual flow rate measurements, and the
error involved in estimating the average flow rate by a simple average
of the two measurements. The latter error may be larger than the former.
(For example, battery-operated pumps have been observed to maintain
constant flow for a number of hours and then enter a steep decline
in flow rate; in such cases, a simple average, which presumes a linear
decline, is likely to underestimate the actual volume sampled.) Both
types of errors will cause identical relative errors in estimating the
concentrations of all chemicals. Changes in beginning and ending flow
rates were usually <10%. Therefore, the error in taking the simple
average is likely to be less than 5%. Reproducibility of flow rate
measurements was usually <5%. Thus the combined error associated
with flow rate measurements is not likely to exceed \/52 + 52 ~ 7%.
2. Injection of external standard. For each sample cartridge, an external
standard (perfluorobenzene or perfluorotoluene) is injected. The
response (area counts) of the GC-MS system is then applied to all
target compounds based on the amount of standard injected. An error
in estimating this amount will affect all target chemicals in the sample
in the same way. The magnitude of the error associated with this
operation is unknown.
3. Flow rate of permeation system. An error in measuring the flow rate
of the carrier gas used to load chemicals on the cartridge will cause
errors affecting all chemicals equally. (This is not to be confused with
the individual permeation tube rates, which are chemical-specific and
are described below.) Errors are not expected to exceed 10%.
Errors Affecting Individual Chemicals
4. Contamination of Tenax during preparation, transport, and storage.
Blank values for most chemicals were consistently below 10 ng (the
160
-------
equivalent of 0.5 /ug/m3) The chemical with the highest blank levels
(the equivalent of 1-5 /jg/m3) was benzene With relative standard
deviations of up to 3 /ug/m3 equivalent, errors in estimating low
concentrations (<5 fjg/m3) could easily exceed 100%. Other chemicals
with relatively high background levels were chloroform and 1,1,1-
trichloroethane. For most other chemicals, the error due to blank
contamination should have been negligible
5. Losses for gains) during transportation and storage (recovery
efficiencies). These are determined by loading known amounts of each
target chemical on laboratory and field control cartridges, the difference
between the laboratory and field values is the loss (or gam) of the
chemical during transportation. The difference between the amount
loaded and the amount recovered on the laboratory cartridge is the
amount lost (or gamed) during storage. The amount of a chemical found
on a field cartridge is corrected for the average percent lost (or gained)
during transportation and storage, as determined from the control
cartridges; thus the magnitude of the error is dependent on the
variability of the observed loss or gain, and is different for each chemical.
These coefficients of variation ranged between 8 and 37% during the
three California visits (Table C-1).
6 Calculation of relative response factor Calculation of concentrations
depends on a "relative response factor" (RRFa) determined for each
chemical from a known amount (generally about 200 ng) loaded onto
a test cartridge At least seven determinations of the response (peak
height or area) of the known chemical compared to the response to
an external standard are made and the average ratio is used in all
calculations. One error associated with the RRFa is its variability, as
determined from the standard deviation of the observed values. Another
source of error connected with the RRFa is the assumption that the
response is linear; if the response is not linear with respect to
concentration, then an error will occur. The coefficients of variation
of the mean RRFa calculated for eleven prevalent chemicals varied from
4% to 29% during the three California visits (Table C-1). Although this
error is listed under the "chemical-specific" category, it is possible
that day-to-day variations in instrumental response affect many
chemicals similarly If so, this error would appear under the first
category.
7 Breakthrough volume. Breakthrough volume of a particular chemical
is the volume sampled (from an atmosphere at constant concentration)
at which 50% of the chemical is lost through the rear of the cartridge.
Breakthrough volumes vary according to chemical, temperature, and
the geometry of the sampling system. For the TEAM sampling geometry
and nominal volume (—20 L), the only prevalent target chemicals with
breakthrough volumes at room temperature less than the 20 L sampling
volume are chloroform (15 L) and 1,1,1-trichloroethane (19 L). If the
sample volume exceeds the breakthrough volume for a given chemical,
the concent ration is deter mined by dividing by the breakthrough volume.
Breakthrough volume is determined from previous experiments, and
is a steep function of temperature. Thus errors in the published
breakthrough volumes or in estimating temperature throughout the
sampling period will lead to chemical-specific errors.
Another potentially significant source of error associated with dividing
by the breakthrough volume is the assumption that the concentration
161
-------
•1
s
»2
^
cS
1
Uj
x
«
o
u
^
c
1
u
•2
cu
*2
o
1
1
ctj
"3>
it
o
.0
A
•a
c
>?
^
(^
0
£
s
1
6
ii
v>
,cu
o
C
• *
CM QO
co CD
^ 00
CM CM
CM co
Benzene
Carbon tetrachloride
00
*-
00
0
CM
CO
r—
r-
CO
01
00
0)
CM
Trichloroethylene
o
CM
«*
oo
0)
U)
S —
0)
^
Q
CM
i —
*•»
Tetrachloroethylene
oo
rx
rr\
* J
V-
CM
CO
CO
CO
CO
CM
if)
CM
Styrene
00 CQ IN IX
CM *~
O O Oo DO
CO Co Co ^x
co o 05 o)
*-^
co CM IN CQ
CO O 0) 05
^•t* ^J" ^i ^J"
•«- ^ CM •*
CM *-«-•-
CO 00 0) CQ
m,p-Dichlorobenzene
Ethylbenzene
o-Xylene
m,p-Xy/ene
. °°
rvt **• ^t"
GO QQ
* — CO
§"
o 7 cu
a> \ c
1 '§"^
•R^ x
a Los Angeles — first ti
bLos Angeles— secon
c Contra Costa Count
^
cu
4—
E
8
dFinnegan mass spec
eModel number.
