United States
Environmental Protection
Agency
Office of Pesticides and
Toxic Substances
Washington, DC 20460
EPA-560/13-80-028
December 1980
Toxic Substances
Measurement of Benzene
Body-Burden for Populations
Potentially Exposed to Benzene
in the Environment

-------
       MEASUREMENT OF BENZENE BODY-BURDEN FOR POPULATIONS
        POTENTIALLY EXPOSED TO BENZENE IN THE ENVIRONMENT
                               by

Ruth A. Zweidinger, Stephen D. Cooper, Benjamin S.  H. Harris, III,
    Tyler D. Hartwell, Ralph E. Folsom, Jr., Edo D. Pellizzari
      Albert W. Sherdon, Thomas K. Wong and Harvey S. Zelon
                     Contract No. 68-01-3849
                             Task 1
                 Project Officer:   Joseph Breen

                      Field Studies Branch
                  Exposure Evaluation Division
            Office of Pesticides and Toxic Substances
                      Washington,  DC  20A60
                           August 1980

-------
                                 DISCLAIMER
     This report has been reviewed by the Office of Pesticides and Toxic
Substances, U. S. Environmental Protection Agency, and approved for publica-
tion.  Approval does not signify that the contents necessarily reflect the
views and policies of the U.  S. Environmental Protection Agency,  nor does
mention of trade names or commercial products constitute endorsement or
recommendation for use.
                                      ii

-------
                                  ABSTRACT
     A pilot study was performed to assess the measurement of benzene body-
burden for populations potentially exposed environmentally to benzene.
Probability sampling was used to select the participants in the two geographi-
cal sites, (1) Harris County, TX, and (2) St. Louis County, MO, parts of
Wood River, Roxana, South Roxana, and Hartford, IL.

     Benzene levels were measured for the air and water environmental exposure
for each participant and the benzene body-burden was measured through breath
levels and, in a subsample, blood levels.

     A pretest of occupationally exposed and nonexposed individuals was used
to test analytical methodology and the concept of breath as an indicator of
body-burden.  The blood benzene levels expected and observed required analyti-
cal methods capable of measuring 1 |Jg/L or below.  This methodology did not
exist and had to be developed for the pretest and pilot study.

     The range of air benzene levels found in the Harris County study (49
participants) was 2 to 45 JJg/m3 with a weighted mean of 16.1 pg/m3; breath
levels ranged from 0 to 14 JJg/m3 with a weighted mean of 2.9 [Jg/m3.  In the
St. Louis study (68 participants) the range of air benzene levels was 3 to
125 |Jg/m3 with a weighted mean of 26.8 |Jg/m3; breath levels ranged from 1 to
26 fJg/m3 with a weighted mean of 8.5 (Jg/m3.

     This report was submitted in fulfillment of Contract No. 68-01-3849
by the Research Triangle Institute under the sponsorship of the U. S. Environ-
mental Protection Agency.  This report covers the period September 17, 1977
through September 18, 1980 and the work was completed as of December 1980.
                                      111

-------
iv

-------
                                   CONTENTS

Abstract	   iii
Figures	    vi
Tables	viii
Acknowledgments	    xi

     1.   Introduction	     1
     2.   Results and Conclusions 	     5
     3.   Recommendations 	 .....     7
     4.   Exposure of Populations to Benzene	     9
     5.   Method Development and Pretest Evaluation of Filling Station
               Attendants (and Tank Truck Drivers)	    21
     6.   Sample Design and Selection 	    30
     7.   Field Operations	    60
     8.   Sample Collection and Chemical Analysis 	    64
     9.   Data Analysis	    83
    10.   Discussion of the Health Effects of Benzene 	   112

References	   126
Appendices

     A.   Sampling and Analytical Protocols for Benzene in Air, Water,
               Blood, Breath and Urine	   130
     B.   Spirometer for the Determination of Benzene in Breath 	   146
     C.   Sequential Without Replacement Sampling for Attributes and
               Subpopulation Means	   152
     D.   Data Collection Instruments for the Household Survey and
               Material Submitted to OMB	   166
     E.   Data Listing of Air and Breath Benzene Level Distributions
               by Site and by Exposure Strata	   187

-------
                                   FIGURES

Number                                                                     Page

 1   Map of Harris County, TX, showing the various exposure areas.  ...    33

 2   High exposure strata for Wood River, Roxana , South Roxana and
          Hartford, IL .........  . ...............    46

 3   Strata for the medium exposure (1, 2, and 3) and low exposure
          strata  (I1 , 2'  , and 3' ) ..... .  .......  .  ......    47

 4   Location of  segments for the low exposure sample in Harris
          County, TX  .....  .......  ..............    69

 5   Location of  segments for the medium  exposure sample in
          Houston, TX ......  ............... .....    70

 6   Location of  segments for the high exposure  sample in Houston,
          TX  .........  .....  .........  .......    71

 7   Location of  segments for the high exposure  sample in Wood
          River/Roxana/Hartford,  IL (St.  Louis study) ..........    77

 8   Location of  segment for medium exposure  sample in St. Louis, MO        78

 9   Location of  segments in the  low exposure area in St. Louis, MO.        79

 10   Frequency bar  charts for benzene.  ...  ........  . .....    gg

 11   Percentile plots of benzene  levels in air and breath; by site and
          exposure  strata ......................        g^
 12    Site  =  Houston.   Plot  of breath vs . air

 13    Site  =  St.  Louis.   Plot of breath vs. air
 14   Site  =  St.  Louis.   Plot  of breath vs. air  ....  ......         no

 15   Pathways  of benzene metabolism  and elimination.  .  .....           -inr

 A-l   Thermal desorption  inlet-manifold  ..............         -^2

 A- 2   Analytical  system for  analysis  of organic  vapors in ambient air  .  .   133
                                      VI

-------
                                FIGURES CONT'D.

Number                                                                    Pag(

A-3  Six-port, two-position valve for the introduction of headspace
          samples	142

B-l  Schematic diagram of the spirometer 	   148

B-2  Specifications for 50-L Tedlar bag with duct	149

E-l  Site = Houston.  Plot of air exposed = *	191

E-2  Site = Houston.  Plot of breath exposed = *	192

E-3  Site = St. Louis.  Plot of air exposed = *	193

E-4  Site = St. Louis.  Plot of breath exposed = *	194
                                      VII

-------
                                    TABLES

Number                                                                    Page

 1   Organization and Responsibilities of the Centers and Divisions
          Participating in the Study. ........ 	     3

 2   Petroleum Refineries Producing Aromatics, by State  	    10

 3   Estimated Size and Production Capacity of By-Product Coke Plants
          In the United States on December 31, 1975	    11

 4   Plants Using Benzene as an Intermediary in the Manufacture of
          Other Chemical Compounds	    12

 5   Estimation of Atmospheric Emissions Due to By-Product Manufacturing
          Facilities in 1976. . 	 .....    15

 6   Ambient Monitoring Data for Benzene. ...............    17

 7   Concentrations of Benzene Found in Ambient Air by RTI.......    18

 8   Estimated Exposure of Populations to Benzene 	 ...    20

 9   Benzene in Breath—Validation of Collection.	    24

 10   Results of Pretest Study of Benzene Body-Burden. .........    25

 11   Linear Regression Analysis of the Benzene Air Exposure, Breath and
          Blood Benzene Levels in the Pretest	    26

 12   Sample Statistics for Blood and Breath for Two Intervals of
          Air Exposure	         29

 13   Estimated Levels of Benzene in Human Breath from "Old Love Canal"
          in Niagara Falls, NY	          29

 14   High Exposure Area, Houston, Texas—Minor Strata .....             09

 15   Stratum Housing Unit Data and Sample Allocation. ......          or

 16   Selected Samples—Houston, TX	          07

 17   Housing Counts Per Segment for Houston, TX	      38
                                     Vlll

-------
                                TABLES CONT'D.

Number                                                                   Page

 18  Sample Sizes of Eligible Persons for Houston, TX	    39

 19  Sampling Weights for Houston, TX	    42

 20  Stratum Housing Unit Data and Sample Allocation St.  Louis,  MO .  .    49

 21  Selected Samples for St. Louis, MO	    51

 22  Housing Counts Per Segment for St. Louis, MO	    52

 23  Selected Samples and Supplementary Samples for St.  Louis, MO. .  .    53

 24  Sample Sizes of Eligible Persons	    57

 25  Sample Weights for St. Louis, MO	    59

 26  Benzene Levels Found in the Houston Study 4/3/79-5/5/79 	    66

 27  Quality Control Data for Houston Study	    72

 28  Benzene Levels Found in the St. Louis Study 9/18/79-10/20/79. .  .    73

 29  Confirmation by Gas Chromatography/Mass Spectrometry	    80

 30  Quality Control Data for St. Louis Study	    82

 31  Unweighted Summary Statistics for Benzene Levels in Air, Breath,
          Water, and Blood Samples:  By Site	    84

 32  Summary of Relative Benzene Levels	    86

 33  Weighted Means and Standard Errors for Air and Breath Benzene
          Levels:  By Site	    87

 34  Unweighted Sample Medians and Ranges for Air vs. Breath Levels
          by Site and Sampling Strata	    90

 35  Power of Detecting Various Differences Between the High and Low
          Sampling Strata	    92

 36  Correlations Between Air and Breath Levels vs. Exposure Strata:
          By Site	    93

 37  Pearson and Spearman Correlations Between Air and Breath Benzene
          Levels:  By Site	    95
                                      IX

-------
                                TABLES CONT'D.

Number                                                                    Page

 38  Summary Statistics for Benzene Levels in Breath for Two Categories
          of Air Levels:  By Site.	   99

 39  Sample Medians for Air and Breath Levels Upwind and Downwind of
          Benzene Sources in Houston  ... 	  100

 40  Sample Medians for Air and Breath Levels Upwind and Downwind of
          Benzene Sources in Wood River  (The High Exposure  Strata of
          St. Louis)	101

 41  Frequencies for  Questionnaire Variables:  By Site	103

 42  Results of Stepwise Regressions  by  Site with Air and Breath Benzene
          Levels As the Dependent Variables  (Also Log [Air] and Log
           [Breath] Benzene Levels)  .	  108

 43  Answers to Questionnaire for Selected Individuals with High Air  or
          Breath Benzene Levels	109

 44  General Effects  of Benzene Intoxication  .....  	  ...  117

 45  Bone Marrow Changes Due to Benzene  Intoxication	  119

 46  Effects of Age and Sex on Benzene Toxicity	120

 47  Possible Leukemogenic and Tumorigenic Effects of Benzene	123

 48  Possible Mutagenic and Teratogenic  Effects of Benzene  	  124

 A-l  Operating Parameters for Thermal Desorption GC/FID Analysis of
           Benzene	

 A-2  Operating Parameters for GC/FID  Analysis of Benzene in Blood and
           Urine	

 E-l  Air and Breath Benzene Levels  and Sample Weights for Houston and
           St. Louis	                    TOO
                                                 	•	loo

-------
                               ACKNOWLEDGMENTS
     The authors wish to thank the following RTI personnel for their partici-
pation in the program:  Jesse McDaniel, Doris Smith,  Dick Waddell,  Dorothy
Grossman, and Mary Van Sciver.  Drs.  Joseph Breen and Vincent DeCarlo of the
Environmental Protection Agency provided assistance and guidance for the
implementation of this work.

     Special appreciation of the assistance of local  health officials in
support of the sampling effort is acknowledged.   Specifically we would like
to thank Dr. Richard K. Donelson of the Texas Department of Health  and Dr.
Robert A. McLean, City of Houston Health Department,  for supporting the
sampling effort in Harris County.  Robert L. Wheatley and Genelle Moore of
the Illinois Department of Public Health provided the support of the Wood
River/Roxana/ Hartford, Illinois, sampling.  Rich L.  Roberts of the Missouri
Department of Natural Resources, Charles M. Copley and W. L. Hagar  of the
City of St. Louis, and Clifford Mitchell of the  St. Louis County Health
Department supported the sampling effort in St.  Louis County.
                                      XI

-------
                                  SECTION 1

                                INTRODUCTION
     Benzene is not only a fundamental and well-known organic  chemical,  it
is also a chemical of major industrial importance.  In the United  States,
benzene ranks 13th in volume (1) with a production for 1976 of 1.5 x  109 gal
(2).   Most (88%) of the domestic benzene production is from petroleum
sources with the remainder from coal (3).  The largest benzene source is the
catalytic reforming process at oil refineries.  Other major production
routes are dealkylation of toluene and coproduction, with ethylene, from
steam crackers  (2).

     The primary uses of benzene are chemical manufacturing, solvent  opera-
tions, and as an additive in gasoline.  Chemical processing is  the major use.
Although benzene is used in the commercial production of literally hundreds
of compounds (4),  its major uses are as the starting material  for  styrene
(45%), cumene/phenol  (20%), and cyclohexane (17%) (2).  These  compounds,  in
turn, are used  in  production of polystyrene plastics and rubbers and  other
fabricated plastic products.  Despite its widespread industrial use,  more of
the population  is  exposed -to benzene from the use of automobiles than from
all other uses  combined.  This includes benzene exposure from  gasoline
service  stations and  from general automobile exhaust (5).

     Although benzene is widespread in the environment, the levels of environ-
mental exposure are significantly less than the levels of industrial  exposure.
Risk assessment is hampered by the lack of reliable dose response toxicitv
data,  especially in humans.  Therefore, the health consequences at ambient
benzene  concentrations  are speculative.

     The health effects of benzene have been reviewed extensively recentlv
 (6), especially with  respect to its potential carcinogenic effects.   Even-
though benzene  toxicity has been recognized for over 50 years, much of its
action is  still poorly  understood.  It is clear that the most'serious  effect
of chronic  exposure is  depression of the hematopoietic system,  ranging from
mild reversible depression of some of the formed elements to aplastic  anemia
and leukemia.   The latter has been particularly difficult to study since no
animal model has been found in which benzene induces leukemia despite eDidemio-
logical  evidence linking the two in humans.  Other toxic effects of
are central nervous system depression, and histochemical changes in
liver, small intestine, spinal cord, and heart (7).

-------
     In view of these serious consequences from chronic benzene exposure, an
evaluation of the exposure/body-burden of benzene in the general human
population in areas of relatively high (industrial and urban), medium (urban),
and low (rural) exposure was undertaken.   Relatively little was known at the
project's initiation about benzene levels in ambient air and especially
about measurement techniques for assessing body-burden at very low benzene
levels.  Blood and breath levels had been measured (8) after 25-ppm exposure
for 2 hr.  The extrapolation of these data to exposures of 1 to 40 ppb
indicated that new and innovative sampling and chemical analysis techniques
would have to be developed to accomplish the objectives of the study, i.e.,
to measure benzene body-burden at ambient benzene exposures.

     To test the methodology that was developed, a pretest was performed
using nine individuals both occupationally exposed to benzene and nonoccupa-
tionally exposed.   This pretest included smokers as well as nonsmokers.   A
semiportable spirometer (Figure B->1) was  developed that permitted the collec-
tion of breath samples, and analytical methods capable of detecting benzene
in blood at 1 ng/mL were developed.   This technology was evaluated on the
pretest subjects.

     Using the information derived from the pretest,  and air monitoring  data
that became available, two sites were selected for study:  (1)  Harris County,
TX (Houston), and (2) St.  Louis County, MO, with parts of the municipalities
of Wood River, Roxana and Hartford,  IL, included (St. Louis).

     A team of RTI specialists was brought together for the execution of
these studies.  A diagram of the responsibilities of each center or division
is given in Table 1.

     The Sampling Research and Design Center (SRDC) devised the survey
design used.  The sampling frame was constructed and the areas  designated
for counting and listing of housing units.  The sampling frame  information
and maps were transmitted to Survey Operations Center (SOC) staff who executed
the counting and listing of housing units.  SRDC then selected  housing units
to be screened for eligibility and willingness to participate.   A random
sample was selected by SRDC from the eligible individuals identified during
screening.  SOC staff then arranged for the actual sample collection (breath
and air) by Analytical Sciences Division personnel.  In addition,  SOC
personnel administered the study questionnaire and collected tapwater.  If
blood collection was scheduled for the participant, SOC arranged for a
phlebotomist to collect the sample.

     Analytical Sciences Division (ASD) staff, in addition to collecting
the air and breath samples, maintained custody of all other samples and
provided for their transport to the laboratory.  All of the benzene deter-
minations in each medium were performed by ASD and the data were supplied
to the Statistical Methodology and Analysis Center (SMAC) for analysis.
The sampling weights, study questionnaire, and chemical determinations were
also analyzed by SMAC.

     One of the significant contributions of this program was the integration
of sampling design, field operations, chemical sampling and analysis, and

-------
                 Table  1.   ORGANIZATION AND  RESPONSIBILITIES OF  THE  CENTERS AND DIVISIONS
                                                  PARTICIPATING IN  THE STUDY
Sampling Research and
   Design Center
     (SRDC)
   Survey Operations
        Center
        (SOC)
    Analytical  Sciences
         Division
          (ASD)
Statistical Methodology and
       Analysis Center
           (SMAC)
•  Designed survey and
  sampling frame

•  Drew the sample

•  Assigned sampling
  weights
•  Coordinated  all field
  activities

•  Counted and  listed housing
  units in areas designed  in
  the sampling frame

  Conducted the household
  screen for eligible and
  willing participants

  Scheduled appointments
  for air monitoring and
  breath and blood collec-
  tion in coordination with
  ASD staff

  Administered the study
  questionnaire to partici-
 pants

 Collected  tapwater samples
 and arranged for a phle-
 botomist  to collect blood
 samples
• Coordinated with SRDC and
 SOC all activities

 Developed analytical and
 chemical sampling techniques
 for the study

 Tested methods in pretest

 Collected air and breath
 samples

 Analyzed air, breath, blood,
 and water samples for
 benzene
•  Analyzed the chemical
  data from ASD

•  Analyzed the question-
  naire data from SOC

•  Applied the sampling
  weights from SRDC

-------
statistical data analysis with experts in each of these fields performing
their respective roles.

-------
                                  SECTION 2

                           RESULTS AND CONCLUSIONS
     Sample collection and chemical analysis methods were  developed  and/or
improved in order to execute this study.  The pretest was  performed  to
validate the chemical sampling and analysis methods developed.  Nine indivi-
duals participated, including smokers and nonsmokers and occupationally
exposed and nonexposed individuals.  As a result of this pretest,  a  spirometer
for collection of breath samples and analytical methods capable of detecting
benzene in blood at 1 ng/mL were developed.

     The execution of this study required clearance by the Office  of Manage-
ment and Budget (OMB) under the Federal Reports Act because  a questionnaire
was to be administered to the human subjects.  Approval was  granted  with  the
understanding that:

     1.   "These surveys are being conducted as a pre-test of the  feasibility
          of the information collection procedures;

     2.   That the information collected will not be used  to generalize to
          either local areas or the nation as a whole."

 In deference to these stipulations, only conclusions about the methods used
 will be  made and the studies conducted in Houston, and St. Louis,  will be
 referred to  as pilot studies.

     The survey had a two-stage design with stratification imposed at the
 first  stage.  First-stage  sampling units were clusters of  housing  units
 called segments, from which  a sample of eligible persons was selected for
 the second  stage of sampling.  The degree of stratification  of the exposure
 areas  had strong implications on the level of effort required to collect  the
 air and body-burden samples.  In the Harris County, TX, pilot studv   15
 first-stage  strata were  used.  The distances between sampling points  were
 great  and sample collection  was slow.  In the St. Louis, MO-Wood River/Roxana/
 Hartford, IL, pilot study, 9 first-stage strata were used  and the  level of
 effort to collect  samples  from a greater number of participants was  reduced
 by about one-third.