162
-------
is constant over the sampling period. (This assumption is necessary
to justify dividing by the breakthrough volume rather than the sample
volume.) However, depending on the time-varying profile of the
concentration, dividing by the breakthrough volume can yield over- or
underestimates. For example, if the concentration is relatively high
during the first part of the sampling period, most of the chemical will
breakthrough, and the average concentration will be underestimated.
If the concentration reaches high values late in the sampling period,
little breakthrough will occur, and the average concentration will be
overestimated. Since extreme short-term peaks occur for both of these
chemicals (chloroform in showers, 1,1,1 -trichloroethane in various
spray can propellant uses) this error is potentially large in these
situations.
8. Permeation tube rates. Most of the target chemicals are loaded on
control cartridges by permeation tubes. Variations in the permeation
rates will lead to chemical-specific errors in estimating recovery
efficiencies. However, historical records of permeation-tube variabilities
indicate that they are quite stable, with variations ranging from 1-
5%.
9. Chemical reactions. Artifact formation has been observed on Tenax
(Pellizzari, 1984)1 as have effects of NO2, Os, and humidity (Pellizzari,
1 984)2 The main artifacts observed were benzaldehyde, acetophenone,
and phenol, none of which were selected as TEAM target compounds
forthat reason. However, particularly in California, IN02and ozone levels
were high, and may have led to errors of unknown magnitude due
to chemical reactions occurring during sampling.
10. Calculation of area of GC/MS peaks. These areas can be affected by
variable background heights, asymmetrical shapes, saturation, and
interferences. In most cases, the error should be of small magnitude
Table C-2 summarizes these ten types of errors, the chemicals affected,
and (when possible) the approximate magnitude of the errors.
To determine the propagation of these errors, we examine the equation
for the concentration.
Ca = (Ma - Mb)/VRa (1)
where Ma = total mass of analyte (ng/cartridge)
Mb = average mass on field blanks (ng/cartridge)
V = volume sampled or breakthrough volume, whichever is
smaller (L)
Ra = average recovery efficiency for the given analyte
In turn, Ma is determined from a mean relative response factor (RRFa)
calculated for each analyte for a given mass spectrometer from a minimum
of seven cartridges that have been loaded with known amounts of the target
chemicals and of the external standards1
1Pellizzan, E D, and Krost, K J (1984) Chemical transformations during ambient air sampling
for organic vapors Analytical Chemistry 56 1813-1819
2Pellizzan, E D , Demian, B , and Krost. K (1984) Sampling of organic compounds in the presence
of reactive inorganic gases with Tenax GC Analytical Chemistry 56 793-798
163
-------
Table C-2. Estimated Magnitude of Errors Associated with
Air Measurements
Errors Affecting All Chemicals Equally
Percent Error
1. Flow rate measurement
2. Injection of external standard
3. Flow rate of carrier gas for permeation system
5. Recovery efficiencies
6. Relative response factor
7. Breakthrough volume
All
All
Chloroform
1,1,1-Trichloroethane
8. Permeation tube rates All
9. Chemical reactions Unknown
10. Measurement of peak area All
<7
Errors Affecting
Individual Chemicals
4. Blank contamination
Chemicals
Affected
Benzene
1, 1, 1-Trichloroethane
Chloroform
m,p-Xylene
Range
of Error
up to 5 \ng/m3
up to 2 \ng/m3
up to 1 \n.g/m3
up to 1 pg/m3
8-37%
4-29%
Could be large
for the two
chemicals
affected
1-5%
7
1 n Akl/Mk,
RRFa = - I
n i=1 AsiMs,
(2)
where Ak,s = system response (integrated peak area) to the known
chemical (k) or the standard (s)
Mk,s = mass of known chemical (k) or standard (s)
i = ith of n RRF cartridges
If we assume that the RRFa determined for the known mass IVk of a
particular chemical also applies to any unknown concentration Ma, (that is,
that the response is linear with respect to concentration) we may write:
RRFa =
Aa/Ma
As/Ms
(3)
where Aa is the system response to the unknown mass of analyte Ma and
As is the system response on the day of analysis to the mass Ms of the
external standard. Solving for Ma, we have-
Aa/RRF
AS/MS
(4)
164
-------
Similarly, the average mass Mb for the blank cartridges is calculated from
several determinations at different times (indexed by t) of a set of m blank
cartridges
_ 1 m
Mb = - I (5)
m t=1 AstMst
Thus, the concentration of an analyte a on a particular field sample is'
Aa Ms m Abt Mst
RRFa As t=1 RRFa Ast
: _ (6,
or
Ms Ab Ms
A -
"a
A A
C_ s r\s
a —
RRFaVRa
where the term Aa MS/AS refers to the single determination of the analyte
area Aa and the system response As to the external standard Ms on the
sample; the term Ab MS/AS refers to an average of system responses to
external standards loaded on m blank cartridges; RRFa is an average relative
response factor determined from at least 7 RRF cartridges; Ra is the average
recovery efficiency from several control cartridges; and Vis the sample volume
or breakthrough volume of the analyte, whichever is smaller
For those chemicals with negligible blank levels, the equation reduces
to
_ AaMs
Ca " AsVRaRRFa (8)
Of the six factors on the right-hand side, errors in the measurements
of the external standard parameters (Ms, As) and the sample volume V will
affect all chemicals equally, while errors in measuring the peak area of the
target chemical and its recovery efficiency are specific to the chemical. (For
those cases with sample volumes exceeding the breakthrough volumes of
certain chemicals—primarily chloroform and 1,1,1 -trichloroethane—the error
m the breakthrough volume is of course also specific to the chemical.)