      Of the  exposure  and body-burden samples evaluated, only air and  breath
 showed detectable  benzene  levels in the majority of samples  for environmen-
 tally  exposed  individuals.   In the other two matrices, water and blood
 benzene was  at  or  near  the detection limit for all samples.  Hence,  water
 and blood samples  do  not appear to be  good indicators of either exposure  or

-------
body-burden at environmental levels.  To compound the problem of low level
exposure/body-burden, benzene is ubiquitous in ambient air.  Its presence
in all air samples limits the lower end of the dynamic range.  The differen
tiation must then be made between population body-burdens differing by
factors of 2 or less instead of by orders of magnitude as might be expected
with some pollutants.

     The range of air benzene levels in the Harris County study was 2 to 45
pg/m3 with a weighted mean of 16.1 [ig/m3; breath levels ranged from 0 to 14
Hg/m3 with a weighted mean of 2.9 |Jg/m3.   In the St.  Louis study the range
of air benzene levels was 3 to 125 Mg/m3 with a weighted mean of 26.8
[Jg/rn3; breath levels ranged from 1 to 26 [Jg/m3 with a weighted mean of 8.5

-------
                                  SECTION 3

                               RECOMMENDATIONS
     Further study in the area of exposure and body-burden  should  include  a
wider activity range of the participants.  Personal monitors  should be
attached to the participant to accurately assess their exposure during
outdoor activities.  Ambient air benzene levels have been recorded in the
St. Louis, MO-Wood River/Roxana/Hartford, IL, study area that were a factor
of 2 above the highest air level found in this study (9).

     Retesting of participants would provide better information concerning
individual variation and response to variations in exposure.  The  additional
burden on the participant would require careful consideration, however.

     Several recommendations concerning the field procedures have  developed
from discussions at RTI and with the field staff.  The first  concerns the
timing of the study.  The break between screening and final sample enrollment
should be made as  short as possible, since people's health  and activity
patterns  change and a person screened as eligible may no longer be so at the
time of actual sampling.  The shorter the time interval, the  less  likely the
change.

     The  second area of concern is the size and location of the areas being
sampled and  the size of the field staff.  It is vital once  the sample frame
is developed and the participants screened and selected that the activities
of the  field staff be examined in terms of time and distance between areas
and that  the size, composition, and schedule of the field study staff be
determined  accordingly.  Careful scheduling of activities can alleviate
problems  to  some extent.  However, the outcome of the study will depend on
 the ability of  the field team to cover all areas and all persons within the
 time scheduled.

      Analytical methods  capable of detecting benzene in blood at 0.05 ug/L
 or less need to be developed to evaluate blood as an indicator of  benze
body-burden. Examples  of such methods are photoionization  detection optimized
 for benzene and use  of  on-line headspace purge and cryogenic focusine   '
 to gas  chromatography.

      Any  study  of  this  general scope should not be undertaken without a f  11
 complement  of experts  in sampling design, survey operations, chemical
 sampling  and analysis,  and  design and statistical analysis,  inexperi
 any of these areas may  lead to errors that could invalidate or bias th     ^

-------
entire study.  Each of these areas is essential to the design, execution,
and interpretation of studies relating to environmental exposure and body-
burden.

-------
                                  SECTION 4

                     EXPOSURE OF POPULATIONS TO BENZENE


BENZENE MANUFACTURING SITES AND INDUSTRIAL USERS

     The benzene production capacities are listed by  state  in  Table  2  for
petroleum-based benzene and in Table 3 for coal-derived benzene  (3,10).
These facilities are highly concentrated in the Texas  Gulf  Coast  region,
with several in the Northeast, and others scattered nationwide.

     The most extensive industrial use of benzene is  as a starting material
for the production of other chemicals.  Table 4 lists  the benzene user
facilities, their location, and their processes (5).   Again, the  geographical
distribution is the same—high concentration of benzene utilization  in the
Gulf Coast area, less in the Northeast, and a scattering of facilities
nationwide.

EMISSIONS AND AMBIENT MONITORING DATA

     Because of its volatility, benzene is undoubtedly released to the
environment at many steps  of production, storage, transportation, and  use.
Petroleum  refining accounts for approximately 27 x 106 kg/yr of total
benzene  emissions.  Approximately 4 x 106 kg/yr of benzene  is  emitted  by
solvent  operations.  Estimates of benzene emissions from several  of  the
major  by-product manufacturing operations are given in Table 5 (5).  An
upper  limit was estimated  using a material balance and all  losses were
assumed  to be benzene,  although certain processes are  multistep and  the
material lost may not be benzene.  The lower limit was obtained from esti-
mates  of the  losses based  on the processes with the applicable control
technology.

      Emissions  estimates are useful for predicting where ambient  air levels
may be high;  however, they are no substitute for air monitoring.  For
 example, the  Houston, TX,  area would appear to be a potentially'benzene-
 contaminated  area because  of the industrial activity.  A study by the  Te
Air Control  Board  (11)  in  Houston, TX, of ambient air  benzene  levels showed
 levels that  ranged  from 2.2 to 30 ppb.  To differentiate between  benzene
 derived from combustion processes  (auto exhaust) and  that from industrial
 sources, acetylene,  another combustion product, was monitored  simultaneou 1
 The results  were  ratioed and the ratio and absolute values  were correlat d ^'
 with wind direction.  The  results were used to identify components of  the
 ambient benzene levels  due to  traffic patterns and industrial  activity on
 the Houston  ship  channel.

-------
   Table  2.   PETROLEUM REFINERIES PRODUCING AROMATICS, BY  STATE
State
California
Illinois
Kansas
Kentucky
Louisiana
Mississippi
New York
Oklahoma
Pennsylvania
Texas
Total
Number of plants
3
2
1
1
3
1
1
1
3
18
34
«•
Quantity (bbl/ stream day)
5,990
6,700
1,400
4,000
19,100
6,000
3,000
2,000
9,700
122,525
180,415
JL,

 Total quantity of  benzene,  toluene, and  xylene produced.
    • 1977, Oil & Gas Journal, used by permission of Petro-
     leum Publishing Company
                                   10

-------
        Table 3.  ESTIMATED SIZE AND PRODUCTION CAPACITY OF BY-PRODUCT COKE PLANTS IN THE
                               UNITED STATES ON DECEMBER 31,  1975
State
Alabama
California
Colorado
Illinois
Indiana
Kentucky
Maryland
Michigan
Minnesota
Missouri
New York
Ohio
Pennsylvania
Tennessee
Texas
Utah
West Virginia
Wisconsin
Undistributed
Total
Number of
o
plants
7
1
1
4
6(7)
1
1
3
2
1
3
12
12(13)
1
2
1
3(4)
1
-
62(65)
Number of
batteries
28
7
4
9
31
2
12
10
5
3
10
35
51
2
3
4
13
2
-
231
Number of
ovens
1,401
315
206
424
2,108
146
758
561
200
93
648
1,795
3,391
44
140
252
742
100
-
13,324
Maximum annual
theoretical
production
capacity (tons)
6,961,000
1,547,000
1,261,000
2,523,000
11,925,000
1,050,000
3,857,000
3,774,000
784,000
257,000
4,053,000
9,960,000
18,836,000
216,000
839,000
1,300,000
4,878,000
245,000
-
74,266,000
Coke
production
in 1974
• (tons)
5,122,000
(b)
(b)
1,912,000
9,073,000
(b)
(b)
3,259,000
(b)
(b)
(b)
8,842,000
16,318,000
(b)
(b)
(b)
3,555,000
(b)
12,656,000
60,737,000
 Three plants are collocated.

 Included in undistributed.

Source:   Sheridan,  E. T.,  "Supply and Demand for United States Coking  Coals  and Metallurgical Coke,"
         U.S.  Department of the Interior, Washington, DC  (1976).

-------
                   Table 4.  PLANTS USING BENZENE AS AN INTERMEDIARY IN THE MANUFACTURE OF
                                          OTHER CHEMICAL COMPOUNDS
N>
Company
Reichhold Chen., Inc.
Witco Chem.
Std. Oil Co. of CA
Speciality Organics, Inc.
Std. Oil Co. of CA
Ferho Corp.
Std. Chlorine Chem Co., Inc.
Chem. Products Corp.
Clark Oil & Refining
Koppers Co., Inc.
Reichhold Chem., Inc.
Monsanto
Skelly Oil Co.
Ashland Oil, Inc.
Foster Grant Co.
Cos-Mar, Inc.
Tenneco , Inc .
Rubicon Chem. , Inc.
Georgia Pacific Corp.
Gulf Oil Corp.
Continental Oil Co.
Solvent Chem. Co., Inc.
Dow Chemical
City
Tuscaloosa
Carson
El Segundo
Irwindale
Richmond
Santa Fe Springs
Delaware City
Cartersville
Blue Island
Cicero
Morris
Sauget
El Dorado
Ashland
Baton Rouge
Corville
Chalmette
Gelsmar
Plaquemine
Welcome
Baltimore
Maiden
Midland
Statv:
Alabama
California
Delaware
Georgia
Illinois
Kansas
Kentucky
Louisiana

Maryland
Massachusetts
Michigan
Chemical (s)
phenol
detergent alkylate
cumene
dichlorobenzene
phenol, detergent alkylate
phenol
mono- and dichlorobenzene
dichlorobenzene
cumene , phenol
maleic anhydride
maleic anhydride
nitrobenzene, mono- and dichlorobenzene
cumene, phenol
cumene
ethylbenzene, styrene
ethylbenzene, styrene
ethylbenzene
nitrobenzene, aniline
phenol
ethylbenzene, styrene
detergent alkylate
mono- and dichlorobenzene
ethylbenzene, styrene, cumene, phenol, mono- and dichlorobenzent
                                                                                      (continued)

-------
Table 4 (continued)
Company
First Mississippi Corp.
Monsanto
Montrose Chem. Corp. of Cal.
American Cyanamid
Union Carbide
Relchhold Chem. , Inc.
Tenneco , Inc .
E. I. Du Pont
Std. Chlorine Chem. Co.
Texaco, Inc.
ICC Industries, Inc.
Occidental Petroleum
Solvent Chera. Co.
Allied Chem. Corp.
United States Steel
Ar co /Polymers, Inc.
Koppers Co., Inc.
United States Steel
Allied Chemical Corp.
United States Steel
Gulf Oil Corp.
Phillips Petroleum
Commonwealth Oil
Union Carbide Corp.
Exxon Corp.
E. I. Du Pont
Union Oil Co. of CA
American Petrofina
Phillips Petroleum
City
Pascagoula
St. Louis
Henderson
Bound Brook
Bound Brook
Elizabeth
Fords
Gibbstown
Kearny
Westville
Niagara Falls
Niagara Falls
Niagara Falls
Syracuse
Haverhill
Beaver Valley
Bridgeville
Clairton
Frankford
Neville Island
Philadelphia
Guayama
Penuelas
Penuelas
Bay town
Beaumont
Beaumont
Big Spring
Borger
State Chemical (s)
Mississippi nitrobenzene, aniline
Missouri maleic anhydride
Nevada monochlorobenzene
New Jersey nitrobenzene, aniline
phenol
maleic anhydride
maleic anhydride
nitrobenzene, aniline
dichlorobenzene
cumene
New York mono- and dichlorobenzene
monochlorobenzene
mono- and dichlorobenzene
mono- and dichlorobenzene
Ohio phenol
Pennsylvania styrene
maleic anhydride
phenol
phenol
maleic anhydride
cumene, chlorohexane
Puerto Rico cyclohexane
ethylbenzene, cyclohexane
cumene, phenol
Texas cyclohexane
nitrobenzene, aniline
cyclohexane
ethylbenzene, styrene, cyclohexane
cyclohexane
                                         (continued)

-------
Table 4 (continued)
Company
Monsanto
Coastal States Gas
Sun Oil Co.
Union Pacific Corp.
Dow Chemical
Arco/Polymers, Inc.
The Charter Co.
Joe Oil, Inc.
The Herichem Co.
Petro-Tex Chem. Corp.
El Paso Natural Gas
Dow Chemical
Phillips Petroleum Co.
Arco/Polymers, Inc.
Gulf Oil Corp.
Texaco
Union Carbide Corp.
Phillips Petroleum Co.
Marathon Oil Co.
Monsanto
Standard Oil (Indiana)
Union Carbide Corp.
Koppers Co . , Inc .
Allied Chem. Corp.
PPG Industries, Inc.
Mobay Chem. Corp.
American Cyanamide
Stlmson Lumber Co.
Kalama Chemical
City
Chocolate Bayou
Corpus Christ!
Corpus Christ!
Corpus Christ!
Freeport
Hous ton
Houston
Houston
Houston
Houston
Odessa
Oyster Creek
Phillips
Port Arthur
Port Arthur
Port Arthur
Seadrlft
Sweeney
Texas City
Texas City
Texas City
Charleston
Follansbee
Moundsville
Natrium
New Martinsville
Willow Island
Anacortes
Kalama
State Chemical (s)
Texas (con.) cumene, phenol, detergent
cumene

alkylate

ethylbenzene, styrene, cumene
cyclohexane
ethylbenzene, styrene
ethylbenzene, styrene
ethylbenzene
no data
phenol
maleic anhydride
ethylbenzene, styrene
phenol
detergent alkylate
ethylbenzene
cumene
cumene
ethylbenzene, styrene
detergent alkylate
cumene
ethylbenzene , styrene

















ethylbenzene, styrene, cumene
West Virginia detergent alkylate
monochlorobenzene


nitrobenzene, maleic anhydride
mono- and dichlorobenzene
nitrobenzene, aniline
nitrobenzene, aniline
Washington phenol
phenol






-------
               Table 5   ESTIMATION OF  ATMOSPHERIC EMISSIONS DUE TO BY-PRODUCT
                              MANUFACTURING FACILITIES IN 1976
Production capacity3 Amount of benzene
By-product (106 kg) unaccounted for (106 kg)
Ethylbenzene
Styrene
Cyclohexane
Cumene
Phenol
Nitrobenzene
Detergent alkylate
Monochlorobenzene
Aniline
Maleic anhydride
Dichlorobenzene
Total
3894
3211
2076
1720
1252
483
393
313
314
189
120

120
-
0
-
275
-
98
-
24
142
22
682
Benzene emission using
emissivity factor (106 kg)
2.41
4.82
7.58
0.43
1.25
3.38
0.86
1.10
7.41
18.28
1.03
48.55
 Source:  SRI, 1976 Directory of Chemical Producers, as cited in PEDCO, Environmental, "Atmos-
 pheric Benzene Emissions," prepared for U.S.  Environmental Protection Agency, RTF  (1977).

 Calculated using material balances in Benzene Environmental Sources of Contamination, Ambient Levels
 and Fate,  Life Sciences Division,  Syracuse University Research Corporation, 1974.
Calculated  using emission factors  for benzene per kg of product.

-------
     The contribution of auto exhaust to ambient benzene levels has been
estimated at 1 to 4 ppb in inner city areas and less than 1 ppb in suburban
areas (5).  These estimates are in general agreement with the ambient air
levels, although downtown Houston levels have been measured at an average of
8 ppb for the 0600 to 0900 hr period (an average of 29 days).  The pervasive-
ness of auto-exhaust-derived atmospheric benzene contributes more to higher
total benzene emissions than any other source.  Yet emissions from motor
vehicles are widely distributed and contribute to relatively low ambient
levels.  Emissions from benzene production, transport, storage, and use are
localized and the exposure of the population near these sites may be greater
than that of the general population.  In a study of the exposure of service
station and bulk loading operators (12,13), ambient air concentrations were
measured at 0.1 to 9.4 ppm.  The highest urinary phenol concentrations (a
benzene metabolite) were 18 mg/L for service station operators, 10 mg/L for
bulk loading facilities workers, and 48 mg/L for workers loading gasoline
containing added (10 to 33 percent) benzene.  Another recent study of benzene
exposure in self-service filling stations found levels of 43,  121,  and 647
ppb in the breathing zone of three customers (14).

     One major source of benzene not considered in the above discussion is
tobacco smoke.  Cigarette smoke has been reported (15) to contain 47 ppm,
(6.1 [Jg/40 mL puff) which is higher than the NIOSH-recommended maximum
allowable concentration of 10 ppm.  These levels of benzene are important to
this study since smokers and their families would be exposed to higher than
ambient concentrations of benzene.

     Benzene is ubiquitous in air samples and is nearly always found in
water samples.  Table 6 summarizes the ambient monitoring data through 1974
(16-27).  Benzene was found (17) in finished water from the Mississippi
River, but not in the effluents from 60 industries  that discharge into the
Mississippi, suggesting that the benzene source is  nonindustrial (3).
Benzene is one of the organic compounds identified in drinking water in the
United States (28) at a minimum concentration of <5.0 |Jg/L.   It has been
identified in the drinking water in Miami, FL, Ottumwa, IA,  Philadelphia,
PA, Cincinnati, OH, Washington, DC, and the White House (28,29) but was not
detected in the drinking water of Seattle, WA.

     In addition to the air sampling data in Table 6, RTI has quantitated
benzene in air samples from a variety of locations  all over the United
States (1).  In over 500 analyses to date, benzene has been detected in all
samples.  Sampling locations have included the Los  Angeles Basin, Houston,
TX, Kanawha Valley, WV, El Dorado, AR, northern NJ, and Baltimore,  MD.
Table 7 presents some of the quantitative results found by RTI.

     Battelle (30) recently measured benzene in air, water,  and soil at
several facilities employing benzene.  They found ambient air benzene
levels ranged from 0.4 to 34.8 ppb.  Water levels ranged from <1 ppb to a
high of 179 ppb, which was found in a plant effluent.  Soil levels also
ranged from less than detectable (<2 ppb) to a high of 191 ppb.  Care must
be exercised in comparing data for ambient air because seasonal variations
may be an important factor in the emission rate and air-soil partitioning of
the benzene.


                                    16

-------
                         Table 6.  AMBIENT MONITORING DATA FOR BENZENE
Reference
Gordon and Goodley
(1971)
U.S. EPA (1972)
Prlllows (1971)
Novak et al. (1971)
Williams (1965)
Sonyer et^ al. (1971)
Helllgan e± al. (1975)
Altschuller and Bellar
(1963)
Lonneman et al. (1968)
Grob and Grbb (1971)
Stephens (1973)
Filar and Graydon (1973)
Type of sample
water and mud
finished water
finished water
"polluted" and
"pure" drinking
water
ambient air
ambient air
ambient air
ambient air
ambient air
ambient air
ambient air
ambient air
Geographical
location
Lower Tennessee River
Carrollton Plant,
New Orleans
U.S. PHS Hospital
Carville, LA
Prague, Czechoslovakia
Vancouver, Canada
vicinity of solvent
reclamation plant
Los Angeles basin
Downtown Los Angeles
Los Angeles basin
Zurich, Switzerland
Riverside, CA
Toronto, Canada
Sampling
method used
CCE liquid-liquid
CCE

inert gas
stripping
cold trap -
GC column
grab sample
cold trap -
firebrick
grab sample
cold trap -
glass beads
charcoal trap -
carbon disulfide
extract
cold trap -
GC column
cold trap -
GC column
, , .a
Analysis
technique
GC-MS
preparative GC

GC, GC-MS
rapid heating
into GC
direct injection
into GC; MS, IR
rapid heating
into GC
direct injection
into GC
rapid heating
into GC
GC-MS
GC-FI
GC-FI
GC-FI
Quantities
detected
not reported
not attempted
"trace" ppb-ppm
range
"^0.1 ppb
1-10 ppb
23 ppm
0.005-0.022 ppm (V/V)
0.015-0.06 ppm (V/V)
aver. 0.015 ppm
highest 0.057 ppm
(V/V)
0.054 ppm
0.007-0.008 ppm
aver. 0.013 ppm;
highest 0.097 ppm
CCE - carbon chloroform extract;  GC = gas chromatography; FI = flame ionization;  IR =  infrared
spectroscopy; MS = mass spectrometry.