Depending on the behavior of the particular mass spectrometer, the daily
variation of the relative response factor may or may not affect all chemicals
similarly. Since all six quantities are related by multiplication or division,
the total error associated with determining the concentration is simply the
square root of the sums of the squares of the individual errors, assuming
all errors are additive and normally distributed
The coefficients of variances (CVs) of the recovery efficiencies and of the
relative response factors are compared for all eleven prevalent chemicals
(Table C-3). For every chemical, the average CVs associated with the recovery
efficiencies (13-23%) were slightly larger than the average CVs associated
with the relative response factors (9-1 6%). The combined CVs for these two
major sources of error are compared with the observed CVs of all personal
165
-------
Table C-3. Coefficients of Variation (%) of Measurement Errors:
TEAM-California Study
Observed
Recovery Precision of
Chemical RRFa Efficiency*1 Combined0 Duplicates'1
Chloroform
1, 1, 1-Trichloroethane
Benzene
Carbon tetrachloride
Trichloroethylene
Tetrachloroethylene
Styrene
m,p-Dich/orobenzene
Ethylbenzene
o-Xylene
m,p-Xy/ene
9
12
12
9
16
13
10
10
10
10
10
13
15
18
18
17
17
21
23
18
19
17
16
19
22
20
23
21
23
25
21
21
20
27
18
26
11
32
16
30
24
14
13
15
aMean relative response factor CV averaged for two mass spectrometers
during three California visits.
bRecovery efficiency CVs averaged for 48 air and breath cartridges,
cCombined error: the square root of the sums of the squares of columns
(1) and (21
dMean CVs of personal air and breath samples only !N= 74).
air and breath samples also in Table C-3 Except for chloroform,
tnchloroethylene, and styrene, these two sources of error alone appear to
account for most of the observed variation.
The errors associated with the area and volume terms are expected to
be small (<5%). The error associated with injection of the external standard
(Ms) has not been quantitated. However, we may use our knowledge of the
different ways in which these errors affect a sample to determine whether
the chemical-specific errors or the "constant-multiple" errors are dominant.
If most duplicate pairs show consistent ratios among most or all chemicals
on a sample, then the dominant errors are of the constant-multiple type
(such as Ms and V); if most duplicate pairs show no such consistent bias,
then the chemical-specific errors (such as Ra) are dominant.
The relative importance of the two categories of errors was determined
by the following scheme: a "typical" ratio R (for example, the median ratio
of the eleven prevalent chemicals on one sample to their counterparts on
the duplicate) was determined for each pair of duplicates. Then all chemicals
on one sample were multiplied by R to remove that part of the variance
due to this single multiplicative constant R. The remaining variance represents
the chemical-specific variance. If the chemical-specific variance is small
compared to the original variance, the importance of background
contamination or possible chemical reactions is minimal.
The results of carrying out this calculation for duplicate air and breath
samples in New Jersey and two of the three California trips are displayed
in Tables C-4 to C-8. As can be seen for the New Jersey data, chemical-
specific errors are small (median CVs < 10%) for about six chemicals (styrene,
166
-------
0 O)
c *••
f!
I3'
I
o
p
0
(0
*
to
§
V,
o
§
>>
°
5
V,
5
~*
^
V.
O
§
^
"^
^
^S
s^
f .
*^
^
(j
5
.0
-^
°
nj
^
Chemical
IV
»-
O
CM
0
*
1
CM
IV
O
ci
to
PO
ci
CD
0)
O
0)
1
O
00
CM
o
PO
O
sj.
00
Chloroform
CM
O
0
O
oo
CN
CN
ci
0)
CM
0
Q
^.
CN
ci
to
xj-
ci
o
CM
CM
O
O
^
o
o
PO
CM
0
to
f\i
VN
Q
Q
tb
C
to
^
Hi
1,1,1-Trichloroi
10 to
CM CM
O O
to O
Ci Ci
0) Oo
rv oj
Ci Ci
Oj Q
•"*• 00
0 0
*-~ Oj
CD
*•*
ci
^
^1" 1
Ci
PO CM
CN
iv oj
CM O
0 Ci
CD CM
o o
O tv
10
O) to
CM *--
Ci O
03 0)
*" *O
0 Ci
*-• 10
10 CM
o
't
,0
^
u
Benzene
Carbon Terra
O
o
CM
PO
O
00
0
ci
10
0
IV
CD
0
Ci
00
Ci
CD
to
o
o
0)
o
^
*-*
to
o
o
0
o
o
Q)
c
0}
Trichloroethyi
^
0
o
CD
0
*-
CN
ci
0
o
0)
^
*-.
ci
CO
Ci
^J-
CM
S
O
IV
C)
co
^f
00
c.
o
03
o
IV
10
Qj
C
-S)
^
-c
Tetrachloroet
^
CN
o
IV
PO
o
00
o
ci
,q.
*-
o
^~
00
*-
ci
CM
CN
Ci
CM
'•-
o
o
o
to
PO
0)
O
o
CM
O
O
10
CD
c
1
•kj
oo
CO
CN
O
DO
CM
O
PO
rx
O
SJ.
*~
o
10
CM
CM
Ci
to
*-
ci
PO
*~-
Ci
o
00
o
IV.
PO
v^
*s^
o
00
o
0)
.
•*. ""> Q)
"m C ^
^ B)"^3
|'i s
-Q 'o 'Q
•£ c c
•|-g'§
.to ," '*- '^
o ^t s^
a Number of t
b Median coe,
c Median coe,
-------
•SJ
9.
1
(/}
a
u
1
Q
•S
•Q
2
5
(B
J>
^
I;
^
O
4>4
i
o
1
o
G
'to
1^
*
o
"3
•§
I
£
§
s
^
i
«
o
c5
'ft
:>
^ S
.O O)
^ v
O Q)
P E
•a s
is1'3
!S
10
•4
1
^
O
5
^* *S.
Q ^J
.^
^ ^
o
1 ^
Q ^
S
-C" ">v
•S> 0
§
>
X,
o
5
*^ X
S C
CQ
5
>
0
•S
^ *S.
Q °
•i= ^
^
"cO
§
5 s
tf ^
§
* -e
Cjj i>
$ O
nj
<
Chemical
05
O
CM
0
CM
05
CO
O
rx
rx
O
^fr
^
"--
d
^j.
xj-
d
to
o
CO
d
rx
CM
d
CM
CO
*-
O
00
CO
o
CO
Chloroform
^
*-
0
to
co
O
to
CO
CM
0
05
CO
o
05
,_
10
d
0
10
d
CO
^j.
»-
d
^
CM
d
05
*-.
o
CM
O
!O
CO
o
^O
CM
cu
1
a!
1, 1, 1-Trichloro
»-,
CM
O
^.
10
O
rx
en
i —
0
CM
O
rx
rx
CM
d
CM
10
d
10
*^_
CO
d
CO
d
i —
to
CO
0
(^
^J-
0
00
1 —
Benzene
rx
*-
O
10
0
O
M-
CO
*•*.
C5
CM
CO
0
rx
\
\
O
*-
C5
CO
0
Y-^.
CM
0
^j.
hloride
o
g
ha;
§
•8
<2
O to
M' CM
d d
> •*
rx <5j-
0 0
to •*-
CO >0
CM O
o o
rx io
^ CM
C5 O
to O
0) »-
»- CO
O O
0 co
d d
^ O
* —
0) CM
»-. *^
d d
rx rx
co CM
d d
to CM
•^ CM
to IN
CM O
O O
•^ Cf)
to CM
O O
00 Co
i-- CM
c
CD CD
c "5^
Q) -C
Trichloroethy
Tetrachloroet
CM
CM
O
to
CM
0
•st-
05
O
O
^
CM
O
rx
^
O
O
xj-
CM
d
to
to
O
d
^.
CM
d
05
^j-
o
o
10
o
0
0>
i
X
c/5
05
0
Q
CO
o
00
^
CM
0
CM
0
CM
0
CM
O
10
CM
d
O
CM
d
to
T-»
d
IO
CM
CM
CM
O
05
<--
O
rx
CN
n\
enzent
-Q
m,f>-Dichloro
rx o
0 0
0 0
05 O
10 to
0 O
rx co
rx cj)
O O
O O
05 CM
10 io
O O
co ^h
O IN
0 0
05 Q
CO tO
d d
05 00
•<*• "*
0 0
d d
00 to
CM CM
d d
0 C5
CO Co
CM to
O O
O C)
\}- r\i
CM
Q.
to
"co
?
0)
-c
^tj
"
^
i.
0
0)
»^
^ Q
.5 CD
-S> "g
-Q ^
S^ Co
c §
CD il
0- . ^
A ^ -g
f> ~O -^
J "co ^
^ -5 ^
1 °l
•Q 'o 'o
-C t; C:
1 -S •§
CO CQ
•^ CO CO
co <; ^
Q) O O
111
§••§ •§
"^ i^^ t+^
a Number of *
b Median coe
c Median coe
168
-------
Table C-6. Comparison of Total Variance with "Chemical-Specific"
Component of All Measurable Duplicate Samples:
NJ, Winter 1983
Persona/ Air
Chemical
Chloroform
1, 1, 1-Trichloroethane
Benzene
Carbon Tetrachloride
Trichloroethylene
Tetrach/omethy/ene
Styrene
m,p-Dichlorobenzene
Ethylbenzene
o-Xylene
m,p-Xy/ene
Na
8
8
8
2.
7
8
9
9
9
9
9
Night
cvb
0.17
0.18
0.16
—
0.09
0.09
0.14
0.09
0.21
0.17
0.12
MCVC
0
.06
0.20
0
0
0
0.
0.
0,
0.
0
.13
—
.10
.06
.08
.10
.10
.13
.03
N
8
9
9
—
6
9
9
9
9
9
9
Day
CV
0.44
0.55
0.34
—
0.34
0.25
0.18
0.18
0.21
0.17
0.24
MCV
0.
0.
0.
0.
0.
0.
0.
16
43
07
15
08
05
13
0.03
0.
0.
04
05
N
2
2
5
—
1
5
4
5
5
5
5
Breath
CV
—
-
0.35
—
-
0.28
0.19
0.22
0.22
0.18
0.22
MCV
—
~
0.13
—
-
0. 10
0.05
0.05
0.01
0.05
0.04
8Number of duplicate pairs with both values '/2 quantifiable limit.
bMedian coefficient of variation of original data.
cMedian coefficient of variation of modified data (constant factor
removed): "chemical-specific" CV.
Note: Outdoor air duplicate samples too few to calculate statistics.