-------
 Table 7.  CONCENTRATIONS OF BENZENE FOUND IN AMBIENT AIR BY RTI
Site
Bound Brook, NJ
Paterson, NJ
Clifton, NJ
Newark, NJ
Fords, NJ
Passaic, NJ
Hoboken, NJ
Edison, NJ
Staten Island, NY
Los Angeles , CA
South Charleston, WV
Belle, WV
Dominguez , CA
Houston, TX
Pasadena, TX
Deer Park, TX
St. Louis, MO
Magnolia, AR
El Dorado, AR
Baton Rouge , LA
No. samples
1
1
1
1
1
1
1
29
1
1
3
6
1
3
2
7
-
12
34
14
Concentration
Mean
-
-
-
-
-
-
-
60
-
-
34 + 21
28 + 48
-
2.5 + 2.4
-
4.0 + 4.5
-
0.39 + 0.38C
0.06 + 0.40d 0
1.0 + 1.1
(ppb)
Range
3.3
0.8
trace
111
1.1
0.8
trace
to trace
0.8
6.8b
58 - 20
125 - 0.3
12.7
5.4 - 1.2
2.4 - 0.5
0.8 - 12.2
72 - 0.2
0.08 - 1.5
.002 - 0.16
0.02 - 3.4
 Fence line monitoring downwind of a chemical dump,  levels as high as
 480 ppb were found.
 Downwind of a petrochemical facility, 260 ppb benzene was found.
p
 Samples for 12 24-hour periods were collected on the roof of a
 three-story building.
 Samples for 32 24-hour periods were collected on a  water tower ^-20
 to 30 m above ground.

Source:  E. D. Pellizzari, Quantification of Benzene in 150 Ambient
         Air Samples, Final Report, EPA P.O. No. DA-7-43205, August
         1977.
                              18

-------
ROUTES OF EXPOSURE

     For the general population,  exposure  occurs  primarily  through  respira-
tion of the benzene generated  by  motor  vehicle  use.   The  presence of  high
concentrations of benzene  in cigarette  smoke, previously  mentioned  affects
not only smokers, but people around  them  in congested areas.   These two
routes of exposure represent a background  that  is highly  variable among
individuals and with time  according  to  their personal daily routine.   In
addition, there are indications  (17,18,28,29) that benzene  may be present in
significant concentrations in  the drinking water  in certain areas.   Some
individuals may receive  large  doses  of  benzene  while using  it as a  solvent,
through the use of gasoline for nonmotor vehicle  uses,  through the  use of
benzene-containing commercial  products, or via  other unidentified sources.

      Information  on benzene levels in food is sparce.   It apparently  occurs
naturally  in  fruit, fish,  vegetables,  nuts,  dairy products, beverages, and
eggs.  Quantitative data exist only  for cooked  meat, rum, and eggs.  A
report by  the  National  Cancer  Institute (31) estimated that an individual
could ingest  as many as  250 [Jg/day from these foods.  This  compares to 320
|jg/day respired from continuous exposure  to 1 ppb of benzene in air.   The
relative  absorption of  benzene via these  two routes of exposure is  uncertain.

BASIS OF  SITE SELECTION

      The  sites for measurement of benzene  body-burden levels for populations
 in the vicinity of benzene manufacturing plants and/or benzene industrial
 user facilities were  selected  based  on meteorological,  geographical,  topo-
 logical,  and demographic data  for the area surrounding the  facility.   Exposure
 via water and food was  considered to be reasonably constant within  a  locale.
 High, medium, and low  exposure sites were  then  selected based on the  proximity
 to an identifiable benzene source.  Individuals were monitored with personal
 samplers  for their  ambient air exposure for 6 to  8 hours  prior to collecting
 the biological sample.   Because  benzene is excreted very  rapidly during the
 first 3 to 4 hours  after exposure, this exposure  period is  the most relevant
 to the study "(6,8).

      The data in Table 8 were  used to assess the  locations  most likely to
 define the upper limit of the  human  body-burden and three locations were
 recommended for study.   The first location to be  evaluated  was Houston  TX
 Although it does  not rank highest for human exposure,  the time frame  of the
 study required a southern location for the winter/early spring sampling
 Subsequent sampling  in summer  was less restricted.  St. Louis  MO   was the
 second choice.  Several factors  were considered in site selection in  addition
 to a potential benzene-exposed population.  For example,  each of these
 locations is a metropolitan area where general  urban exposure may be  evaluated
 and, in each location, a  suburban-to-rural population exists within a reason-
 able distance to serve as a control.
                                     19

-------
                            Table  8.   ESTIMATED EXPOSURE OF POPULATIONS TO BENZENE
N>
O
Location
St. Louis, MO
Elizabeth, NJ
Houston, TX
Cicero, IL
Philadelphia, PA
Morris, IL
Texas City, TX
Freeport, TX
Neville Island, PA
Bridgeville, PA
Company
Monsanto
Shell
Reichhold
Petrotex
Joe Oil
Arco/Polymers
Koppers
Gulf Oil Corp.
Reichhold
Monsanto
Marathon Oil Co.
Standard Oil (Ind.)
Dow
U.S. Steel
Koppers
Emission rate
(106 kg/yr)
4.64
1.35
2.22
NA
0.09
0.483
0.325
2.61
1.78
0.02
1.53
1.74
1.45
Population exposed to
>10 ppb benzene
190,200
34,100
20,900
10,200
7,800
5,200
5,100
2,500
2,400
2,300
        Source:   S.  J.  Mara and S.  S.  Lee,  Human Exposures  to Atmospheric Benzene, EPA Contract 68-01-4314,
                 Final  Report,  October 1977.


        NA =  not avaiable.

-------
                                  SECTION 5

   METHOD DEVELOPMENT AND PRETEST EVALUATION OF FILLING STATION ATTENDANTS

                           (AND TANK TRUCK DRIVERS)
     The feasibility of measuring benzene body-burden required testing with
both known exposed individuals and nonexposed individuals.  A local popula-
tion of occupationally exposed individuals was available in the form of
filling station attendants and tank truck drivers.  Additional information
about the ability to assess body-burden could be obtained by retest of the
individuals after a period away from the occupational exposure.  This approach
was incorporated into the pilot study described below.

     In addition to designing the test protocol for the participants, there
were several technical problems in the sampling and analysis that had to be
solved before the pilot test could be implemented.  Benzene must be analyzed
in blood at 1 pg/L or less on a sample size of 10 mL or less.  Previously
reported methods (8,32) were sensitive down to 10 Hg/L.  The second technical
development required was a portable or semiportable breath sampling device
or spirometer.  The solutions to these technical problems are discussed as a
preface to the pilot study results.

EXPERIMENTAL PROCEDURES

     The sampling and analysis protocols developed for this study are given
in Appendix A and special equipment developed for breath sampling is described
in Appendix B.  The background, development, and validation is discussed
below.

Benzene in Blood

     After an initial attempt to use the Volatile Organic Analysis (VOA)
purge technique for the recovery of benzene from blood, a headspace analysis
was used similar to that described by Sato et al. (8,32) in which an aliquot
of blood sample was sealed in a hypodermic syringe and equilibrated at 37°C
with subsequent analysis of the headspace over the blood.  To increase the
sensitivity of this approach, a larger blood sample, 1 mL instead of 0.1 mL,
and a larger syringe, 10 mL vs. 1 or 2 mL, were used.  In addition, instead'
of taking a small aliquot of the headspace, the entire headspace was purged
through a cryogenic trap, which was part of a sample loop of a gas chromato-
graph and could subsequently be placed in line and warmed to volatilize the
benzene and inject it as a small discrete sample onto the column.   Preliminary
verification indicated that the laboratory air, which was introduced into


                                    21

-------
the headspace prior to equilibration, contributed less than 0.3 ng total
mass to the sample.  Using outdated blood from a local blood bank, both
spiked and blank samples were prepared.  The control samples contained 500
|Jg/L.  Blank samples were used as they were obtained from the blood bank.
Duplicate samples of the control and blanks were analyzed.  The benzene
found in the blank was 1.8 + 0.6 (Jg/L and in the controls, 123 + 8 |Jg/L.
Although the reproducibility on the controls was very good, improvement in
the 25 percent recovery was desired.

     The blood analysis procedure was modified by reducing the time for
equilibration (see Appendix A).  To speed the attainment of thermal equili-
brium, the 10-mL glass syringes were preequilibrated at 37°, 1 mL of the
blood was introduced, and the syringes were sealed and reequilibrated for 20
min.  After the 20-min equilibration, the entire headspace was purged into
a cryogenic trap, the contents of which can be injected onto a GC.  The
sample was then injected and the trap heated.  This procedure was evaluated
using pooled blood from two nonsmoking laboratory workers.  Benzene was
added to portions of the pooled blood to give concentration of 1, 5, 10, and
20 (Jg/L.  Each of these concentrations was analyzed in duplicate along with
the blank pooled blood.  The average percent relative deviation of the
benzene spiked samples was 4 percent over this range and a least squares
linear regression of the values gave a correlation coefficient of 0.991.
The principal difficulty with the detection and quantitation of benzene in
blood at low levels is the appearance of a benzene peak in "blank" blood.
This blank has been assessed by analyzing blood samples taken from three
nonsmoker laboratory workers (the two used from the calibration above plus
subject F).  These blood samples were analyzed in duplicate to yield a blank
value of 1.1 + 1.2 [Jg/L (standard deviation).  Based on this variability in
the blank, 3.4 [Jg/L would have to be present to detect benzene with a 95
percent confidence level.  A further modification of the procedure was made
using "zero" grade air for filling the headspace rather than laboratory air.
The result of this modification was the reduction of blank blood values from
1.1 + 1.2 to 0.49 + 0.39 M8/I"  By this method, 1.6 (jg/L of benzene would
have to be found for the results to be significant at the 95 percent confi-
dence level.

Benzene in Breath

     A breath collection apparatus (spirometer), diagrammed in Figure B-l,
was constructed for the collection of breath samples.  The spirometer was
tested using two smokers for a positive test.  The blank used in this evalua-
tion was a volume of purified air comparable to that collected during the
tests drawn through the apparatus.  Sample collections were performed in the
laboratory.  Subject A was a 100-kg male who normally smoked between 1 and
1.5 packs of cigarettes per day.  Subject A was a laboratory worker; however,
all uses of benzene were restricted and no exceptions had been made for the
particular place of work for this individual.  Subject B was a 60-kg male
who smoked 1.5 packs of cigarettes per day.  He was a shop worker fabricating
both metal and wood pieces.  There was some possibility of exposure to
benzene through solvents or adhesives that he may have handled during the
course of his activities.  These subjects were chosen primarily to provide a
positive test for the apparatus; however, the information obtained from


                                    22

-------
these tests served to establish the dynamic range of benzene levels which
might be encountered in later studies.  The results of these tests are given
in Table 9.  Based on these preliminary data the breath sampling protocol in
Appendix B was developed.

PRETEST EVALUATION OF FILLING STATION ATTENDANTS AND TANK TRUCK DRIVERS

     The pretest was designed to measure as many of the exposure and body-
burden parameters as possible for the nine participants.  Participants A and
B were selected as controls for smokers and participants F and G were selected
as control nonsmokers.  The other participants were occupationally exposed
for one sample collection period and presumably unexposed (nonworking)
during the second sample collection.

     After soliciting participation in the study, a schedule was set up with
the subject.  The schedule consisted of (1) initiation of personnel monitoring
of benzene levels in air, (2) 6 to 8 hours after initiation of air monitoring,
collection of a blood sample followed immediately by collection of a breath
sample, and  (3) collection of urine samples at this time for some partici-
pants.  This sequence was performed first on a workday, then on a nonworkday.
At least 18 hours lapsed between the last work exposure and the nonworkday
blood  collection.

Results and Discussion of the Pretest

     The results of the pretest of benzene body-burden are given in Table
10.  Inspection of the data indicates one line that is unique and uncharac-
teristic.  Subject C on the work day test had unusually high blood and
breath values.  The exposure of this participant was probably atypical due
to direct  contact with gasoline containing benzene.  The personnel monitor
may not have fully registered this exposure because of possible skin absorp-
tion and high air concentration gradients possible with this type of exposure.
The blood  and breath levels are not those expected to be characteristic of
environmentally exposed individuals.  For these reasons, Subject C's data
were excluded from the statistical evaluation.  Excluding the Subject C
workday data set, the correlation coefficients, slopes, and intercepts were
calculated for the various parameters.  Given sufficient data, the slopes
would  indicate the partition of benzene between the various phases such as
blood  and  breath.  The intercept indicates any residual benzene that is not
related to-the concentration in the phase being compared.   For example  the
intercepts for air vs. blood and air vs. breath are larger where smokers are
included in  the regression analyses indicating an exposure,  smoking, that is
not dependent on air benzene concentration.  These parameters are given in
Table  11.  In every category, the correlation is better for nonsmokers than
for smokers.  The uncertain nature of the benzene exposure through smoking
is the probable source of this variability.  Air exposure and blood levels
also correlate at the 95 percent confidence level in each category

Breath vs. Blood--
     The correlation of breath and blood levels is of particular concern
since  it would permit estimation of body-burden from breath levels   Sat'
al.  (8) measured blood and breath levels kinetically following a 2-hr


                                    23

-------
      Table  9.   BENZENE  IN BREATH—VALIDATION OF COLLECTION
Benzene
found in
breath
Sample
A (laboratory worker)
B (shop worker)
Cigarettes
(packs/day)
1.5
1.5
Sex
M
M
Weight
(kg) |Jg/m3
100 9.3b
60 19b
ppb
3.1
6.5
 Laboratory air was  found to  contain 8.2  Hg/m3  (2.7  ppb).
 Corrected for 1.7 M8/m3 in tne  air supply blank.
"No data on air levels.
                              24

-------
                   Table 10.  RESULTS OF PRETEST STUDY OF BENZENE BODY-BURDEN
Benzene

Sample
A
B
C

D

E

F

G

H

I


Occupation
Laboratory worker
Shop worker
Filling station
attendant
Filling station
attendant
Filling station
attendant
Laboratory worker

Laboratory worker

Filling station
attendant
Gasoline tanker
driver
Smoker
smoker
smoker
smoker
smoker

smoker

non-
smoker
non-
smoker
non-
smoker
smoker

non-
smoker

Sex (Kg)
M 100
M 60
M 60-70

M 80-90

M 90-100

M 60

M 70-80

M 60-70

M 60-70


preceding test
worked
worked
worked
off
worked
off
worked
off
worked

worked

worked
off
worked
off
Air monitor
(pg/m )
a
a
153
-
260
13
c
13
a

a

190
1.4
54
85
(ppb)
a
a
48
-
82
4
c
4
a

a

60
0.4
17
27
(lig/m3)
9,
19.
432.
6,
17.
11.
7.
1.
1.

3.

27
15
58
28
.3
.0
,0
,8
.0
,0
,2
3
,1

7





Breath
ppb
3
6
136.
2
5.
3,
2,
0.
0.

1.

8.
4.
18
8.

.1
.5
.0
.1
.4
.5
.3
.4
.4

,2

.4
.7

.4
levels found
Blood Urine
(ng/min) yg/L gg/L
87
110
2300 186.6 + 1.4
60 0.42 + 0.07
98 1.88
57 <0.35
40 0.28 + 0.06
8.4 <0.34
7.3 0.1

10.4 d 0.72 + 0.14e

200 1.14 0.93 + 0.14
130 0.40 + 0.06 0.41 + 0
513 1.30 + 0.34 2.4 + 0.4
246 1.93 + 0.12 1.6 + 0
 The individuals were not monitored; however,  air monitoring in the vicinity of their work
 areas indicated 8.2 yg/m3 of benzene in the air.

 Subject C had been repairing a fuel line on an automobile.   It is probable that he was
 exposed to more benzene than indicated by the personnel monitor.

 Cartridge lost due to breakage.

 Interfering peak on the gas  chromatogram.
^
"Sample  taken on the same subject  at a later date.

-------
                 Table 11.   LINEAR REGRESSION ANALYSIS OF THE BENZENE AIR EXPOSURE, BREATH AND
                                      BLOOD BENZENE LEVELS IN THE PRETEST

                                    Smokers                  Nonsmokers           All subjects
                             Slope  Intcp   r    n  Slope   Intcp   r     n   Slope  Intcp   r    n
                   	     ___...*     	      	           	             	                            	

        Breath3 vs.  Bloodb  0.051   0.011  0.55  5  0.025   0.28   0.75   5  0.026    0.32  0.64  10

        Aira'C vs.  Bloodb   0.0058  0.22   0.97  4  0.023  -0.018  0.997  4  0.0060   0.42  0.75   8

        Aira'C vs.  Breath3  0.032  14      0.57  6  0.52    1.0    0.72   5  0.056   14     0.30  11

        Blood vs.  Urine      -      -       -    --       -      -     -0.90     0.26  0.66   4

         Benzene concentration in pg/m3 was used.
isj        "Less than" values treated as one-half the given value.
Ox
         Some air data were estimated by the value found for laboratory air (8.2 |Jg/m3).


        Key:   intcp = intercept of linear regression
                  r = correlation coefficient
                  n = number of measurements

-------
exposure to 25 ppm of benzene.  From these data, the ratio of blood
concentration (HgA) to breath level (|jg/m3) is 0.016  (range, 0.012 to
0.018).  This ratio represents the average of the ratios for 0  60  120, and
240 min after the termination of the exposure.  A comparison of Sato's ratio
and the slopes obtained from the linear regression of  breath and blood
levels shows similar values in the case of nonsmokers  and all subjects.  The
similarity of the blood-breath ratios indicates that the blood/breath relation-
ship at the ppb level and ppm level of benzene exposure is similar.  The
correlation of the pretest blood/breath data (r = 0.64) is significant at
the 95 percent confidence level for all subjects (n =  10).

     Further justification for the use of breath as a  measurement  of body-
burden may be made based on the prediction that a large number of  blood
values will be less than the detection limit.  For blood, the significant
value has been estimated by obtaining blood from "unexposed" nonsmokers and
determining the benzene.  The average value of a total of six measurements
of blood from three individuals was 0.49+0.39.  At the 95 percent confidence
level, 1.6 [Jg/L of benzene would need to be found to be significantly differ-
ent from 0.49 MS/L-  It is improbable that every individual in even the
highest environmentally exposed population would have blood benzene levels
above  1.6 (Jg/L.  On the other hand, breath samples have shown benzene levels
discernible above the blank at all exposure levels.  The dynamic range is
thirtyfold  (highest over lowest values) for breath relative to about fourfold
over  a background of 0.49 Mg/L for blood.  This dynamic range is especially
important because three exposure groups were to be examined.  Blood levels
would not be  able to differentiate between the low and medium exposure
groups  and  only  if  the high exposure groups have individuals exposed to
significantly greater than 30 (Jg/m3 would detectable levels of blood benzene
be found.   It should be pointed out that the detection limit reported here
for blood  is  more than an order of magnitude lower than that previously
reported  in the  literature.  Reproducibility of the breath analyses is good
as indicated  by  the analysis of the replicate breath cartridge from Subject
A where the difference between the two determinations was 5.8 percent.

Urine--
      Urine  analysis was included for the last two subjects.  The values are
 similar to  those of blood as would be anticipated if no active transport
occurred  in the  kidney.  Current data are insufficient to assess the validity
of this relationship.

Air vs. Blood and Breath--
      Although air exposure and breath levels do not correlate well for the
nine  individual  samples, measurements of air exposure  are still necessarv
 for proper assessment of subject exposure in the study because of  the dailv
 fluctuations  of  benzene concentration in air and the short biological half-
 life  of benzene.  The cartridge used in the pretest was too bulky  to locate
 in the subject's breathing zone so it was attached at  the subject's waist)
For an environmentally  exposed subject, this would not be a serious problem
 since local concentration gradients would be small.  For the pretest subiect
however,  highly  localized exposures probably occurred  (e.g., height of gas    '
pump  nozzle), which may contribute to the lack of good correlation with
 exposure.