ethylbenzene, o-xylene, m,p-xylene, tnchloroethylene, and tetrachloroeth-
ylene) For these chemicals, it appears likely that multiplicative errors, such
as errors in flow rate measurement, injection of standards, or variation of
the relative response factor, were the major sources of error Background
contamination appeared to be an important source of error for benzene,
chloroform, and 1,1,1-trichloroethane
In California, the first Los Angeles visit had very clean Tenax backgrounds
and only chloroform and styrene had "chemical-specific" coefficients of
variation consistently exceeding 10% (Table C-7) In Contra Costa, 1,1,1-
trichloroethane, tnchloroethylene, and tetrachloroethylene sometimes
exceeded "chemical-specific" coefficients of variation of 20% (Table C-8),
however, the number of measurable duplicates was very small
These results indicate that a major portion of the error affects most
chemicals similarly. Since the combined errors due to recovery efficiency
and relative response factors are large enough to account for most of the
observed error, we conclude that the error due to the relative response factor
must be of the first category (affecting all chemicals equally) rather than
the second
In summary, of the ten sources of error discussed, two have measured
ranges of variability considerably larger than most of the rest: recovery
efficiencies and relative response factors. These two alone appear capable
of causing a significant portion of the observed variation in precision of the
duplicate samples. Several other sources of error—blank contamination,
breakthrough volume, chemical reaction—could be large on occasion,
169
-------
Table C-7. Comparison of Total Variance with "Chemical-Specific"
Component of All Measurable Duplicate Samples:
Los Angeles, Winter 1984
Persona/ Air
Night
Chemical
Chloroform
1, 1, 1-Trichloroethane
Benzene
Carbon tetrachloride
Trichloroethylene
Tetrachloroethylene
Styrene
m , p -Dichlorob enzene
Ethylbenzene
o-Xylene
m,p-Xylene
Na
9
11
1
1
11
1
1
1
1
1
9
1
1
1
1
1
11
CV
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
19
05
17
14
12
10
16
11
11
13
0.09
Adj
CV
0.22
0.08
0.07
0.04
0.09
0.02
0.11
0.06
0.05
0.04
0.04
N
12
12
12
11
9
11
11
11
12
12
11
Day
CV
0.26
0.18
0.15
0.20
0.12
0.13
0.36
0.45
0.18
0.20
0.22
Adj
CV
0.10
0.06
0.05
0.06
0.10
0.05
0.14
0.15
0.04
0.03
0.06
N
3
12
12
2
7
11
3
8
8
11
12
Breath
CV
0.25
0.11
0.20
0.07
0.10
0.15
0.24
0.11
0.26
0.14
0.21
Adj
CV
0.11
0.04
0.08
0.11
0.06
0.10
0.17
0.12
0.14
0.09
0.09
aNumber of duplicate pairs with both values greater than the quantifiable
limit.
bMedian coefficient of variation.
cMedian coefficient of variation after adjusting for multiplicative errors.
however, they are unlikely to affect a large portion of the samples because
the observed precision of the duplicates (11 -32%) appears to allow little room
for additional unknown errors.
170
-------
Table C-8. Comparison of Total Variance with "Chemical-Specific"
Component of All Measurable Duplicate Samples:
Contra Costa, June 1984
Personal Air
Chemical
Chloroform
1, 1, 1-Trichloroethane
Benzene
Carbon tetrachloride
Trichloroethylene
Tetrach/oroethyfene
Styrene
m ,p-Dichlorobenzene
Ethylbenzene
o-Xylene
m,p-Xylene
Na
2
6
7
6
7
6
4
4
6
7
7
Night
CV
0.31
0.25
0.45
0.20
0.54
0.15
0.33
0.22
0.17
0.12
0.07
Breath
Day
Adj
CV
0.33
0.20
0.15
0.09
0.46
0.02
0.13
0.05
0.03
0.01
0.02
N
2
7
7
6
4
7
4
3
7
7
7
CV
0.
0.
0.
11
13
47
0.21
0.34
0.
0.
0.
0.
0.
0.
13
45
19
13
07
11
Adj
CV
0.09
0.08
0.13
0.04
0.22
0.27
0.23
0.14
0.12
0.09
0.05
N
0
4
5
1
0
5
5
3
5
4
6
CV
0.
0.
0.
0.
0.
0.
0.
0.
43
18
18
17
25
06
14
15
Adi
CV
—
0.23
0.06
—
-
0.64
0.11
0.01
0.05
0.17
0.06
aNumber of duplicate pairs with both values greater than the quantifiable
limit.
b Median coefficient of variation.
cMedian coefficient of variation after adjusting for multiplicative errors.
171
-------
Appendix D
Corrections to the Estimated Frequency Distributions
Due to Measurement Error
Random measurement errors cause increases m the observed variance
of any distribution. These increases lead to overestimates of the number
of people exposed to concentrations greater than any concentration above
the median. For a normal distribution of exposures and a normal distribution
of (additive) measurement errors, the variance of the observed distribution
equals the sum of the variances of the true distribution and of the
measurement errors:
222
ffobs - CTtrue + (Term
Similarly, for a log-normal distribution of exposures and a log-normal
distribution of (multiplicative) errors, the same formula holds for the
logarithms of the quantities. Multiplicative (concentration-dependent) errors
are commonly encountered in environmental measurements, particularly
those having a dependence on flow rates (Evans, 1 984)1. Since the observed
concentrations (breath, personal air, and outdoor air) are reasonable
approximations to log-normal distributions, at least between the 10th and
the 90th percentiles (Figures 4 to 14), an attempt has been made to calculate
the correction factor associated with the 90th percentile for all air and breath
measurements and all prevalent chemicals during all three seasons in New
Jersey using the observed quality control data on duplicate measurements.
(The duplicate measurements also show evidence of being drawn from
distributions whose central regions can be approximated by a log-normal
fit: Figures D-1 and D-2 )
A multiplicative measurement error is defined as the ratio of one member
(for example, XT) of a duplicate pair of observations (x-i, x2) to the geometric
mean of the pair
E =
The logarithms of these errors were then calculated for all duplicates collected
during the three seasons. For certain chemicals, these errors have been
plotted on log-normal probability graph paper (Figures D-3 to D-4). The results
indicate that the measurement errors as defined above are in fact reasonable
approximations to log-normal distributions, at least between the 10th and
90th percentiles. Thus, by calculating the variance of the measurement errors
(crE2 = 2s2, where s2 is the variance of the duplicates), the true variance
can be estimated:
_ 2 _ 2 2
-------
Figure D-1. Cumulative frequency distribution of geometric means of 62 pairs
of duplicate measurements of overnight personal air exposures to
1,1.1-trichloroethane (New Jersey. Fall 198J).