                                     27

-------
     If the test subjects are divided into two groups, those exposed to <30
Mg/m3 and those exposed to >30 [jg/m3, and one computes the coefficients of
variation (Table 12), an idea of the type of information one might obtain
for a high vs. low exposure group is obtained.  Table 12 indicates that the
relative variability in the breath levels (80.2 percent and 54.4 percent) is
much greater than for the blood levels (43.3 percent and 25.7 percent).
Since the sample individuals were selected to represent a wide range of
exposure levels, this would indicate that breath measurements may be more
sensitive than blood measurements to changes in exposure.

     The complicated kinetics of absorption and excretion of benzene make
detailed analysis of body-burden difficult.   It is clear from the pretest
that if exposure is sufficiently high, breath and blood levels are elevated
and the two correlate at the 95 percent confidence level.  Because of its
greater range, breath is more likely to differentiate between the high,
medium, and low exposure populations.  Although blood and breath correlate
at the 95 percent confidence level with 10 paired measurements, a larger
number of measurements would be desirable.  Since only the high exposure
population of the performance sites was expected to contain individuals with
measurable blood levels of benzene, the incorporation of this matrix in the
study was made in the interest of improving the sample size for the blood/
breath correlation.  The added effort to include the blood samples was small
compared to the information content.  The inclusion of blood samples from
the other two groups was unlikely to produce any significant additional data
because of the large number of nondetectable levels anticipated.

Effects of Smoking--
     Smokers had consistently higher benzene levels than nonsmokers.
Further evidence of the effect of smoking on benzene levels in breath is
found in the results of the pilot study presented in "Formulation of a
Preliminary Assessment of Halogenated Organic Compounds in Man and Environmen-
tal Media" (32).  In this pilot study, breath samples from residents of the
Love Canal area of Niagara Falls, NY, were analyzed for halogenated hydrocar-
bons and other toxic organics.  Quantitation of the benzene levels was also
performed.  The results are given in Table 13.  Smokers in this group had
breath concentrations ranging from 1.8 to 7.0 (Jg/m3 while nonsmokers had
breath benzene levels ranging from 0.69 to 0.90 pg/m3.  Pooling all the data
from the pretest and the Love Canal study for those individuals where occupa-
tional exposure was not contributory to benzene body-burden, one finds a
mean breath benzene level of 1.3i3 MS/1"3 with a standard deviation of 1.06 (n
= 7) for nonsmokers while the mean was 8.42 (Jg/m3 for smokers (standard
deviation was 5.38 with n = 10).  The difference is significant at greater
than 99.5 percent confidence level.

     Clearly, smoking is a major contributory factor to benzene body-burden.
Since the objectives of this study were to evaluate environmentally related
(nonsmoking) benzene body-burden, smokers were excluded from the main
study.  It was felt that inclusion of smokers would confound the data and
require a much larger sample size.
                                    28

-------
    Table  12.   SAMPLE  STATISTICS FOR BLOOD AND BREATH FOR
                  TWO INTERVALS OF AIR EXPOSURE
Air exposure
<30 (Jg/m3
mean
std. dev.
range
C.V.
>30 |Jg/m3
mean
std. dev.
range
C.V.
Breath

8.6
6.9
1.1 to 19
80.2

32.5
17.7
17 to 52
54.4
Blood (Mg/L)

.30
.13
.1 to .4
43.3

1.56
.40
1.14 to 1.93
25.7
C.V. = (std. dev./mean) x 100
   Table 13.  ESTIMATED LEVELS OF BENZENE IN HUMAN BREATH FROM
              "OLD LOVE CANAL" IN NIAGARA FALLS, NY
Participant no.
10009
10017
10025
10033
10066
10041
10058
10074
10090
Smoker or nonsmoker
smoker
smoker
nonsmoker
smoker
nonsmoker
smoker
nonsmoker
smoker
nonsmoker
Benzene
1.8 + 0
2.6 + 1
0.90
6.8 + 0
0.69
7.0 + 1
0.90
4.9
0.74 +
(|Jg/m3)
.12
.6

.8

.6


0.13
 Substances, U.S. Environmental Protection Agency, concerning the
 Pilot  Study at  Love Canal under EPA Contract No. 68-01-4731
 September  1978.
                              29

-------
                                  SECTION 6   .

                        SAMPLE DESIGN AND SELECTION
     The pretest with the filling station attendants and others at the RTI
laboratory served to test the chemical sampling and analysis methods.  The
pilot studies conducted in Harris County, TX (Houston), and St. Louis,
MO-Woodriver/Roxana/Hartford, IL (St. Louis), served as tests of the sample
design and field operations.  The strategy and rationale for the sample
design are discussed below with improvements that were made between the two
sites.

HOUSTON, TX, AREA SITE

Overview

     The sample design can be described as a two stage-design with stratifi-
cation imposed at the first-stage.   First-stage sampling units were clusters
of housing units, called segments,  within which a sample of eligible persons
was selected at a second stage of sampling.

     Two dimensions of stratification were imposed on the first-stage frame.
The dimensions employed controlled the distribution of the sample with
respect to geography and degree of exposure to benzene.  A total of 15
separate stratum cells were defined over each dimension.  A total of 30
first-stage units were selected with equal probability and without replacement.
The first-stage sample was equally allocated among the three exposure areas
making up the first dimension of stratification.

     For each first-stage unit, a second-stage sample of eligible person was
selected so that equal weights would apply within the three exposure areas.
The sample was selected with equal probability and without replacement.  A
total of 151 eligible persons were selected at the second stage.

Target Population

     The target population was the residents of Harris County, TX.  Three
geographical areas were designated as high, medium, and low exposure areas.

     The population was restricted to those individuals residing in the
target areas during the data collection period.  They also had to be in
their places of residence at night.
                                    30

-------
     The population was restricted to specific types of individuals.  The
following criteria were used in defining the target population:

     1.    25 to 50 years old;

     2.    Nonsmokers;

     3.    Nonoccupationally exposed; specifically, a person was not considered
          eligible if employed as a painter or in a service station,  garage,
          furniture repair shop, or chemical plant;

     4.    Not engaged in any hobbies involving exposure to high levels  of
          benzene; specifically, painting, building models, gardening,  or
          refinishing furniture;

     5.    Healthy individuals, i.e., taking no prescription medicine.

First-Stage Area Sample

Construction of the First-Stage Frame--
     The  sample was a probability sample of area segments.  Being an  area
sample, the sampling frame had to be constructed so that all  land area
defined in Harris County, TX, was included in the frame and no area was
included  more than once.  For this purpose, 1970 Census Enumeration Districts
 (EDs) and Block Groups  (BGs) were used.  These are units for  which information
about the number of housing units was available for use in defining sampling
units.

Stratification of the First-Stage Frame--
     The  first dimension of stratification was designation of three geographi-
 cal areas as high, medium, and low exposure areas.  The high  exposure
 stratum consisted of the blocks in the city of Houston, TX, shown crosshatched
 in Figure 1.  The tracts and block groups are listed in Table 14.

     The  medium exposure stratum was all of the area in the city of Houston,
 TX, that  was not part of the high exposure stratum.  The areas are shown in
Figure  1.

     The  low exposure stratum was all the area in Harris County, TX,  not  in
 the city  of Houston.  The areas are shown in Figure 1.

     To compensate  for  meteorological variability, the areas  radiate  in all
 directions.  This was accomplished by using a second dimension of stratifica-
 tion.   Each of the  three exposure areas  (major strata), were  divided  into
 five strata (minor  strata) as follows:

      1.   The  total  number of housing units in each of the three manor
           stratum was determined from the 1970 census data.

      2.   These numbers were divided by  5.  This was the approximate  size  of
           each of the five minor strata.
                                     31

-------
   Table 14.  HIGH EXPOSURE AREA, HOUSTON,  TEXAS -- MINOR STRATA
Tract    BGa   Tract   BG    Tract   BG    Tract   BG    Tract   BG
320 101
103
113
105
109
116
117
118
120
121
122
123
124
125
126
127
128
129
130
131
132
133
201
.219
220










320 221
222
223
225
224
226
301
323 102
504
505
506
























322 507
106
801
721
720
803
806
807
810
811
814
815
813
812
809
808
805
804
719
217
216
205
206
215
214










322 207
202
208
213
212
211
210
209
201
120
118
116
115
119
114
113
112
111
110
109
107
106
105
817
820
819
321 102
103
226
225
104
105
106


,321 107
108
109
110
224
223
221
220
111
112
113
114
115
116
219
311
312
320
321
322
323
324
227
228
229
301
302
303
304
305
332
331
330
329
406
a!970 Census.
 BG = Block Groups.
                              32

-------
CO
to
                                                       2milM
       Figure  1.   Map of Harris  County,  TX,  showing the various exposure areas; shaded = high exposure,
                  area enclosed  in bold  line = medium exposure, and remainder = low exposure.

-------
     3.   The file for each of the major strata was ordered by Tract, Block
          Group, and/or Enumeration District.

     4.   Starting with the lowest numbered Tract, Enumeration District
          and/or Block Group, the number of housing units was accumulated
          until this number was less than or equal to this minor stratum
          size.  The next Block Group or Enumeration District was added or
          omitted depending on whether the admission or omission made the
          stratum size closer to the ideal stratum size.  These tracts with
          their BG/EDs were called minor stratum 1.  This procedure was
          continued until all five minor strata were formed.

This stratification scheme defined 15 unique minor strata.

Selection of the First-Stage Sample--
     Sampling units were assigned to each stratum so that the sampling unit
contained approximately 25 housing units.  This was accomplished by dividing
the stratum's total number of housing units by 25 then rounding to the
nearest integer.  The total number of sampling units assigned to each
stratum is shown in Table 15.  Each stratum's total sampling unit was then
distributed across the stratum's EDs and BGs based on the ED's or BG's total
number of housing units.  To accomplish this step, a list of all EDs and BGs
in the stratum and their associated population was prepared from 1970
census data.  The housing units were then accumulated across this list so
that each ED and BG had an accumulated number of housing units representing
the sum of its total number of housing units and all housing units in EDs
and BGs previous to it on the list.  The accumulated number of housing units
was divided by the expected number of housing units in the stratum then
rounded to the nearest integer to determine the accumulated number of sampling
units.  The number of sampling units assigned to each ED and BG was then
determined by subtracting the EDs or BGs accumulated number of sampling
units from the immediately preceding EDs or BGs accumulative sampling
units.  Any ED or BG that did not have enough housing units to be assigned
one sampling unit was combined with other EDs or BGs in close geographic
proximity until at least one sampling unit could be assigned to the combina-
tion.  The two sampling units allocated to the stratum were then selected
from all sampling units in the stratum with equal probabilities and without
replacement.  This was done by selecting two random numbers in the range
from one to the total number of sampling units in the stratum.

Identification of Sampling Units Within EDs and BGs—
     When a selected sampling unit fell within a BG or combination of BGs,
its location was determined and housing unit data for the individual blocks
forming the BG were compiled.  The total number of sampling units assigned
to the BG was then distributed across the individual blocks based on their
total housing units.  Thus, if t were the total number of sampling units
assigned to a selected BG and t. was the number of sampling units assigned
to block i within the selected BG, then 0 < t. < t and It. = t.  Any block
                                          —  i —          i
assigned t. = 0 sampling units was combined with another block or block so
that the combination had a positive number of sampling units.  Any units,
single block, or combination of blocks having a positive number of sampling
units was called a segment with u. sampling units.  A sampling unit, k,


                                    34

-------
                          Table 15.  STRATUM HOUSING UNIT DATA AND  SAMPLE  ALLOCATION

Stratum
1
2
3
4
5
Total
High
Number of
HUs
670
721
674
718
687
3,470
exposure
Number of
SUs
27
29
27
29
29
141
Medium exposure
Expected
size of SU
24.82
24.86
24.96
24.76
23.69
24.61
Stratum
1
2
3
4
5
Total
Number of
113867
114003
113905
113880
113753
569,408
Number of
SUs
4555
4560
4556
4555
4550
22,776
Expected
size of SU
25.00
25.00
25.00
25.00
25.00
25.00
Stratum
1
2
3
4
5
Total
Low
Number of
HUs
4661
5042
4934
4834
4717
24,188
exposure
Number of
SUs
186
202
197
193
189
967

Expected
Size of SU
25.06
24.96
25.05
25.05
24.96
25.02
         HU = Housing unit.
         SU = Sampling unit.
CO
Cn

-------
within a segment, i, was defined as the cluster of housing units beginning
with the kth housing unit in a list of housing units to be compiled by field
personnel, 1 £ k < u., and then taking every u.th housing unit thereafter.
Thus, the random number identified the selected sampling unit located in a
particular segment and determined the particular sampling unit within the
segment by identifying it by a start number, k, and a rate 1/u.^.  The selected
samples are shown in Table 16.

     The selected first-stage sampling units were counted and listed (see
Section 7).  The actual housing unit counts are shown in Table 17.  The
expected number of housing units and the actual number of housing units in
both the high exposure and medium exposure areas were very close; however,
in the low exposure area the actual number was much greater than the expected
number of housing units.  There were too many housing units to screen.   To
reduce the screening costs, these segments were subsampled by taking an
equal probability without replacement samples.  The sample sizes are shown
in Table 16.

Selection Of the Second-Stage Sample

Construction of the Second-Stage Frame—
     The sampling frame at this stage was simply the list of all household
members who met the criteria specified above and who agreed to participate
in the Survey.

Selection of the Second-Stage Sample—
     From the list of eligible persons in each segment, an equal probability
without replacement sample was selected.  The sample sizes are shown in
Table 18.

Calculation of Sampling Weights

     For convenience, the notations used in this selection are summarized
below:

         h = 1, 2, . .,5 indexes the minor stratum cells.

         i = indexes the first stage units; values of i are nested within h.

      N(h) = the number of first-stage units, cluster of housing units
             called segments, in the h-th stratum.

      n(h) = the number of first-stage units, segments, selected from
             the h-th stratum (two).

     M(hi) = the number of households in the (hi)-th first-stage unit.

     m(hi) = the number of households selected from the (hi)-th first-stage
             unit.

     N(hi) = the number of eligible persons ,who agreed to participate at
             the time of screening in the (hi)-th first-stage unit.


                                    36

-------
                                  Table 16.  SELECTED SAMPLES -- HOUSTON, TX
CO
High exposure
Segment
Stratum "j° C5,?
1 35
35
2 116
116
3 13
31
4 21
119
5 22
17
Tract
320
320
320
320
322
322
322
322
321
321
Blocks
124
127,128
222
222
814
813
119
106,107
110
111
Sampling
units
1/1-1
1/1-1
1/5-4
1/5-4
1/1-1
1/1-1
1/1-1
1/4-3
1/1-1
1/1-1
Number of
rlUs (1)
24
25
71
37
34
574
17
78
21
37
Medium exposure
Segment
Tract
233
208
327
302
409
403
433
431
506
519
BG/ED
203
217,218
207
205
309,310
502
106
501
212
217
Sampling
units
1/1-1
1/1-1
1/3-1
1/1-1
1/1-1
1/23-12
1/1-1
1/3-1
1/1-1
1/2-1
Number of
HUs (1)
126
91
106 (2)
37
121
32
104 (2)
102 (2)
129 (2)
132 (2)
Low exposure
Segment
Tract
250
228
258
251
452
533
538
545
551
556
BG/ED
4
202,909
910
81A
310,311
431
106
144
174
175
142
Sampling
units
1/4-1
1/3-2
1/4-3
1/1-1
1/4-4
1/2-2
1/4-2
1/4-3
1/5-S
1/6-2
         HU = Housing unit.
         SU = Sampling unit.

-------
                            Table  17.  HOUSING  COUNTS  PER SEGMENT FOR HOUSTON,  TX
00
High exposure
No. housing units
Stratum Segment
1 11
12
2 13
19
3 14
15
4 16
17
5 18
10
70 Census
35
35
23
23
13
31
21
30
22
17
Actual
34
34
24
24
12
31
15
29
16
21
Medium exposure
No. housing units
Segment
28
29
27
26
25
24
20
21
22
23
70 .Census
24
25
24
37
34
25
17
26
21
19
Actual
27
22
24
35
34
25
21
26
19
14
Segment
38
33
39
37
36
34
31
35
30
32
Low exposure
No. housing units
70 Census
32
30
27
37
30
16
26
26
26
22
Actual
73
561
71
56
168
18
197
190
174
253
Sample Size
20
52
20
20
30
18
39
22
20
23

-------
Table 18.  SAMPLE SIZES OF ELIGIBLE PERSONS FOR HOUSTON,  TX
High exposure

Stratum
1

2

3

4

5



Segment
11
12
13
19
14
15
16
17
18
10
Total

Total
10
7
3
2
5
4
2
7
7
_2
49
Eligibles
Selected
8
7
3
2
5
4
2
7
7
	 2
47
Medium exposure
Low exposure
Eligible*
Responded
5
4
0
1
0
2
0
4
0
0
16
Segment
28
29
27
26
25
24
20
21
22
23
Total
Total
7
10
8
9
7
4
5
11
9
6
76
Selected
7
10
8
9
7
4
5
11
4
_6
71
Responded
0
1
3
2
2
2
1
3
3
1
18
Segment
38
33
39
37
36
34
31
35
30
32
Total
Total
0
22
8
11
9
4
9
10
7
12
90
Eligibles
Selected
0
10
2
2
3
2
2
3
3
6
33

Responded
0
4
0
1
2
1
0
1
2
5
16

-------
     n(hi) = the number of eligible persons who participated from the  (hi)-
             th segment.

     As discussed previously, the actual number of housing units found in
the selected first-stage units in the low exposure area was much greater
than the expected number; therefore, the first-stage units were subsampled
before they were screened.  Thus, the calculation of the weights for the low
exposure area requires a different formulation than the calculation of the
weights for both the high exposure and low exposure areas.

     Since the calculation of the weights for both the high exposure and
medium exposure areas is simpler than for the low exposure area, it will be
presented first followed by the weight calculation for the low exposure
area.

Calculation of the Sampling Weights for Both the High and Medium Exposure
Areas —
     The selection probability for an eligible person is given by

               2   n(hi)
for all

     j = 1, 2, . . ., n(hi)

eligible persons who participated in the (hi)-th segment.

     The sample weights are the reciprocal of the selection probabilities
given by

     w(hij) = l/7t(hij)                                                  (2)

for all

     j = 1, 2, . . ., n(hi)

eligible persons in the (hi)-th segment.

Calculation of the Sampling Weights for the Low Exposure Area —
     The selection probability for an eligible person is given by

     TrfM-n -  2    m(hi)  n(hi)
     nthljj ~ n(h)  M(hi)  N(hi)

for all

     j = 1, 2, . . ., n(hi)

eligible persons who participated in the (hi)-th segment.
                                    40

-------
     The sample weights are the reciprocal of the selection probabilities
given by

     w(hij) = l/7T(hij)                                                   (4)

for all

     j = 1,2, . .  ., n(hi)

eligible persons in the (hi)-th segment.  The sample weights are  shown  in
Table 19.

Nonresponse Adjustment

     A nonresponse compensation was made by using n(hi), the number of
eligible persons who participated in the sample, rather than the  number of
eligible persons selected for the sample in the weight calculations.  This
is the same as using the average value of the response variable of the
responding eligible persons for the nonrespondents in the same segment.

ST. LOUIS, MO, AREA SITE

Overview

      The sample design can be described as a two-stage design with stratifi-
cation imposed at the first-stage.  First-stage sampling units were clusters
of housing units, called segments, within which a sample of eligible persons
was selected  at a second stage of sampling.

      Two dimensions of stratification were imposed on the first-stage frame.
The dimensions employed controlled the distribution of the sample with
respect  to geography and degree of exposure to benzene.  A total  of nine sepa-
rate  stratum  cells were defined over each dimension.  A total of  18 first-stage
units were selected with equal probability and without replacement.  The
first-stage  sample was equally allocated among the three exposure areas
making up  the first dimension of stratification.

      For each first-stage unit, a second-stage sample of eligible persons
was selected  so that equal weights would apply within the three exposure
areas.   The  sample was selected with equal probability and without replace-
ment.  A total sample size of 75 eligible persons was designated  to be
selected at  the second stage.  Sample persons were defined as individuals
between  the  ages of 25 and 50 years old.

Target Population

      The target population consisted of the human population that resides in
St. Louis  City, St. Louis County, MO, and parts of Wood River, Roxana   South
Roxana,  and Hartford, IL.  Three geographical areas were designated as  high
medium,  and  low exposure areas.
                                    41

-------
Table 19.  SAMPLING WEIGHTS FOR HOUSTON, TX

City
1
1
1
1
1
1
1
1
1
1

1
1
1
1
1
1
1
1
1
1

Exposure
1
1
1
1
1
1
1
1
1
1

2
2
2
2
2
2
2
2
2
2

Stratum
1
1
2
2
3
3
4
4
5
5

1
1
2
2
3
3
4
4
5
5

Segment
11
12
13
19
14
15
16
17
18
10

28
29
27
26
25
24
20
21
22
23
High exposure
N(h)
27
27
29
29
27
27
. 29
29
29
29
Medium exposure
4,555
4,555
4,560
4,560
4,556
4,556
4,555
4,555
4,550
4,550

n(h)
2
2
2
2
2
2
2
2
2
2

2
2
2
2
2
2
2
2
2
2

N(hi)
10
7
3
2
5
4
2
7
7
2

7
10
8
9
7
4
5
11
9
6

n(hi)
5
4
0
1
0
2
0
4
0
0

0
1
3
3
2
2
1
3
3
1

Weight
27.000
23.625
__
29.000
—
27.000
__
25.375
__
--

__
22,775.000
6,080.000
10,260.000
7,973.000
4,556.000
11,387.500
8,350.833
6,825.000
13,650.000
                                                    (continued)

-------
00
                                              Table 19 (continued)

City
1
1
1
1
1
1
1
1
1
1

Exposure
3
3
3
3
3
3
3
3
3
3

Stratum
1
1
2
2
3
3
4
4
5
5

Segment
38
33
39
37
36
34
31
35
30
32
Low
N(h)
186
186
202
202
197
197
193
193
189
189
exposure
n(h)
2
2
2
2
2
2
2
2
2
2
M(hi)
73
561
71
56
168
18
197
190
174
253
m(hi)
20
52
20
20
30
18
39
22
20
23
N(hi)
0
22
8
11
9
4
9
10
7
10
n(hi)
0
4
0
1
2
1
0
1
2
5
Weight
	
5,518.298
—
3,110.800
2,482.200
394.000
_ _
8,334.091
2,877.525
2,079.000

-------
     The population was restricted to those individuals residing in the
target areas during the data collection period.  They also had to be in
their places of residence at least 6 hours prior to the time the measurement
was to be taken.  The population was restricted to specific types of indivi-
duals.  The following criteria were used in defining the target population:

     1.   25 to 50 years old;

     2.   Nonsmokers;

     3.   Nonoccupationally exposed; specifically, a person was not considered
          eligible if employed as a painter or in a service station, garage,
          furniture rapair shop, or chemical plant;

     4.   Did not engage in any hobbies involving exposure to high levels
          of benzene; specifically, painting, building models or refinishing
          furniture;

     5.   Healthy individuals, i.e., taking no prescription medicine.

First-Stage Area Sample

Construction of the First-Stage Frame—
     The sample is a probability sample of area segments.   Being an area
sample, the sampling frame was constructed so that all land area defined
below was included in the frame and no area was included more than once.
For this purpose, 1970 Census Enumeration Districts (EDs)  and Block groups
(BGs) were used.  These are units for which information about the number of
housing units was available for use in defining sampling units.

Stratification of the First-Stage Frame--
     The first dimension of stratificaton was the designation of three
geographical areas as high, medium, and low exposure areas.  The high expo-
sure stratum consisted of the following blocks in Wood River, Roxana,  South
Roxana, and Hartford, IL:

Stratum       Part                         Boundary Streets

  1            1         Satier Place, Clark, Thomas Street, Chaffer Ave.,
                         8th Street, State Highway 111, Old Edwardsville
                         Road, 6th Street, Esther Avenue

  2            1         State Highway 143, Town Limits, Kindall Drive,
                         Crestview Road, Arbitrary Line

               2         Arbitrary Line, Alton-Edwardsville Road, Town
                         Limits
                                    44

-------
Stratum       Part                         Boundary Streets

               3         Hedge Road, High Street, Washington Street, Park
                         Street, Roxanna Street, Park Street, ID Boundary,
                         State Highway 111, Roxana City Limits, Alton-
                         Edwardsville Road, Madison Street

               4         Rand Avenue, Olive Road, 7th Street, IT Railway

   3           1         Main Street, Evans Avenue, Wolcott Street, Arbi-
                         trary line

               2         Dulany Avenue, Ferguson Avenue, Madison Avenue,
                         8th Street, Esther Avenue, Old Edwardsville Road,
                         6th Street, Madison Avenue, Wood River Avenue,
                         E. Lorena Avenue

               3         13th Street, Chaffer Street, Thomas Street,
                         Dofier Avenue, 12th Street, Esther Avenue

               4         Trailer Park on Alton-Edwardsville Road

These  areas are shown crosshatched in Figure 2.

     The medium exposure stratum is all the area in St. Louis, MO, and St.
Louis  County roughly east of Interstate 1-244 and 1-270.  The census tracts
are  shown  in Figure 3.

     The low exposure stratum is all the area in St. Louis County, MO, roughly
west of Interstate 1-244 and 1-270.  The census tracts are shown in Figure
3.

     To compensate for meteorological variability, the areas radiate in all
directions.  This was accomplished by using a second dimension of stratifica-
tion.  Each of the two exposure areas (major strata), medium and low, were
subdivided into three strata (minor strata) as follows:

     1.    The total number of housing units in each of the two major strata
           was determined from the 1970 census data.

     2.    These numbers were divided by 3.  This was the approximate size of
           each of the three minor strata.

     3.    The file for each of the major strata was ordered by Tract  Block
           Group, and/or Enumeration District.

     4.    Using a tract map along with the file having data, the three minor
           strata were formed as shown in Figure 3.

     The high exposure area was subdivided into three strata (minor strata)
as  follows:
                                    45

-------
Figure 2.  High exposure strata for Wood River, Roxana, South Roxana and Hartford, IL.

-------

                   I —^   'fUlBsl 2188 I2193/2192\J   I
                   •   41B4   II    MltNOAU I m* I . _ _  i    '
                                  EASTERN ST. LOUIS COUNTY
Figure 3.   Strata for the medium exposure (1,  2,  and 3)  and low
            exposure strata  (I1, 2',  and 3').
                                   47

-------
     1.    The whole high exposure area was counted and listed.

     2.    The total number of housing units actually found was divided by 3.
          This was the approximate size of each of the three minor strata.

     3.    Using a tract map along with the listing data, the three minor
          strata were formed as shown in Figure 2.

     This stratification scheme defined nine unique strata.  Two segments of
approximately 50 housing units were selected from each minor stratum.

Selection of the First-Stage Sample--
     For the medium and low exposure areas, sampling units were assigned to
each stratum so that the sampling unit contained approximately 50 housing
units.  This was accomplished by dividing the stratum's total number of
housing units by 50 then rounding to the nearest integer.  The total number
of sampling units assigned to each stratum is shown in Table 20.  Each
stratum's total sampling units were then distributed across the stratum's
EDs and BGs based on the EDs or BGs total number of housing units.  To
accomplish this step, a list of all EDs and BGs in the stratum and their
associated number of eligibles was prepared from 1970 census data.  The
housing units were then accumulated across this list so that each ED and BG
had an accumulated number of housing units representing the sum of its total
number of housing units and all housing units in EDs and BGs previous to it
on the list.  The accumulated number of housing units was divided by the
expected number of housing units in the stratum, then rounded to the nearest
integer to determine the accumulated number of sampling units.  The number
of sampling units assigned to each ED and BG was then determined by subtract-
ing the EDs or BGs accumulated number of sampling units from the immediately
preceding EDs or BGs accumulative sampling units.  Any ED or BG that did
not have enough housing units to be assigned one sampling unit was combined
with other EDs or BGs in close geographic proximity until at least one
sampling unit could be assigned to the combination.  The two sampling units
allocated to the stratum were then selected from all sampling units in the
stratum with equal probabilities and without replacement.  This was done by
selecting two random numbers in the range from one to the total number of
sampling units in the stratum.

     For the high exposure area, the sample selection was essentially the
same with the exception that the housing units were accumulated over the
stratum.  The accumulated number of housing units was divided by the number
of housing units in the stratum, then rounded to the nearest integer to
determine the accumulated number of sampling units.  The two sampling units
allocated to the stratum were then selected from all sampling units in the
stratum with equal probabilities and without replacement.  This was done by
selecting two random numbers in the range from one to the total number of
sampling units in the stratum.

Identification of Sampling Units Within EDs and BGs—
     Recall that two random numbers were selected in the range from one to
the total number of sampling units in the stratum, thus establishing a one-
to-one correspondence between the sampling units and the random numbers.


                                    48

-------
               Table 20.  STRATUM HOUSING UNIT DATA AND SAMPLE ALLOCATION
                                      ST. LOUIS, MO
High exposure
Stratum
1
2
3
Total
Number of
593
651
615
1859
Number of
SUs
12
13
12
37
Expected
size of SU
49.42
50.08
51.25
50.24
Stratum
1
2
3
Total
Medium exposure
Number of
HUs
151125
151215
149561
451901
Number of
SUs
3023
3024
2991
9038
Expected
size of SU
49.99
50.00
50.00
50.00
Stratum
1
2
3
Total
Low exposure
Number of
HUs
26375
25854
26005
78234
Number of
SUs
528
517
520
1565
Expected
size of SU
49.95
50.01
50.01
49.99
HU = Housing unit.
SU = Sampling unit.

-------
Each random number, therefore, uniquely identified one sampling unit.  Once
the random number had been selected, it completely determined the selected
sampling unit.  The sampling unit then had to be located in an ED or BG
defined in the 1970 census.  If the sample fell partially or entirely within
a BG or combination of BGs, then the location of the sampling unit was
pinpointed using 1970 block statistics and their associated maps.  If the
sample fell within an ED or combination of EDs then the sampling unit was
located using 1970 county ED maps.

     When a selected sampling unit fell within a BG or combination of BGs,
its location was determined and housing unit data for the individual blocks
forming the BG were compiled.  The total number of sampling units assigned
to the BG was then distributed across the individual blocks based on their
total housing units.  Thus, if t was the total number of sampling units
assigned to a selected BG and t. was the number of sampling units assigned
to block i within the selected BG, then 0 < t. < t and It. = t.  Any block
                                          —  i            i
assigned t. = 0 sampling units was combined with another block or blocks so
that the combination had a positive number of sampling units.  Any unit,
single block or combination blocks, having a positive number of sampling
units was called a segment with u  sampling units.  A sampling unit, k,
within a segment, i, was defined as the cluster of housing units beginning
with the kth housing unit in a list of housing units to be compiled by field
personnel, 1 £ k £ u., and then taking every u.th housing unit thereafter.
Thus, the random number identified the selected sampling unit located in a
particular segment and determined the particular sampling unit within the
segment by identifying it by a start number, k, and a rate 1/u..  The
selected samples are shown in Table 21.

     The selected first-stage sampling units in the medium and low exposure
areas were counted and listed.  The high exposure areas had been cruised and
listed before the sample was selected.  The actual housing unit counts are
shown in Table 22.

Selection of the Second Stage Sample

Construction of the Second-Stage Frame—
     The sampling frame at this stage was simply the list of all household
members who met the criteria specified above and who agreed to participate
in the survey.

Selection of the Second-Stage Sample--
     The field supervisor was sent the screening questionnaires with the
selected participant's household member number circled in red for each of
the exposure areas  (high, medium, and low).  The sample for each segment was
divided into two parts.  The first part was made up of the blue screening
questionnaires and the second part was made up of white copies  of the blue
screening questionnaires.  The blue part was called the Sample  and  the white
part was called the Supplementary Sample.  The Samples and Supplementary
Samples for all three exposure areas are listed in Table 23.  The Sample was
given to the field  interviewer, while the Supplementary Sample  was  retained
by the supervisor.
                                    50

-------
                            Table  21.    SELECTED  SAMPLES FOR ST.  LOUIS,  MO
High exposure Medium exposure
Stratum Number of Location Sampling Number of Segment Sampling
HUs (all in Madison County, IL) unit HUs Tract BG/ED unit
1 56 Bounded
111, and
55 Bounded
Chaffer
Roxanna
by Elm St., State Highway 1/1-1 92 1122 102 1/2-2
Maple Ave. in Roxanna
by Central Ave., Thomas St., 1/1-1 91 1171 304 1/2-1
Ave., and Tydeman Ave. in
Low exposure
Number of Segment Sampling
HUs Tract BG/ED unit
53 2108.02 221,223 1/1-1
56 2112 815 1/1-1
          52      Bounded by  town limits, Kindall       1/1-1       43
                  Drive,  and  Big Bend Drive in
                  Kendall Hills

          54      Bounded by  Railroad, Maple St.,       1/1-1       49
                  Olive Road, East 1st St., Market
                  St.,  and Hawthorne St. in Hart-
                  ford.
         2188    604,605  1/1-1       47
                   606
         2201    503,504  1/1-1       57
2114    318,319    1/1-1
2152.02  103,104    1/1-1
         105
          55       Bounded hy Doerr Ave., South 12th    1/1-1
                  St.,  Esther Ave., 13th St., Chaffer
                  Ave.,  and Thomas St.  in Roxanna

          62       Bounded by Wood River Ave., East      1/1-1
                  Lorena Ave., 2nd St., Ferguson
                  Ave.,  3rd St.,  and Madison Ave.
                  in Roxanna
 57       1064     104     1/1-1
109       2106    105,106  1/2-1
                                    340      2212.02   114      1/6-4
                                     58      2204.01   309      1/1-1
HU = Housing  unit.

-------
                            Table 22.  HOUSING COUNTS PER SEGMENT FOR ST. LOUIS, MO
High exposure
No. housing units
Stratum
1
2
3
Segment
31
32
33
34
35
36
Actual
56
55
52
54
56
61
Segment
21
22
23
24
25
26
Medium exposure
No. housing
70 census
53
56
47
57
57
58
Units
Actual
33
53
48
86
86
52
Low exposure
No. housing Units
Segment
11
12
13
14
15
16
70 census
46
46
43
49
57
55
Actual
31
46
45
51
56
83
ro

-------
Table 23.  SELECTED SAMPLES AND SUPPLEMENTARY SAMPLES
                  FOR ST. LOUIS. MO


Segment
iiumber
Sample
Housing
unit
number
Supplementary sample
Household
member
number
Housing
unit
number
-Household
member
number

Selection
order
High exposure
31



32


33






34



35




36

11

12



3
16
33
34
9
43
50
11
17
20
26
35
43
43
19
37
40
46
9
13
13
39
43
61

20
44
17
43
65

02
02
01
01
01
01
02
02
01
02
02
01
01
02
01
01
01
02
02
01
02
02
01
02
Medium
02
03
01
01
01

3
39
25

16
43
9
19
39
15
35
33


19
41


5
24
36
53
43
--
exposure
40

17
27
35
35
01
02
03

02
02
02
02
01
01
02
02


02
02


01
01
02
01
02
__ ,

01

02
01
02
01
1
2
3

1
2
3
1
2
3
4
5


1
2


1
2
3
4
5
_

1

1
2
3
4
                                                 (continued)
                        53

-------
                     Table 23 (continued)


Segment
number
13

14

15







16








Sample
Housing
unit
number
22
33
8
44
2
9
22
34
45
52
56

12
18
20
22
32
38
60
72
156
Supplementary
Household
member
number
01
02
02
01
02
03
02
01
01
02
01

03
02
02
02
02
01
02
01
01
Hous ing
unit
number
4
22
23
43
14
22
6
9
51
1
54
51
52
10
128
60
72
58
56
138
92
'Household
member
number
02
02
01
02
01
01
02
02
02
01
01
01
02
03
02
01
02
01
01
02
02
sample

Selection
order
1
2
1
2
1
2
3
4
5
6
7
8
1
2
3
4
5
6
7
8
9
Low exposure
21








2
4
7
10
19
32



01
01
01
01
01
02



5
10
33
6
33
9
9
32
25
01
02
01
01
02
01
02
01
02
1
2
3
4
5
6
7
8
9
22
20
01
                                                        (continued)
                             54

-------
Table 23 (continued)

Segment
number
23










24









25



26




Sample
Housing
unit
number
24
35
35
37
39
42
47




46
48
48
50
64
70




90
94


15
15
38


Supplementary
Household
member
number
03
03
04
02
02
03
02




01
01
02
01
01
01




02
01


01
02
01


Hous ing
unit
number
22
25
47
31
46
29
1
39
46
2
18
52
1
62
46
70
54
34
6
68
50
44
2
96
94
10
26
12
33
38
Household
member
number
01
01
01
02
02
01
02
01
01
01
02
02
02
01
02
02
02
01
02
02
02
02
01
02
01
02
03
01
02
02
sample
Selection
Border
1
2
3
4
5
6
7
8
9
10
11
1
2
3
4
5
6
7
8
9
10
1
2
3
4
1
2
3
4
5
        55

-------
     In the field, the eligible persons listed in the Sample were contacted
in any order.  Every effort was made to get these people to participate.
After all procedures had been exhausted and still k, say, refused to parti-
cipate, then the supervisor selected the first k people from the Supplemen-
tary Sample.  These k people were then contacted to see if they would
participate.  Again, if p of these would not participate, then the supervisor
selected the next p from the Supplementary Sample.  This procedure was
continued until the required number of people from the segment agreed to
participate.

     After the required number of people agreed to participate, the inter-
viewer continued contacting people from the Supplementary Sample until one
more than was required agreed to participate.  The number of people contacted
to get the number was recorded.  The last person was not part of the survey;
however, if a member of the panel dropped out the last person was made part
of the panel.  If this happened, the interviewer continued to contact
people from the Supplementary Sample until one more than was required
agreed to participate, then the number contacted was updated.  The inter-
viewer was required to contact this extra person to fulfill the requirements
of sequential without replacement sampling (see Appendix C).  The sample
sizes are shown in Table 24.

Calculation of Sampling Weights

     The selection procedure of the second stage unit (i.e., the sample of
eligible persons) described above is called Inverse Sampling (See Appendix
C).  This leads to the Negative Hypergeometric distribution.  Using the
properties of this Negative Hypergeometric distribution, the sampling
weights were estimated.

     For convenience, the notation used in this section is summarized
below:

         h = 1, 2, 3 indexes the minor stratum cells.

         i = indexes the first-stage units; values of i are nested within h.

      N(h) = the number of first-stage units, cluster of housing units
             called segments, in the h-th stratum.

      n(h) = the number of first-stage units, segments, selected from
             the h-th stratum (2).