7000
500
300
T Measurement 1
f Geometric Mean
J- Measurement 2
1 2 510 20 30 50 70 90 95 98 9999 5 99.9
Cumulative Frequency, percent
We can then define the correction factor at one standard deviation above
the mean (the 84th percentile) as the ratio of the true value /j exp (OT) to
the observed value /j exp (ffobs):
Correction factor =
exp (
exp (a0bs)
= exp (crT
Figure D-5 illustrates the effect of a correction factor of 0.93 on a distribution
with an observed geometric standard deviation of exp (
-------
Figure D-2. Cumulative frequency distribution of geometric means of 62 pairs
of duplicate measurements of overnight personal air exposures to
benzene (New Jersey, Fall 1981).
400
300
200
too
50
^ 30
g
cS
25%
error) is benzene. For all other chemicals, personal air measurements are
consistently good. Outdoor air errors are occasionally large for chloroform,
styrene, and carbon tetrachloride. Breath values are poor for 1,1,1-
trichloroethane and four of the five aromatics. In 6 of 55 cases, measurement
errors were large enough to account for all of the observed variance.
174
-------
Figure D-3. Cumulative frequency distribution of measurement errors (defined
as the ratio of one measurement to the geometric mean of the
pair) for 62 pairs of duplicate overnight personal air samples: 1,1,1-
trichloroethane (New Jersey. Fall 1981).
995 99 98 95 90 80 70 605040 30 20 10 5 2 J 0.5
1.0
0.5
0.4
0.3
n—r
1,1,1 - Trichloroethane
i—\—r
0.5 1 2 5 10 20 304050607080 90 95 989999.5
Cumulative Frequency, percent
Figure D-4. Cumulative frequency distribution of measurement errors for 62
pairs of duplicate overnight personal air samples: benzene (New
Jersey, Fall 1981).
99.5 9998 95 90 8O 7O6O5040 30 2O 10 5 21 \ 0.5
1.0
(18)
Benzene
0.5 1 2 5 10 20304050607080 90 95 989999.5
Cumulative Frequency, percent
175
-------
Figure D-5. Effect of correction factor of 0.93 (Table D-1} on observed
cumulative frequency distribution of overnight personal air
exposures to tetrachloroethylene for 350 residents of Elizabeth-
Bayonne, New Jersey.
Population Exceeding Concentration Shown (103)
96 64 32 128 64
c-P*1
1
c
o
I
o
0
^
c
o
«
<
Q.
•C
0)
c
1
o
70
60
50
40
30
20
10
7
5
4
3
2
-
Tetrachloroethylene
10 20 40 60 80 90 95 99 995
Cumulative Frequency, percent
999
Table D-1.
Correction Factors Due to Measurement Errors—Fall 1981
Chemical
Chloroform
1, 1, 1-Trichloroethane
Benzene
Carbon tetrachloride
Trichloroethylene
Tetrachloroethylene
Styrene
m,p-Dich/orobenzene
Ethylbenzene
o-Xylene
m,p-Xy/ene
Breath3
0.70
0.60
—
0.97
0.84
0.85
0.90
0.96
-
0.55
0.50
Personal
Night3
0.96
0.93
0.75
0.92
0.96
0.93
0.89
0.96
0.89
0.74
0.81
Air
Day3
0.92
0.81
0.62
0.63
0.84
0.96
0.68
0.92
0.92
0.92
0.84
Outdoor
Nigh?
C
0.82
—
0.95
0.98
0.92
-
0.98
0.98
0.95
0.93
Air
Day*
0.87
0.91
0.66
—
0.86
0.97
0.77
0.97
0.92
0.92
0.75
a Corrected 90th percentile value/observed 90th percentile.
b Corrected 75th percentile value/observed 75th percentile.
c Corrected value cannot be calculated—measurement errors too large.
176
-------
In the summer of 1982, contamination of the Tenax cartridges during
storage occurred. This caused a general worsening of the measurement errors
for most chemicals (Table D-2) However, the majority of observations were
not invalidated—only 9 cases out of 50 had to be discarded.
Finally, the corrections due to measurement error are quite small in the
winter 1983 season (Table D-3). The Tenax batch was quite clean (with
the exception of benzene) and no problems of contamination were
encountered in the field. Thus, the overestimates at the 90th percentile are
usually 5-15% for all chemicals except 1,1,1-trichloroethane, which is
overestimated by a factor of 2 in the personal air determination. All 27 cases
gave useful information on the variance of exposures.
Since these calculations rest on criteria that we know to be violated (the
distributions of observations and of errors seldom meet the Kolmogorov-
Smirnov criterion for log-normality; the errors probably include additive as
well as multiplicative components), a numerical simulation was run to
determine whether the originally observed distribution could be recovered
by convoluting the observed errors with the corrected distribution. At the
time of publication, only one set of simulations has been run on one
compound—the results indicated that the original distribution could in fact
be recovered to within 5% of the observed values over most of the distribution
by this method.
Thus, although it is not possible to state that the method has been validated,
preliminary indications are encouraging.
Table D-2. Correction Factors Due to Measurement Errors—
Summer 1982
Personal Air
Outdoor Air
Chemical
Breath3
Night3 Day3 Nigh? Day''
Chloroform
1,1,1-Trichloroethane
Carbon Tetrachloride
Trichloroethylene
Tetrach/oroethy/ene
Styrene
m,p-Dich/orobenzene
Ethylbenzene
o-Xylene
m,p-Xylene
0.64
C
0.83
0.54
0.56
0.44
0.87
-
0.46
—
0.70
0.72
-
0.58
0.69
0.76
0.92
0.64
-
0.73
0.73
0.86
0.65
0.76
0.89
0.69
0.80
0.53
0.74
0.64
0.95
0.92
0.94
0.69
0.96
0.87
-
0.80
-
0.88
0.98
0.66
0.93
0.94
0.74
0.83
0.92
—
-
0.75
a Corrected 90th percentile value/observed 90th percentile.
bCorrected 75th percentile value/observed 75th percentile.