     N(hi) = the number of eligible persons who agreed to participate at
             the time of screening in the (hi)-th first stage unit, segment

     n(hi) = the number of eligible persons who participated from the
             (hi)-th segment

     p(hi) = one less than the number contacted to get one more than
             needed to agree to participate in the (hi)-th segment
                                    56

-------
Table 24.   SAMPLE SIZES OF ELIGIBLE PERSONS

High
exposure
Medium exposure
Eligibles
Stratum Segment
1 31
32
2 33
34
3 35
36
Total
Total
9
6
12
6
10
44
Selected
5
4
8
4
0
1
22
Responded
5
3
7
4
6
_0
25
Segment
11
12
13
14
15
16
Total
Total
3
7
4
4
15
18
51
Low exposure
Eligibles
Selected
2
6
3
3
14
17
45
Responded
2
2
2
2
4
8
20
Segment
21
22
23
24
25
26
Total
Total
15
1
18
16
6
8
64
Eligibles
Selected
14
1
7
10
3
42

Responded
5
0
7
6
2
23

-------
     An unbiased estimate of the sampling weight for an eligible person is
given by

     u(hij).


for all

     i = 1, 2, . .  ., n(hi)

eligible persons who participated in the (hi)-th segment.   The sample
weights are shown in Table 25.

Nonresponse Adjustment

     A nonresponse compensation was made by using p(hi) based on the number
of eligible persons who participated in the weight calculations.  This is
the same as using the average value of the response variable of the
responding eligible persons for the nonrespondents in the  same segment.
                                    58

-------

City

2
2
2
2
2
2

2
2
2
2
2
2

Exposure

1
1
1
1
1
1

2
2
2
2
2
2
Table 25
Stratum

1
1
2
2
3
3

1
1
2
2
3
3
. SAMPLE WEIGHTS FOR ST. LOUIS,
Segment
High
31
32
33
34
35
36
Medium
11
12
13
14
15
16
N(h)
exposure
12
12
13
13
12
12
exposure
3023
3023
3024
3024
2991
2991
n(h)

2
2
2
2
2
2

2
2
2
2
2
2
MO
N(hi)

9
6
12
6
10
1

3
7
4
4
15
18

p(hi)

5
4
8
4
9
-

2
6
3
3
14
17

Weight

10.800
9.000
9.750
9.750
6.667
--

2267.250
1763.417
2016.000
2016.000
1602.321
1583.471
Low Exposure
2
2
2
2
2
2
3
3
3
3
3
3
1
1
2
2
3
3
•21
22
23
24
25
26
528
528
517
517
520
520
2
2
2
2
2
2
15
1
18
16
6
8
14
-
7
10
3
7
282.857
--
664.714
413.600
520.000
297.143

-------
                                  SECTION 7

                              FIELD OPERATIONS
GENERAL PRINCIPLES

     The Survey Operations Center (SOC) was responsible for the conduct of
all field activities.  This included development of data collection instru-
ments; recruiting and training of field staff; supervision of actual data
collection; and receipt, editing, and preparation of data for entry and
conversion to machine-readable form.  These activities were the basis of SOC
involvement at the two data collection sites, Houston, TX, and St. Louis,
MO.  Field operations consisted of several steps leading to the sample
selection and actual data collection.  These included "counting and listing"
the selected sample areas and screening households for eligible individuals.
The sample, composed of eligible individuals who expressed willingness to
participate, was recontacted, the questionnaires administered, and appoint-
ments made to collect the biological samples.  At each site, appropriate
contacts were made with local health officials and EPA regional offices to
garner support for the activities.  Public relations contacts were made
through local officials.

DEVELOPMENT OF DATA COLLECTION INSTRUMENTS

     Appendix D of this report contains copies of the household data collec-
tion instruments developed for this study.  These instruments and supporting
information and rationale (also included in Appendix D) were provided to
EPA's Office of Pesticides and Toxic Substances for the submission to OMB
for clearance to execute the study.  While OMB approval was being obtained,
several pretest activities were conducted to validate methodology.  Two data
forms were developed.  The first, a Household Screening Questionnaire (HSQ),
was constructed to assist in the development of the sampling frame by provid-
ing a complete list of household residents that indicated their eligibility
and willingness to participate.  The matrix in Question 10 provided the
major input on eligibility by displaying age, smoking status, health status,
and vocational or avocational exposure to benzene.  Information collected on
the HSQ was summarized on a Household Screening Log (HSL) created to summarize
each segment of each stratum at each site.  A study questionnaire was develo-
ped to obtain demographic, occupational, and household information, as well
as the pertinent data on the samples that were collected.  Participant
Consent Forms were developed and specifically prepared for each stratum and
each site.  For each site, the appropriate local officials were listed as
                                    60

-------
contact persons.  The high exposure area consent form reflected the need for
a blood sample and the concomitant larger incentive.

RECRUITING AND TRAINING OF FIELD INTERVIEWERS

     RTI has a staff of field supervisors available in several sites across
the country.  St. Louis and Houston, the study sites, are the home areas for
two of the supervisors.  These supervisors were responsible for recruiting
and retraining interviewers for the study.  In addition, the supervisors
assisted the Survey Director in training the interviewers.  Two training
sessions were held at each site.  The first was a general introduction to
RTI and the study and specific details of the screening procedure.  The
second training session was held during the break between screening and the
administration of the study questionnaire.  This second training focused on
the study questionnaire and how the interviewers were to assist in the
scheduling and collection of biological samples.  This training involved the
chemistry group's field staff and the phlebotomist (nurse or medical technolo-
gist trained in collecting blood samples) from the health department in the
high exposure area.

     Throughout the screening process and the administration of the study
questionnaire, the field supervisor maintained day-to-day contact with the
interviewers and provided guidance, coordination, and problem resolution.
The RTI Survey Director maintained contact with the supervisors.  All forms
were examined by the supervisors for quality control and then sent to RTI
for processing.

DATA COLLECTION

     In both the Houston and St. Louis area sites, the field data collection
processes consisted of several steps.  Areas chosen, as described in Section  '
6, were "counted and listed" by the Field Supervisors.  This process consisted
of identifying an area by its boundaries and then driving or walking through
the area in a specified replicable manner and listing the address or identi-
fying  description of each apparent housing unit.  These segment listings
were examined for any necessary subsampling and became the basis for screening.
Each selected housing unit was approached and the screening questionnaire
was administered to any adult resident.  Housing units were visited repeatedly
until  the unit was interviewed, determined to be vacant, or no response was
obtained after multiple visits.

     Based on the data collected, a sample of eligible persons who were
willing to participate was chosen.  The field interviewers then returned to
the field, and  recontacted the selected individuals.  At the time of this
contact, the interviewer provided a more complete description of the study,
had the participant sign the consent form, administered the study question-
naire, collected the tapwater samples, and set up an appointment for the air
monitor to be installed.  A second appointment, the morning after the monitor
was positioned, was made to retrieve the monitor and collect the breath
sample and, in  the high exposure areas, the blood sample.
                                    61

-------
Houston

     Field activities were executed first in Houston.  Based on the experi-
ence gained, the St. Louis activities were planned for and accomplished in
a more effective manner.  January 1979 marked the start of activities.
Segment maps were sent to the field supervisor for the "counting and listing."
This activity was slow due to the rapid growth of the area, but was completed
in February.  Interviewers were recruited and hired during late February so
that the screening activities could be started in March.  By the March 19
training date, sampling of strata for screening was 80 percent completed.
During the last two weeks of March the remaining sampling was finished and
all areas screened.  Household data collection began in early April as the
samples were selected.  The Houston Health Department provided a phlebotomist
who began work in the high exposure areas in mid-April.  By the end of
April, only 45 air and breath samples had been collected and by the end of
the field collection period on May 5, a total of 50 samples had been collected
and 50 questionnaires administered.  Several problems were related to the
difficulties in data collection.  As alluded to before, the size and rate of
growth of the area was not considered when the strata and sample segments
were selected.  The travel time between strata precluded working in more
than one area in any one day, and indeed the travel to and from areas was
often extremely long.  Based on this experience, the selection of strata and
segments in St. Louis was reconsidered and the number of sample areas reduced.

     A second problem, which was partially overcome, was the change in
eligibility status of selected individuals between screening and actual data
collection.  Persons started smoking, became ill, or now had exposure to
benzene from a new job or from activities around the home (e.g., painting,
gardening, or hobbies).  In St. Louis this problem was partially alleviated
by reducing the time between screening and final data collection.  However,
some changes in eligibility were due to improper reporting during the screen-
ing interview.  This source of error is beyond the control of the field
staff.  Some percentage of over-sampling could relieve this problem as well
as the problem of persons no longer willing to participate.

St. Louis

     The timetable of activities was shorter in St. Louis than in Houston.
The "counting and listing" that started in late June was followed in July by
hiring of interviewers.  They were trained and started the screening in
August.  The sample was drawn and data collection was started with the
second training session in early September.  All three sample areas were
finished in early October.  The response rate was higher in St. Louis as was
the ease of data collection.  The interviewers were told to schedule appoint-
ments at specific intervals that allowed for equipment management and
transport.  In addition, appointments were grouped in areas on specified
days.  For example, early field efforts were concentrated in the high exposure
area.  Four days out of six were designated for data collection in the high
area.  After this area was completed, three days a week were spent in the
medium and low exposure areas.  Monday, Tuesday, and Wednesday were designated
for low areas during one week and for medium areas the next.  This made
planning and executing data collection much simpler and allowed for ease of


                                    62

-------
coordination with the phlebotomist from the regional office of the Illinois
Department of Health.  These problem areas and their potential resolution
will be of great assistance in future body-burden studies.
                  i
DATA RECEIPT AND PREPARATION

     As data were received, the questionnaires were logged in and were
subjected to a manual review for completeness.  This review, by the Survey
Director, was followed by an edit and range check.  After problems were
resolved, the data were converted to machine readable form and a data tape
was prepared for use in the analysis.  The only data entered were from the
study questionnaire and the results of the analyses of samples.   The two
data sets were linked by the study number generated on a series  of labels
and attached to all paperwork and samples.  This mechanism works extremely
well in studies requiring multiple questionnaires and multiple sample types
or replicates.
                                    63

-------
                                  SECTION 8

                   SAMPLE COLLECTION AND CHEMICAL ANALYSIS


     The methods developed for the pretest of filling station attendants and
tank truck drivers were used for the sample collection and chemical analysis
for the two study sites.  Some modifications were required to meet the
logistical problems of field sampling.  The sampling and chemical analysis
protocols are given in Appendix A with the modifications indicated below.

SAMPLE COLLECTION METHODOLOGY

Blood Collection

     Because of technical difficulties associated with sterilization of
glass syringes in the field, an alternative collection method for blood was
evaluated.  Venoject tubes (Kimble), which are designed for use with gas
chromatography, were evaluated for their suitability for collecting blood
samples for benzene analysis.  Three nonsmoker laboratory workers with no
known benzene exposure were used as donors for the evaluation.  Each sample
was analyzed in replicate.'  No detectable benzene was found for any of these
individuals.  The limit of detection was 0.1 ng/mL for all but two injections
were it was ~1 ng/mL due to an interferent peak.  Replicate injections did
not show the same peaks hence they were presumed to be an artifact of those
particular injections.  Therefore the Venoject tubes were used for the blood
collection.

Breath Collection - (Mobile Van)

     As indicated in Section 6, the block groups contained two to six partici-
pants and were, for the most part, widely distributed.  To sample this
population effectively, the sampling team had to have considerable mobility.
To sample a sufficient number of participants, at least two of these block
groups had to be sampled each day.  To accomplish this, goal the sampling
equipment was set up in a van.  The primary concern with this approach, as
the contamination of samples with benzene from auto exhaust in and about the
van, was circumvented by placing all parts of the sampling apparatus that
would be subject to contamination in glove bag(s) under a slight positive
pressure of helium (hydrocarbon free).  This excluded the majority of contami-
nated air from the sampling medium and the exterior surfaces of the cartridges,
                                    64

-------
SAMPLE CHEMICAL ANALYSIS--HOUSTON, TX

     The chemical analyses of the sample collected in the Harris County, TX,
study are summarized in Table 26 along with the pertinent meteorological
data.  The meteorological data cover the period of the air sampling.  It
should be noted, however, that the air samples were collected inside the
participant's home.  The locations of the various segments are presented in
Figures 4 through 6.  The segment designations are:  10-19, high exposure
group; 20-29, medium exposure group; and 30-39, low exposure group.

     The control and blank data for air, breath, and water samples are given
in Table 27.  The blank values showed a very high time-dependence.  Blank
values were plotted against time and the breath samples corrected according
to the analysis date.  Since sampling and analysis chronologies were parallel,
the  samples were corrected by the blanks that shared the closest chronology
to the samples.

     Considerable difficulty was encountered with the blood analysis.
Detection limits that had been on the order of 1 ng/mL for the pretest could
not  be duplicated for the samples collected in the field, shipped to the
laboratory, and stored for more than a month.  The primary problem was one
of interference in the GC/FID procedure because of large amounts of other
hydrocarbons.  An attempt to improve the selectivity of the method by using
a photoionization detector, which detects aromatic compounds selectively,
was  made.  The sensitivity, however, was not adequate to detect the ^20 pg
of benzene that would have been introduced from a 1-ng/mL sample.  In fact,
the  limit of detection was 20 ng/mL, approximately that found in current
literature  (34).  Several steps were taken to minimize the problems encoun-
tered with the blood analysis between the two study sites.  In the St. Louis
study, the blood was collected during the first period of the study, flown
back to  the laboratory and analyzed.  The blood samples were the first
analyzed to avoid contamination that might occur during storage.  The use of
glass capillary gas chromatography increased the sensitivity and reduced
interferences.  A limit of detection of less than 1 ng/mL was obtained with
the  GC/FID technique.

SAMPLE CHEMICAL ANALYSIS--ST. LOUIS

     The sample chemical analysis results for the samples collected in the
St.  Louis, MO-Wood River/Roxana/Hartford, IL, study are summarized in Table
28 along with the pertinent meteorological data.  The meteorological data
cover the period of the air sampling.  The air samples were collected in
the  participant's homes.  The locations of the various segments are presented
in Figures 7 through 9.  The segment designations are:  30-36, high exposure
group; 20-26, medium exposure group; and 10-16 low exposure group.

     Three breath samples were confirmed using GC/MS/COMP.  For two of
these, the replicate sample was analyzed by GC/FID and the results are
presented in Table 29.
                                     65

-------
                      Table 26.  BENZENE  LEVELS  FOUND  IN THE HOUSTON  STUDY  4/3/79-5/5/79
ON
ON
Date
4/3,4/79
4/3,4/79
4/3,4/79
4/3,4/79
4/5,6/79
4/5,6/79
4/7,8/79
4/7,8/79
4/10,11/79
4/9,10/79
4/11,12/79
4/11,12/79
4/12,13/79
4/12,13/79
4/12,13/79
4/13,14/79
4/13,14/79
4/15,16/79
4/16,17/79
Subject
no.
10231
10827
10835
10876
10819
10843
10850
10223
10496
10207
10488
10479
10983
11007
10991
10583
10520
10462.
10603
Segment
25
22
22
21
20
23
26
26
36
36
32
32
27
27
37
32
32
32
24

Air
(yg/m3)
12
10
20
26
33
25
10
10
4.5
3.4
27
9
8.0
2.0
13
18
17
13
4.4
Benzene levels
Breath Water
(yg/m3) (yg/L)
4.8 <1
a
1.8 <1
2.5 <1
2.4 <1
4.4 <1
1.0 <1
1.3 <1
<2
2.9 <2
4.0 <2
1.7 <1
a <1
2.0 <1
a <1
1.4 <1
0.7 <2
1.9 3
2.7 <1
Meteorology

Blood Temp.
(yg/L) (°C)
-b 13-17
-
-
-
7-22
-
20-21
-
22-23
18-20
20-28
-
22-27
-
-
12-26
-
18-27
19-24
Wind
Speed (kn)
6-9



0-9

4-8RF

9-16F
5-10F
7-11

5-10


5-8

5-12F
6-10
Direction
35



13-16

15

15
10
SF(10 16 34)

F(08)14(27)


36

09
10
                                                                                       (continued)

-------
Table 26 (continued)
Date
4/16,17/79
4/18,19/79
4/18,19/79
4/19,20/79
4/20,21/79
4/20,21/79
4/20,21/79
4/20,21/79
4/22,23/79
4/22,23/79
4/22,23/79
4/22,23/79
4/22,23/79
4/23,24/79
4/23,24/79
4/24,25/79
4/24,25/79
4/25,26/79
4/25,26/79
Subject
no.
10199
10751
10702
10678
10744
10769
10710
10652
10629
10694
10660
10611
10637
10728
10645
10595
10793
10439
10132
Segment
24
12
15
17
11
17
17
11
15
17
11
12
11
19
11
12
12
25
21

Air
(yg/m3)
11
13
11
18
b
6.7
6.7
8.6
6.1
3.1
5.2
31
32
35
16
146
<146
3.1
4.6
Benzene 1
Breath
(yg/m3)
2.0
3.0
a
3.5
0.8
NDC
1.4
2.8
3.5
14
1.7
5.6
9.0
NDb
3.1
5.1
8.6
3.8
2.3
Meteorology
eve Is 	
Wi i
	 W_LI
Water Blood Temp. 	
(yg/L) (yg/L) (°C) Speed (kn)
<1
<1 <20 18-19 5-12F
<1 <20
<1 <20 19-22 2-9TRF
<1 <20
<1 <20 17-19 3-6F
<1 <20
<2 <20
<1 -d 17-20 5-8F
<1 <20
<2 <20
<1 <20
<1 <20
<1.0 -d 13-23 0-5F
<1 <20
<1 <20 15-27 0-10
<1 <20
<1 - 19-27 5-10
<1
id
Direction

08-15

09

36


30




07-34

16-29

18

                                          (continued)

-------
                                             Table 26 (continued)
ON
00
Benzene levels
Date
4/25/26/79
4/27,28/79
4/27,28/79
4/27,28/79
4/29,30/79
4/30,5/1/79
4/30,5/1/79
5/2,3/79
5/2,3/79
5/3,4/79
5/4,5/79
5/4,5/79
Sub j ect
no.
10165
10447
10454
10074
10033
10090
10082
10272
10066
10041
10256
10371
Segment
21
35
30
30
27
29
22
33
33
34
33
33
Air
(yg/m3)
11
5.6
4.0
21
30
25
45
18
18
17
4.3
9.8
Breath
(yg/m3)
2.0
8.2
2.0
0.3
1.0
4.1
1.0
0.4
7.5
ND
a
a
Water
(yg/L)
1
1
1
1
1
1
1
1
1
1
1
1
Meteorology
Wll
R 1 nnH TVmn
(yg/L) (°C) Speed (kn)
_
12-26 4-9
-
-
15-22 4-9
16-24 3-10FS
22-24 7-10
-
-
18-27 4-13TR
13-16 5-10
-
id
Direction

(13) 02


06
(13) 9 (3)
13


16,32
34

        b
 Subject did not breathe into apparatus correctly.
 Not detected above apparatus blank.
 Sample lost.
 Not analyzed.
Interference peaks on the chromatogram.
Wind Key:
  Directions are those from which wind blows.  Indicated in tens of degrees from the north:
  for east, 18 for south, 27 for west.  Entry of 00 in the direction column indicates calm.
Meteorology Key:
  R = Rain
  F = Fog
  T = Thunderstorm
  S = Smoke
                                                                                                     i.e., 09

-------
                    WIND
                                                2miln
Figure 4.
Location of segments for the low exposure sample in Harris County, TX;  shaded  area is
high exposure area; Houston City limits = bold line.

-------
                                                                              GREENS
                                                                              BAYOU
Figure 5.  Location of segments for the medium exposure sample in Houston, TX.

-------
Figure 6.  Location of segments for the high exposure sample in Houston,
TX,

-------
       Table  27-  QUALITY  CONTROL  DATA  FOR HOUSTON STUDY
Media
Air





Breath












Water



Date analyzed
5/14/79

6/1/79

6/19/79

6/6/79
(4/14/79)3
C4/22/79)3
(4/22/79)a
6/13/79
(4/22/79)a
6/21/79
C4/26/79)3
7/3/79
C5/5/79)3
(S/5/79)3
(5/14/79)3
(5/14/79)3
7/18/79



Sample
Field Blank 1
Field Control 1
Laboratory Blank
Laboratory Control
Field Bank 2
Field Control 2
Field Blank 1

Field Blank 2
Field Control 1
Field Control 1

Field Blank 3

Field Blank 4

Field Control 2
Laboratory Blank
Laboratory Control
Field Blank lb
Field Control 1
Field Blank 2
Field Control 2°
Benzene
(ng)
18
814
28
Lost
97
600
11

39
806
680

56

122

1300
70
800
<1 Mg/L
11 Mg/L
<]_ Mg/L
27 Mg/L
% Recovery

80



50



81
68





130

80

110

270
 Refers to date control or blank was collected on the spirometer.

 Prepared in the laboratory before the sampling trip.
p
 Prepared in the field on 4/14/79.
                              72

-------
Table 28.  BENZENE LEVELS FOUND IN THE ST.  LOUIS STUDY 9/18/79-10/20/79
Date
9/18,19/79


9/19,20/79



9/20,21/79


9/21,22/79


9/23,24/79


9/24,25/79



Subject
No.
20537
20586
20552
20024
20032
20040
20057
20347
20362
20388
20420
20412
20511
20404
20370
20396
20339
20321
20909
20354
Segment
21
21
21
16
16
16
16
34
34
34
35
35
35
35
32
32
31
31
31
31

Air
17
22
32
20
19
18
22
19
3
18
6.2
6.2
23
11
-.
14
86
86
25
17
Benzene
Breath
1 (Hg/m3)
1.1
2.2
11
2.3
7.0
3.8
4.0
12.0
6.2
4.5
8.9
a
3.8
3.0
7.3
3.5
16
9.2
4.6
9.2
levels
Water
(Mg/L)
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
Meteorology
U-i
W J.
TJ "1 — „ J T^rtm-rt
Bioou lemp. 	
(pg/L) (°C) Speed (kn)
13-29 0-10
-
-
16-26 4-12
-
-
-
14-24 3-9
-
ND
11-18 3-12
ND
ND
10-22 0-6
-
-
16-25 0-7
ND
ND
ND

nd
Direction
H 35 (8)


H 12



H 16 (8V) 35


35


HF 12 (calm)


H 13 calm 22



                                                                   (continued)

-------
Table 28 (continued)
Date
9/26,27/79


9/27,28/79




9/28,29/79

10/1,2/79

10/2,3/79

10/3,4/79


10/4,5/79


Subject
No.
20446
20438
20503
20297
20271
20917
20891
20305
20313
20453
20461
20479
20594
20545
20610
20560
20602
20099
20073
20081
Segment
31
35
32
33
33
33
33
33
33
35
33
34
26
26
13
25
13
16
16
16

Air
(M8/m3)
17
17
17
35
3.7
3.8
3.8
25
6.6
52
17
18
3.9
3.9
4.0
9.9
13
_b
10
6.7
Benzene
Breath
(Mg/m3)
4.2
7.7
7.7
a
4.1
6.5
2.1
4.7
2.8
7.7
5.7
2.5
2.6
5.2
_b
5.1
7.0
_b
-
-
levels
Water
(HgA)
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
ND
Meteorology
W-i nH
W J.11U
"D"! /•k^\s4 T^TIVW-* ^ ^ — ^^ ^— ^ — ^— — ^^^— ^^—
rsiooa iemp.
(Mg/L) (°C) Speed (kn) Direction
ND 15-28 0-10 HF 16 (calm) 12
ND
ND
ND 14-28 0-9 HF 16 (calm) 22
ND
ND
-
ND
ND 19-28 3-11 H 17-25
ND
ND 14-23 14-23 31
ND
11-21 0-10 H 15V
-
8-16 6-10 27
-
-
4-15 4-14 F 28
-
-
                                         (continued)

-------
Table 28 (continued)
Date
10/5,6/79




10/8,9/79




10/9,10/79

10/10,11/79
10/11,12/79


10/12,13/79

10/17,18/79


Subject
No.
20115
20149
20131
20107
20123
20628
20644
20651
20669
20636
20156
20164
20172
20214
20198
20206
20677
20693
20248
20230
20222
Segment
23
23
23
23
23
24
24
24
24
14
23
23
15
15
15
11
25
21
12
12
16

Air
((Jg/m3]
10
30
30
44
46
4.2
4.2
3.4
118
125
_b
"b

5.1
b
b

b
4*3
-
_b
~b

b
Benzene
Breath
1 (|Jg/m3)
_b
-
11,
b
4.9
5.0
17,
b
8.8
26
_b
1.4
9.4
3.1
2B3

b
5.5
14
19
9.4
Meteorology
1 evels..
U-i n/1
wind
Water Blood Temp. 	 ~
(pg/L) (Pg/L) (°C) Speed (kn) Direction
ND - 12-19 8-20 (G25) 19-28
ND
ND
ND
ND
ND - 8-31 7-22 (G26) R33
ND
-
-
ND
ND - 3-12 3-16 29-26
ND
ND - 12-14 9-18 (G25) 18-25
ND - 15-23 9-19 32 (calm) 26V
ND
ND
ND - 1-12 6-17 32
ND
ND - 13-18 0-11 12 (calm) 12
ND
ND
                                         (continued)

-------
                                     Table 28  (continued)
Benzene ;
Subject
Date
10/18,19/79



10/19,20/79
10/20,21/79

No.
20578
20719
20735
20727
20685
20834
20255
Segment
26
. 14
24
24
21
15
11
Air
Breath
Meteorology
Levels
U-i nrt

Water Blood Temp.
(|jg/m3)(Mg/m3) Qjg/L) (pg/L) (°C) Speed (kn) Direction
63

-
b
9.1
3.5
7.4
18
4.4
9.3
6.1
6.2
11
7.0
ND - 17-23 7-15 (G23) Fie, 30, 21
ND
_ _
ND
ND - 22-25 12-17 (G23) H 17
ND - 22-25 18-19 (G33) 18
ND
 Sample containers damaged in transit.
 Chromatographic interferences prevented quantitation.
Meteorology Key:
    R = Rain
    F = Fog
    T = Thunderstorm
    S = Smoke

Wind Key:
    Directions are those from which wind blows.  Indicated in tens of degrees from the north:
    i.e., 09 for east, 18 for south, 27 for west.  Entry of 00 in the direction of column indi-
    cates calm.

ND = not detected.

-------
Figure 7. Location of segments for the high exposure sample in Wood River/Roxana/Hartford, IL
                                   (St. Louis study).

-------
00
              Figure 8.   Location of segments for medium exposure sample in St.  Louis,  MO.

-------
Figure 9.  Location of segments in the low exposure area in St.  Louis, MO.

-------
  Table 29.  CONFIRMATION BY GAS CHROMATQGRAPHY/MASS SPECTROMETRY

          Benzene found in replicate breath samples by

GC/FID                    GC/MS
(Mg/m3)                   (Mg/m3)                   Ratio GC/FID:GC/MS


  7.8                      15.4                            0.50

  lost                      8.9

 15 .6                      15.8                            0.99

                                                     Mean  0.74
                              80

-------
     Analysis of the blank and control samples for each of the media are
presented in Table 30.  Breath blank samples showed higher than usual
benzene values.  Excluding those showing obvious interferences, the average
was 4.2 + 1.9 [Jg/m3 while the average air blank corresponded to 1.5 + 1.2
|Jg/m3 (based upon a 30-L air sample).   The breath samples were not corrected
for this large blank because many of them were well below the blank values.
The source of this discrepancy is not obvious.  The tests are necessarily
performed under less than ideal circumstances in a mobile unit.  The opportu-
nities for contamination are always present and the blanks only partially
mimic the test conditions.  Neither the pretest nor the Houston breath
blanks were this high (1.2 |Jg/m3 for the pretest and 0.3 to 3.5 [Jg/m3 for
the Harris County study).
                                    81

-------
       Table  30.  QUALITY CONTROL  DATA FOR ST.  LOUIS  STUDY
Media
Air


Breath




Blood
Water
Date analyzed Sample Benzene (ng) % Recovery
10/31/79
11/1/79
11/12/79
11/12/79
11/14/79
11/14/79
10/27/79
11/2/79
11/7/79
11/8/79
11/16/79
11/16/79
11/21/79
11/21/79
11/21/79
11/21/79
10/11/79
10/11/79
10/26/79
11/26/79
Field Blank 1
Field Control 1
Field Blank I1
Field Control 1'
Field Blank 2
Field Control 2
Field Blank
Field Blank
Field Control
Field Blank
Laboratory Blank
Laboratory Control
Field Blank
Field Control
Field Blank
Laboratory Blank
Field Blank
Field Control
Field Blank
Field Control
34
885 85
13
614 60
89
1070 98
110
290a
1088 82
269a
152
778 63b
216
924 71b
280a
114
<0.2C
4.9C 98

-------
                                  SECTION 9

                                DATA ANALYSIS


     The following data were available for analysis of benzene levels in
both of the sites investigated (Houston and St.  Louis):

     1.   Overnight air samples of benzene levels collected by personal
          samplers that were placed in the respondent's homes,

     2.   Breath samples (5-20 minutes) for respondents,

     3.   Blood samples (only collected in the "high exposure" sampling
          strata),

     4.   Tapwater samples,

     5.   Limited meteorological data (wind direction, wind speed, tem-
          perature) ,

     6.   Individual answers to questions on a sample household question-
          naire  (e.g., demographic characteristics, where spent time, years
          in area, etc.), and

     7.   Individual sampling weights (the computation of these weights is
          described in Section 6).

 A  data  file was  constructed that  contained the above information for each
 individual along with  sampling strata identification (i.e., exposure = high,
 medium,  and low).

 SUMMARY STATISTICS

     Table 31 presents unweighted summary statistics for air, breath, water,
 and blood benzene  levels for the  two  sites (i.e., arithmetic means, medians,
 minimums, 90th percentiles, and maximum).  Also, the percent of samples
 above  minimal detectable is presented.

     Table 31 indicates that water and blood benzene levels in the samples
 in Houston and St. Louis are almost always below minimum detectable  (all
 blood  samples were below minimum  detectable and only 13 water samples in
 Houston were above minimum  detectable).  Thus, for the  remainder of  this
 section, analysis  was  performed only  on  air and breath  benzene levels.  For
 the two sites studied,  St.  Louis  had  higher levels of benzene in both air
                                     83

-------
 Table 31.   UNWEIGHTED SUMMARY STATISTICS FOR BENZENE LEVELS IN
         AIR,  BREATH,  WATER, AND BLOOD SAMPLES:  BY SITE
•a
Type of Sample
sample size
Houston
Air
Breath
Water
Blood
St. Louis
Air
Breath
Water
Blood
49
43
49
14
53
55
66
17
% Detected'
100.0
93.0
26.5
0
100.0
100.0
0
0
Arith.b
mean
14.55
3.07
0.30
0
24.0
7.05
0
0
Median
11.5
2.1
0
0
17
5.6
0
0
Minimum
2.0
0
0
0
3.4
1.1
0
0
90%d
30.1
6.9
1
0
50.2
13
0
0
Maximum
45.1
14.0
3.0
0
125.0
26.0
0
0
a
 Sample size = number of individuals sampled.

 Units:  air, breath = |Jg/m3; blood,.water = pg/L.

 Percent of samples above minimal detectable limit
  Min.  detectable:   Air [Houston] =0.5 jJg/m3
                    Air [St.  Louis] = 1 (Jg/m3
                    Breath = see Section 8
                    Water [Houston] = 1 |Jg/L
                    Water [St. Louis] =0.1 |Jg/L
                    Blood [Houston] = 20 pg/L
                    Blood [St. Louis] =0.2 pg/L

 90% = 90% of observations were less than or equal to this value.
                              84

-------
and breath samples (e.g., Houston arithmetic means = 14.6 |Jg/m3 for air and
3.1 pg/m3 for breath versus St. Louis means of 24.0 for air and 7.1 for
breath).  In almost all samples taken in both sites, air and breath levels
were above minimum detectable levels.

     In general, the benzene levels observed in the Houston and St. Louis
air samples are very low compared to published data relating health effects
to benzene inhalation.  For example, the observed air levels are three
orders of magnitude lower than the benzene odor threshhold and three to five
orders of magnitude lower than the occupational levels associated with
leukemia.  In fact, the current proposed NIOSH-recommended standard for
occupational exposure in air is 3.13 mg/m3 = 1 ppm.  Probable health effects
benzene data are discussed further in Section 10 of this report.

     In addition, the air and breath levels observed were lower than those
observed in RTI's pilot study in Durham of filling station attendents (on
the job attendants had air levels >150 [Jg/m3 and breath levels ranging from
7.2 to 27 ME/m3> see Section 5).  On the other hand, the levels in the
current study were somewhat higher than those observed in the Durham pilot
study and the RTI Love Canal study for nonsmoking individuals where occupa-
tional exposure was not contributory to benzene body-burden (these indivi-
duals had a mean breath level of 1.33 [Jg/m3 with a range of 0.69 to 3.7 (see
Section 5).  Note also that individuals who smoked in the RTI Durham pilot
study and the Love Canal study who were not occupationally exposed had a
mean breath level of 8.42 [Jg/m3 (range 1.8 to 19 pg/m3).  These breath
levels are similar to the levels observed in St. Louis (recall the indivi-
duals in the current study were nonsmokers and not occupationally exposed).
This discussion is summarized in Table 32.

     A  list of the air and breath levels along with sampling weights and
exposure strata for Houston and St. Louis are given in Appendix E, Table E-
1.  The sampling weights are described in Section 6.  Using these data,
Table 33 gives weighted summary statistics for air and breath levels in the
two sites.  In  general, the weighted means and geometric means are similar
to the  unweighted means and medians given in Table 31.  Recall that the
weighted means  give an estimate of the average level of benzene for the
entire  area under study  (i.e.,  (1) Harris County, Texas and (2) St. Louis
City, St. Louis County, and parts of Wood River, Roxana, South Roxana and
Hartford, Illinois).

DISTRIBUTION OF AIR AND BREATH LEVELS

     Figure 10  gives histograms of the air and breath benzene levels in
Houston and St. Louis.   In general, the distributions are skewed indicating
that perhaps the  data are better represented by a lognormal or exponential
distribution rather than the normal distribution.  This would indicate that
when making statistical  tests  of hypotheses on benzene levels, logs of the
data  should be  taken before tests are performed.  Also, sample medians or
geometric means are probably a better measure of central tendency than
simple  arithmetic means.
                                     85

-------
          Table 32.  SUMMARY OF RELATIVE BENZENE LEVELS
                                       Air levels
                  Breath levels
Probable Health Effects

     Leukemia
     Odor

RTI Pilot Study (Durham)

     On-the-job filling station
     attendents

RTI Pilot Study (Durham and
Love Canal)

     Nonsmoking, nonoccupa-
     tional exposure (NOE)

RTI Pilot Study (Durham and
Love Canal)

     Smokers (S), nonoccupa-
     tional exposure (NOE)

Current Study (NS, NOE)

     Houston
     St. Louis
3-15 x 105
   3 x 105
   >150
7.2 - 27
                   .69 - 3.7
   2-45
   3 - 125
                   1.8 - 19
0-14
1-26
                              86

-------
           Table 33.  WEIGHTED MEANS AND STANDARD ERRORS FOR AIR AND BREATH BENZENE LEVELS:  BY SITE
00

Houston
Air
Breath
St. Louis
Air
Breath
0
Sample size
49
43
53
55
Weighted mean
(yg/m3)
16.1
2.93
26.8
8.5
Standard error
of mean
1.84
.48
8.1
1.5
Weighted geometric
mean
(yg/m3)
12.4
2.51
15.0
6.81
 Sample size =  number of individuals  sampled.
 Weighted means are an estimate  of the benzene  level  for  air  and breath  in  the  area sampled (see
 sampling section).
'Standard error is  an estimate of  the standard  error  of the weighted mean.

-------
  a.

  22

  20

  18

  16
0) 1 2

V 10
   6

   4
                                      Breath
                                      Level
                                      (ug/m3)
         6     12    24   36

         Air Midpoint
                          04     8   12

                           Breath Midpoint
 Cum.
16
 Air
Level
                                                          Breath
                                                          Level
                                                          (yg/m3)
        0    40    80  120

         Air Midpoint
                         0     8    16    24

                           Breath Midpoint
                                               Cum.
                      6.22
                     17.0
                     25.0
                     50.2
                    125.0
    Figure 10.  Frequency bar charts  for benzene,   a.  air levels in Houston,
                TX; b. breath levels  in Houston, TX;  c.  air levels in
                St. Louis, MO, and d. breath  levels in St.  Louis,  MO.
                                    88

-------
MEDIANS BY SAMPLING STRATA (EXPOSURE STRATA)

     Table 34 gives sample medians and ranges for air and breath levels by
site and sampling (exposure) strata.  Recall that the sampling strata were
designed to indicate high, medium, and low benzene exposure.  In addition,
Figure 11 gives a plot of the 25th, 50th, 75th, and arithmetic mean for the
three exposure strata and Appendix Figures E-l through E-4 give plots of the
actual data by exposure strata and site.  Tests of significance were also
performed (using the analysis of variance, ANOVA) on the geometric means of
the three sampling strata by site for air and breath levels  (in the case of
the lognormal distribution, the geometric mean estimates the median of the
distribution).  The model for these ANOVA was:
               loge(B)i.j = M + Li + £„                                 CD
where
   log  (B) . . = log of the, benzene concentration.in air or breath  samples
     6    1J   for the j   individual in the i   exposure strata,

           (J = mean benzene level,

          L. = i   exposure effect  (high, medium, low),

        e. . = random error.
          ij
 The ANOVA tests the equality of the exposure effects, L., i.e.,  are the
 benzene levels the same  for the three exposure strata.  The results of these
 tests  indicated no significant differences between the exposure  strata.
 Note,  however, in Figure 11 that in St. Louis the air benzene median  is  the
 highest in the high exposure strata even though this higher median is not
 statistically significant.  Thus,  there is evidence in St. Louis of higher
 benzene exposure at Wood River versus the other two exposure strata.  Also
 note from Table 34 and Figure 11 that the air benzene levels for the  medium
 and  low exposure strata  are approximately the same in St. Louis  and Houston.
 That is,  the high exposure strata  in St. Louis is the principal  reason St.
 Louis  median air benzene levels are higher than those in Houston.

     RTI  also investigated the "power" of detecting various differences
 between the sampling strata based  on the medians, variances, and sample
 sizes  of  the air and breath observed data in Houston and St. Louis.   Here
 "power" is defined as  the probability of detecting a difference  between
 sampling  strata medians  when a difference actually exists.  The  computations
 were based on detecting  a difference between the low versus the  high  sampling
 strata medians.  The results are found in Table 35.

     In addition to the  above, the correlations between exposure strata  and
 air  and breath levels  were computed and are given in Table 36.   Table 36
 shows  that the correlations are relatively small, and Appendix Figures E-l
 through E-4 indicate why this is so.  For example, in St. Louis  for air
 Figure E-3 shows that  two relatively high air values in the medium and low
 exposure  strata  (125 and 118 (Jg/m3, respectively) result in a -0.02 correlation


                                    89

-------
   Table 34.   UNWEIGHTED SAMPLE MEDIANS AND RANGES FOR AIR vs.
           BREATH LEVELS BY SITE AND SAMPLING STRATA


                              Air       Breath
  City     Sampling strata  (yg/m3)    (yg/m3)    Approx.  sample size
Houston
TX


Low
Median
High
Median
Range
Median
Range
Median
T"l <-. « n. A.
13.0
(3.4-27)
11.0
(2-45)
12.0
1.7
(0-8.2)
2
(1-4.8)
3.1
16
18
15
                   Range

St. Louis  Low     Median    10.0        5.4
MO                 Range    (3.4-118)  (1.1-18)

           Median  Median    13.0        7,0
                   Range    (3.5-125)  (2.3-26)

           High    Median    17.0        4.7              7_
                   Range    (3.7-86)   (1.6-16)

 Tests of significance did not indicate significant differences be
 tween estimated medians across sampling strata.
                               90

-------
                        Mean
O = Median
D= 25  or  75 percentile

32-
30-

26-
24-
22-
20-
18-
16-
14-
12.
10.
8-

6.
4-
2.