0 Corrected value cannot be calculated—measurement errors too large.
177
-------
Table D-3. Correction Factors3 for Estimated Frequency Distributions
Based on Measurement Errors— Winter 1983
Chemical
Chloroform
1, 1, 1-Trichloroethane
Trichloroethylene
Tetrachloroethylene
Styrene
m , p -Dichlorobenzene
Ethylbenzene
o-Xylene
m,p-Xy/ene
Breath
0.90
0.87
0.89
0.85
0.97
0.97
0.94
0.96
0.95
Personal
Overnight
0.93
0.42
0.93
0.96
0.91
0.99
0.94
0.91
0.88
Air
Daytime
0.85
0.69
0.70
0.94
0.83
0.96
0.85
0.90
0.85
' Corrected 90th percentile value/observed 90th percentile.
178
-------
Appendix E
A Method for Comparing Weighted and Unweighted
Distributions on Probability Graph Paper, with
Examples from the TEAM Study
The TEAM Study employed a three-stage stratified design One purpose
was to identify persons with potentially high exposures and to overrepresent
them in the sample population to improve the precision of the estimates
of these relatively rare high exposures Thus, stratification variables such
as occupation, residence near major point sources, and socio-economic status
were employed. By selecting persons in these high potential exposure strata
with higher probabilities than the rest of the sample, a relatively larger number
of high exposures should result Since the probabilities of selection are known
for each person, it is straightforward to "weight" the observed values by
the reciprocal of the probability of selection to arrive at an estimate of the
actual distribution of concentrations for the entire target population The
observed distribution is called the unweighted distribution and the corrected
distnbution is called the weighted distribution
It is useful to compare the two distributions in order to determine the
effect of the weighting process For example, did it work as predicted? If
so, the higher exposures should have been concentrated among the people
selected with higher probabilities (smaller weights). One way to compare
the two distributions would be to examine the weights associated with the
highest exposures. If most of the small weights are associated with high
exposures, the persons expected to have high exposures did If a high
proportion of observations with large weights are associated with high
exposures, an unsuspected source of high exposures may be operating
These exposures and their associated weights may be sorted and
frequencies calculated or they may be compared graphically A graphical
comparison has the advantage of displaying the entire distributions at a
glance However, before comparing the two distributions, a method must
be developed capable of comparing two very different population sizes on
the same set of axes
In the first season of the TEAM Study, personal exposures were measured
for 350 volunteers representing 1 28,800 residents of Elizabeth and Bayonne,
New Jersey To compare the distribution of exposures of the same population
with the target population on the same set of axes, a percentile plotting
convention must be adopted For unweighted samples of size N, two popular
plotting conventions for the cumulative probability Pk associated with the
kth ordered point (1 < k < N) are
Pk = k/(N +1) (1)
Pk = (k - 1/2)/N (2)
As shown by Chernoff and Lieberman (1954)2 the second convention
(Pk = (k-1/2>/N) leads to better estimates of the standard deviation of a normal
2Chernoff, H and Lieberman, G J (1954), Use of normal probability paper, J Amer Stat Assoc
49778-785
179
-------
distribution (for N>10) than the first convention. Therefore, we shall employ
this convention in plotting the unweighted frequency distributions of
exposures for the 350-person sample population.
This convention may be visualized as follows: the N observations split
the distribution into N percentile ranges. A reasonable choice for the proper
plotting percentile of the kth ordered observation is the midpoint of the
associated range. (k-!/2)/N.
For the weighted distributions, we must develop a similar plotting
convention For a set of N weighted observations representing a population
of P persons, we know that the sum of the weights is the population P-
N
I W, = P (3)
Generalizing our observation on the N equal percentile ranges, we now
have N unequal percentile ranges, of "width" W,. A natural choice for the
plotting percentile of the kth observation is the midpoint of the associated
width Wk.
k
Pk = I W, - Wk/2 (4)
P
For the maximum observation (k = N), this reduces to:
PN = (P - WN/2)
(5)
This convention has the desired properties, namely,
1 The highest observation will be plotted at the same percentile as on
the unweighted curve if the weight WN equals the "average" weight
P/N; at a higher percentile if the weight is less than average; and
vice versa.
2. The weighted curve will lie below (to the right of) the unweighted curve
if the weights have been properly chosen (i.e., higher weights for low
levels of exposures).
Conversely, if the weighted curve lies above (to the left of) the unweighted
curve, a preponderance of persons expected to have low exposures in fact
had high exposures, a sign that certain characteristics associated with high
exposures may have been overlooked in the sample stratification process.
Applying these considerations to the overnight personal exposures (i.e.,
indoor air concentrations) in the New Jersey Fall 1981 study, we find that
three of five chemicals had weighted curves on the "wrong" side at the
higher percentiles (Figures E-1 to E-5). This may be due to the fact that
the importance of indoor air sources was not well understood when the
study was designed, and therefore potential indoor air sources were not
used to stratify the sample.
Of course, one of the stratification variables, occupation, would not be
expected to affect night-time exposures. A similar comparison of daytime
exposures would be required to determine whether the selected occupations
indeed had most of the high exposures.
180
-------
Figure E-1.
WOO
500
200
700
Weighted vs. unweighted frequency distributions for m,p-
dichlorobenzene. The straight line is a log-normal curve with the
same geometric mean and geometric standard deviation as the
observed distribution.
,g
-c
50
20
Si 70
Q
Q.
E~
T~
"T
\—\
Weighted
Unweighted
1 2 5 10 20 40 60 80 90 95 9899995
Cumulative Frequency, percent
The "trimmed" geometric standard deviations of the weighted and
unweighted curves were approximated by calculating the square root of the
ratio of the 84th to the 16th percentile (Table E-1). All but one of the chemicals
had "trimmed" geometric standard deviations in this central region very near
3; but the dichlorobenzene isomers were distributed in a much more strongly
right-skewed fashion For this chemical, the log-normal approximation was
not as good as for the others. For the other chemicals, however, the log-
normal approximation using these "trimmed" geometric standard deviations
was generally within 5% of the observed values between the 10th and 90th
percentiles. Beyond tho 90th or 95th percentile, exposures to all chemicals
were higher than predicted by the log-normal approximation.
181
-------
Figure E-2.
WOO
Weighted vs. unweighted frequency distributions for 1.1,1-
trichloroethane. The straight line is a log-normal curve with the
same geometric mean and geometric standard deviation as the
observed distribution.
500 -
200-
I
c
o
WO _
c
to
•g
g
o
I
2-
1 2
5 W 20 40 60 80 90 95 9899995
Cumulative Frequency, percent
182
-------
Figure E-3. Weighted vs. unweighted frequency distributions for benzene. The
straight line is a log-normal curve with the same geometric mean
and geometric standard deviation as the observed distribution.
WOO
500
200
> Weighted
Unweighted
100
5
I
c
o
c
01
o
c
o
o
oa
50
20
10
1 2 5 10 20 40 60 80 90 95 98 99 995
Cumulative Frequency, percent
183
-------
Figure E-4. Weighted vs. unweighted frequency distributions for styrene. The
straight line is a log-normal curve with the same geometric mean
and geometric standard deviation as the observed distribution.
WO
50 -
20-
5 10
I 5
Z
I
a
g 2
0.5
0.2
0.1
\—i—I—[—\
Weighted
Unweighted
_i_
i
L_1_J _ I _ L
II III
1 2 5 10 20 40 60 80 90 95 98 99 99.5
Cumulative Frequency, percent
184
-------
Figure E-5.
WOO
500
200
5
I 700
c
o
§ 50
o
c
o
O
Q)
-------
Table E-1. Weighted and Unweighted Overnight Personal Exposures
(Indoor Air Concentrations) and Geometric Standard
Deviations Calculated for Selected Percentiles
Percentile
Chemical
1, 1, 1-Trichloroethane
Tetrachloroethylene
Benzene
m,p-Dichlorobenzene
W
Ua
W
U
W
U
W
U
16
5.9b
6.2
2.2
2.2
5.0
4.7
0.9
0.8
50
16.9b
16.9
6.3
6.3
15.0
15.0
3.8
3.7
84
52b
50
20
20
46
45
39
35
Ratios of
Percentiles
50
16
2.9
2.7
2.9
2.9
3.0
3.2
4.5
4.7
84
50
3.1
3.0
3.2
3.2
3.1
3.0
10.0
9.5
1 84
V 16
3.0
2.8
3.0
3.0
3.0
3.1
6.8
6.7
aW = weighted; U = unweighted.
186
-------
Appendix F
Personal Vs. Outdoor Air Comparisons by Season-
New Jersey
Daily 24-hour personal exposures and 24-hour outdoor air concentrations
of selected chemicals are compared for all three seasons in New Jersey
in Figures F-1 through F-5 (see also Figure 26 in text). Because of quality
assurance problems, benzene values are available only for the Fall 1981
season (Figure F-5). Personal air exposures exceed outdoor air concentrations
at all percentiles for all chemicals in all seasons, with the single exception
of chloroform in summer (Figure F-4) Personal exposures to the four
chemicals with several seasons of valid data appeared to decrease in summer
compared to either fall or winter. However, outdoor concentrations of two
chemicals—chloroform and 1,1,1 -trichloroethane—were highest in summer
Thus, indoor-outdoor differences were generally smallest in summer and
largest in winter
187
-------
Figure F-1. 24-hour personal exposures to 1.1,1-trichloroethane compared to
outdoor air in New Jersey-first three seasons.
1000
500
6
I
o
200
WO
50
5 20
s
o
•s
10
\ \
' • Personal Exposures
' • Outdoor Concentrations
I / I
_ Summer / ^ .
' /
Fall
I
Winter
\ \ \ \
I
\ I
50 60 70 80 90 95 98 99 99.5
Cumulative Frequency, percent
188
-------
Figure F-2. 24-hour personal exposures to tetrachloroethylene compared to
outdoor air in New Jersey-first three seasons.
WOO
500
s
1
c
o
1
c
200
WO
50
20
10
I I I T I
• Personal Exposures
• Outdoor Concentrations
] m
Summer //
"winter '!
f* Fall
Summer
I I I I
I
50 60 70 80 90 95 98 99 99.5
Cumulative Frequency, percent
189
-------
Figure F-3. 24-hour personal exposures to styrene compared to outdoor air
in New Jersey-first three seasons.
100
I I I I
_ • Personal Exposures
• Outdoor Concentrations
SO
20
w
c
-------
Figure F-4. 24-hour personal exposures to chloroform compared to outdoor
air in New Jersey-first three seasons.
200
WO
50
I
c
o
c
01
u
c
o
o
20
JO
r \ \ \
• Personal Exposures
• Outdoor Concentrations
\\
Fall
Winter
Summer
50 60 70 80 90 95 98 99 99 5
Cumulative Frequency, percent
191
-------
Figure F-S.
24-hour personal exposures to benzene compared to outdoor air
in New Jersey-fall season.
200
100
50
20
§
c
o
o
------- |