Air (pg/m3) 28'
n 26-



q


<




c
1
t

)



D
24-
22"
20-
n is-
: : 16-
14-
0 12-
* 10-
,. 8-
CD
] 6-
1 4-
2-
(16) (18) (15)
Air
C



^






(



C
I


X
C
J
Cj


/
V
5



3 [
]
(

\
j



[
1


)






]
3

(18) (12) (23)
       Low     Med    High
             Exposure

           Houston, TX
             Low     Med    High
                  Exposure

               St. Louis,  MO

10-
9-
8-
7-
6-
5-
4-
3-
2-
1-

Breath (yg/m )


(12) (16) (15)





5
c
t



c
A
? 8 (
ft)
CD I
]

5
]
i
        Low    Med    High
              Exposure

           Houston,  TX

10-
9-
8-
7-
6-
5-
4-
3-
2-
1-

Breath
n
c

^

c

c
]

<

)

] '

C
)
5
{

^ [
J


•)
/
]


(17) (15) (23)
              Low    Med    High
                   Exposure

                St.  Louis, MO
                             ( )  " sample size
Figure 11.  Percentile  plots  of benzene levels in air  and  breath; by site
            and exposure  strata.
                                      91

-------
 Table 35.   POWER OF DETECTING VARIOUS DIFFERENCES BETWEEN THE
                 HIGH AND LOW SAMPLING STRATAa

Difference between
medians (%)
50b
100
200
3,
Computations based on
cance.

Houston


Air Breath
.52 .40
.89 .77
.98
a one-tailed test at

• f f\ j_ i • -i
St.


Air
.33
.64
.94
the 0.05 level


Louis


Breath
.63
.96

of signifi


median.
                             92

-------
      Table 36.   CORRELATIONS BETWEEN AIR AND BREATH LEVELS
                  VS.  EXPOSURE STRATA:  BY SITE


                            Houston,  TX             St.  Louis,  MO
                           Air       Breath        Air         Breath
Spearman                 + .06       + .23        - .02         .09

Unweighted Pearson       + .08       + .23          .03         .11
                             93

-------
between air and exposure strata.  Thus, the overall  low benzene  levels
bbserved in the study, coupled with limited sample sizes and a few outlying
values, result in levels in the various exposure strata that are not  signifi-
cantly different.

CORRELATIONS BETWEEN AIR AND BREATH BENZENE LEVELS

     Several correlations between air and breath benzene levels were  also
computed.  The results are given in Table 37  (recall air levels were  by
personal monitor in the participant's home).  In addition, Figures 12 through
14 present plots of breath by air benzene levels by  site.  The scales of all
figures are the same and this is why the Houston plot is clustered in the
left corner (recall Houston had lower benzene levels than did St. Louis).
The table and figures show that there is no apparent relationship between
the relatively low air and breath benzene levels in Houston.  On the  other
hand, in St. Louis there is some evidence of  a positive relationship  (unweigh-
ted Pearson correlation = 0.49).  However, care must be taken not to  overinter-
pret this result since the apparent relationship is based on only a few high
air and breath values; the majority of the values being clustered in  the
left corner of the plots.  The reason the weighted Pearson correlation
(=0.73) is higher than the unweighted correlation can be seen in Figure 13,
which shows the relative weight given each data point.  The figure shows
that the largest data point (air = 125 and breath =  26 |Jg/m3) receives a
relatively large sampling weight resulting in a moderately high-weighted
Pearson correlation.

     To further examine the relationship between the two media, the mean
breath levels were computed for air categories above and below 20 |Jg/m3, by
site.  These results are given in Table 38, which again indicates that in
St. Louis there is some relationship between  air and breath levels.   However,
a test of the breath means in St. Louis (6.33 vs. 8.76 |Jg/m3) for the two
air categories was not significant at the 0.10 level.

EXPOSURE BASED ON WIND DIRECTION

     Using meteorological data based on the prevailing wind, air and  breath
benzene levels upwind and downwind of benzene sources in Houston and  St.
Louis were examined.  The Houston results are given  in Table 39.  Note that
historical prevailing wind was used to define upwind and downwind in  Houston
since benzene sources were quite spread out and it was not considered
feasible to use daily prevailing wind.  In addition, three segments that
were downwind but outside the Houston city limits were not included in the
downwind strata (see Figure 4).  Table 39 shows that there is evidence of
higher levels for air and breath in the downwind strata (e.g., air medians
of 11 versus 14 |jg/m3).  However, tests of significance for these medians
were not significant at the 0.10 level (the model used for testing is the
same as that given in equation [1]).

     Table 40 gives upwind and downwind benzene levels for the Wood River
area (the high exposure strata in St. Louis).  In this case, the prevailing
wind on the day the measurements were taken was used to define upwind and
downwind strata.  Again, there is evidence that the  downwind strata have
                                     94

-------
  Table 37.  PEARSON AND SPEARMAN CORRELATIONS BETWEEN AIR AND
                 BREATH BENZENE LEVELS: BY SITE


Unweighted
Weighted
Unweighted log
Weighted log
Pearson
Houston St
-.08
-.11
-.14
-.09

. Louis
.49b
.73b
.24
.24
Spearman
Houston St.
-.07
X
X
X

Louis
.15
X
X
X
 Pearson correlations assume the data are normally distributed while
 Spearman correlations are based on ranks and do not assume normally.

 Significant from zero at 0.01  level.

X = Since Spearman correlations are based on ranks,  they are not com-
    puted for weighted or logged data.
                              95

-------
 (ng/m1)
BKEATH
                                  ...H..i_lllgh_Exposure
                                  H - Medium Exposure
                                  L - low Exposure
                                                                Spearman Corr.-- -.07
                                                             Unweighted Pearson • -.08
                                                              Weighted Pearson - -.11
                                                 "Sample" Size
                  MH

          M  H         >•
          L       M  H
           l<  H
         H LM  U ML     M
             H   H     L
                     LL
             H        L   L
"«••
 0
                                                                                           100
       Figure  12.   Site  = Houston.   Plot  of breath vs.  air;  symbol  is  value  of exposure stratum.

-------
HllFATM
                                        II = High Exposure	
                                        M = Medium Exposure
                                        L ™ Low Exposure
   Spearman Corr. - .15
Unweighted Pearson = .49
 Weighted Pearson « .73
                                           	Sample Size = 42
            H    H     "
          H   L
                    H
          L    L            H
          M         HH     H
               - M   H  HH
                M
          L H         II M L
          M    M
                    L
                              -L I
                                                                   75
                                                                                       loo
                                                        All
     Figure  13.   Site =  St.  Louis.   Plot of  breath vs. air;  symbol is value  of exposure  stratum.

-------
          -la
IO
00
           BKFATH
           lu
                                                   1 • Low sample weight
                                                   2 - Medium sample weight
                                                   3 - High sample weight
   Spearman Corr. • .15
Unweighted Pearson - .49
 Weighted .Pearson.a_fj.3.
                       I
                           	1  — -
                        1   3     3
                     1    2

                     a    z
                     \          u
                             I   1   31

                     21         132
                     I     I
                                2
                                                                                  ... 	1
                                                           so
                                                                                                    100
                                                                     AIR
                 Figure  14.   Site = St.  Louis.   Plot of breath vs.  air; symbol is  categorized value  of  weights,

-------
 Table 38.  SUMMARY STATISTICS FOR BENZENE  LEVELS  IN BREATH FOR
             TWO CATEGORIES OF AIR LEVELS:   BY  SITE
              Houston, TX, air levels
                       ((jg/m3)

Air levels    Mean    Mi,n.    Max.
             St. Louis, MO, air  levels
                        (|Jg/m3)
             Mean    Min.
                      Max.
                                      n
<20 ug/m3     3.18     0

>20 (Jg/m3     3.11     0
14

 9
30    6.33

11    8.76
1.1

2.2
18

26
29

13
                              99

-------
 Table 39.   SAMPLE MEDIANS FOR AIR AND BREATH LEVELS UPWIND AND
             DOWNWIND OF BENZENE SOURCES IN HOUSTON3


Air


Breath




Median
Range
n
Median
Range
n
Upwind
h
11°
2-33
25
2
0-8.2
22
Downwind

14
3.1-45
24
3.1
0-14
21


(NS)C


(NS)


ft
 Upwind and downwind by historical prevailing wind (not prevailing
 wind on day measurements were taken).   That is, downwind defined
 by segments_<20, 22, 23, 29 and 33.  See data listing in
 Appendix E and Figure 4.

 Units:   MS/™3•

C'NS = Not significant at the 0.10 level.
                               100

-------
Table 40.  SAMPLE MEDIANS FOR AIR AND BREATH LEVELS UPWIND AND
 DOWNWIND OF BENZENE SOURCES IN WOOD RIVER (THE HIGH EXPOSURE
                     STRATA OF ST. LOUIS)3

Air


Breath



Median
Range
n
Median
Range
n
Upwind
14
6.2-23
6
3.8
2.5-8.9
5
Downwind
17 (NS)b
3.7-86
17
5.5 (NS)
1.6-16
18
Upwind and downwind defined by prevailing wind on day the measure-
ments were taken.

NS = Not significant at the 0.10 level.
                             101

-------
higher air and breath levels but tests of significance were not significant
at the 0.10 level.  Thus, in both Houston and St. Louis there is some evidence
that higher air and breath levels are present downwind from the benzene
sources.

QUESTIONNAIRE VARIABLES

     Table 41 presents frequencies of various questionnaire variables for
the individuals sampled, by site.  A copy of the questionnaire is given in
Appendix D.  The table shows that over 60 percent of respondents were white,
almost all respondents considered themselves to be in good health, over 90
percent used air conditioning or an evaporative cooler, most used a municipal
water supply for drinking, 25 percent had individuals in the household who
had hobbies potentially involving benzene products (painting, furniture
refinishing, scale models, or gardening) and approximately 40 percent had
lived at their current address for 0-4 years.

     To further investigate the possible effect of these questionnaire
variables on individual benzene levels a one-way analysis of variance
(ANOVA) was run on air and breath levels for each of the variables in Table
40.

     The model for this analysis was:

               B. . = |J + Q. + £. .                                      (2)

where

         B.. = benzene level in air and breath for the j   individual
          "^   in the i   questionnaire variable category (e.g., age,
               health status, etc.)

           (j = mean benzene level

          Q. = effect of the i   questionnaire variable category,

         £.. = random error.

The ANOVA tested the equality of the Q.'s; i.e., are the benzene levels the
same for the various variable categories.  The results of the ANOVAs are
also given in Table 41 along with sample means by variable categories.  The
table shows that in general very few of the sample means were significantly
different.  This is undoubtedly due to relatively low benzene levels in the
two sites and the limited sample sizes.  The only significant results found
were the following:

     1.   In Houston

          a.   Air levels for "years lived in this area," "grow own  food,"
               and "household smoke";,
                                    102

-------
                         Table 41.  FREQUENCIES FOR QUESTIONNAIRE VARIABLES:  BY SITE
o
CO
Variable
I.

2.




3.


4.

5.

6.

7.

8.


9.


Sex

Race




Avg. hours out-
side each day

Hours away from
home on weekday
Hours away from
home on weekend
Current health
status
Ever treated for
anemia
Years lived in
this area

Way cool home


Categories
Male
Female
White
Black
Asian
Hispanic
Am. Indian/ Alaska
Other
0-2
3-5
±«
0-7
i8
0-4
±5
Excellent
Good
Fair
Yes
No
0-4
5-15
>16
Air cond. or
evap, cooler
fan only
Code
1
2
1
2

3

1
2
3
1
2
1
2
1
2
1
2
1
2
3

1
2
Houston
Results of tests of
significance
(units = Mg/m3)
N % Air means Breath
21
29
30
6
0
13
1
0
14
25
11
22
28
26
24
26
21
3
12
38
23
11
16

49
1
42.0
58.0
60.0
12.0
-
26.0
2.0
-
28.0
50.0
22.0
44.0
56.0
51.0
49.0
52.0
42.0
6.0
24.0
76.0
46.0
22.0
32.0

98.0
2.0
14.
14.
12.
13.

18.

13.
15.
13.
15.
14.
14.
13.
12.
16.
13.
14.
13.
9.
18.

14.
32.
4
7
8
2

7

4
8
3
3
0
6
8
4
8
9
8
9
8 (.09)
4

2*
0
3.
3.
2.
2.

4.

3.
2.
3.
3.
3.
2.
3.
3.
2.
4.
2.
2.
3.
3.

2.
9.
means
0
2
7
7

0

5
9
0
1
1
5
5
3
9
*" <(\ 01
-, \.\j. yj
6
9
1

9
0
St. Louis
N
24
44
45
9
2
0
1
11
38
19
11
21
47
25
35
41
27
0
8
60
11
24
33

63
5
Z
35.3
64.7
66.1
13.23
2.9
1.4
16.1
55.9
27.9
16.1
30.9
69.1
36.8
63.2
60.3
39.7
11.8
88.2
14.9
35.8
49.3

92.6
7.4
Results of tests of
significance
(units = gg/m3)
Air means
24.0
24.0
25.8
8.4

25.1

29.4
18.3
14.2
30.4
21.5
22.7
24.9
21.0
28.6
28.4
23.4
36.2
28.4
18.4

24.6
8.5
Breath means
6.2
7.5
7.4
5.7

6.9

7.2
8.0
4.8
7.6
6.8
6.2
7.6
7.9
6.0
5.2
7.3
7.7
8.7
5.8

7.1
5.8
                                                                                       (continued)

-------
Table 41 (continued)
Houston
Variable
10.

11.


12.

13.


14.


15.

16.


17.




Grow own Food in
home garden
Source of drink-
king water

Does anyone Ise
in household
smoke
Anyone have job
with benzene
prod.
Anyone have hobby
using benzene
prod.
Others in house
treated for
anemia
Age


Weight




Categories
Yes
No
Tap municipal
Supply
Other
Yes
No

Yes
No

Yes
No

Yes
No
Do not know
20-30
31-40
^41
100-125
126-150
151-175
176-200
>201
Code
1
2

1
2
1
2

1
2

1
2

1
2
2
3
4
2
3
4
5
6
Nb
8
42

37
13
10
40

5
45

12
38

5
43
2
19
23
8
15
9
12
13
1
Results of tests of
significance
(units = ug/m3)
% Air means Breath means
16.0
84.0

74.0
26.0
20.0
80.0

10.0
90.0

24.0
76.0

10.0
86.0
4.0
38
46.0
16.0
30.0
18.0
24.0
26.0
2.0
8.5(.08)
15.6

15.2
12.8
S:? <•«>

11.8
14.8

12.4
15.2

14.7
14.5
15.8
13.6
14.6
15.6
9.3
12.4
19.3
11.0
4.5
2.8

3.4
2.3
-(.04)

1.9
3.2

4.7
2.6^'°*'

3.8
3.0
2.8
4.0
1.5
3.6
2.2
2.6
3.6
2.0
St.
N
24
42

66
2
24
44

8
60

17
51

4
64
0
13
29
26
11
22
17
10
8
Louis
%
36.4
63.6

97.1
2.9
35.3
64.7

11.8
88.2

25.0
75.0

5.9
94.1
19.1
42.5
38.3
14.7
32.7
25.1
14.7
11.8
Results of tests of
significance
(units = vig/m3)
Air means
31.9
19.9
J.
24.1
17.0
18.1
27.3

21.8
24.4

23.4
24.2

5.1
25.1
20.2
26.5
22.2
31.2
23.1
17.9
13.9
32.7
Breath means
9.1
6.0

7.1
5.2
5.2
8.1

6.8
7.1

5.0
7.7

4.0
7.3
7.2
7.9
6.0
7.0
7.8
7.0
6.7
5.9

(.03)



(.04)





(.09)











                                         (continued)

-------
                            Table 41  (continued)
Houston
Variable
IS.



19.



20.


Height



Numbers eggs
eaten past
48 .hours

Yrs lived at
this address

Categories
<60
61-66
67-72
Z?2
0
1
2
±3
0-4
5-15
>16
Code
1
2
3
4
0
1
2
3
1
2
3
Nb
3
24
18
5
20
10
14
5
29
16
5
%
6.0
48.0
36.0
10.0
40.8
20.4
28.6
10.2
58.0
32.0
10.0
Results of tests of
significance
(units = ug/m3)
Air means
19.0
14.9
14.2
11.2
14.5
11.5
14.3
15.0
14.8
13.8
15.6
St.
Louis
Breath means N %
6.5
2.5
2.8
4.1
2.8
3.2
2.4
5.7
2.6
4.0
2.5
1
36
28
3
35
10
14
9
26
28
13
1.5
53.1
41.3
4.4
51.5
14.7
20.6
13.3
38.8
41.9
19.6
Results of tests of
significance
(units = ug/m3)
Air means
52.0
23.9
24.6
11.8
31.1
14.4
18.3
13.1
24.4
25.7
21.9
Breath means
7.7
7.4
7.0
3.3
6.9
6.4
7.0
8.6
7.2
7.4
6.6
 than N).
TJ = Number of individuals (Note:  Due to the fact that not all individuals had
 air and breath levels, the air and breath means are based on fewer observations
 The tests of significance tested if the benzene air and breath mean levels were
 the same for the various questionnaire variable categories.  (0.05) = significant
 at the 0.05 level of significance.  Blank = not significant at the 0.10 level of
 significance.

A
 Not tested because of limited sample size in at least one category.

-------
          b.   Breath levels for "treated for anemia," "household smoke,"
               and "hobby".

     2.    In St. Louis

          a.   Breath levels for "grow own food," "household smoke," and
               "hobby."

     ANOVAs were also run using model (2) with the log  of benzene air and
breath levels as the dependent variable.  The results were quite similar to
those given in Table 41 .

     In addition to the ANOVAs, stepwise regressions were also run on the
air and breath levels by site (these regressions were also run on the logs
of air and breath levels).  The general model for this analysis was
     B. = M + B-j^ (Ques. Var^.) + B2 (Ques. Var>2.) +...,+ e.    (3)

where

          BI, 62. .  .  = regression parameters,

          B. = the benzene level in air or breath for the jth individual
           J
          e. = random error
           J
The purpose of the stepwise regressions was to indicate which questionnaire
variables appeared to be the most important in predicting benzene levels.

     In brief, the stepwise regression procedure used consists of the
following approach.   The stepwise computer program finds the single-variable
model (i.e., air benzene level on only one variable) which produces the
largest R2 statistics (where R2 is the square of the multiple correlation
coefficient).  After entering the variable with the largest R2, the program
uses the partial correlation coefficients to select the next variable to
enter the regression.   That is, the program enters the variable with the
highest partial correlation coefficient with air benzene level (given that
the variables with the largest R2 are already in the model).  An F test is
performed to determine if the variable to be entered has a probability
greater than the specified "significance level for entry."  After a variable
is added, the program looks at all the variables already included in the
model and computes a partial F-statistic to determine if these variables
should remain in the model.  Any variable not producing a partial F signifi-
cant at the specified "significance level for inclusion" is then deleted
from the model.  The process then continues by determining if any other
variables should be added to the regression.  The process terminates when no
variable meets the conditions for inclusion or when the next variable to be
added to the model is one just previously deleted from it.  For the present
analysis, all variables in the final regression model were deemed significant
at the 0.10 level of significance.
                                    106

-------
     The results of running the stepwise regression by site are given in
Table 42 for both air and breath benzene levels and the log of air and
breath benzene levels.   The log results are similar to the original unit
results.  In general, the table shows that the correlations for the re-
gressions are relatively small (<0.60 for all cases).   In addition, the
results are difficult to interpret considering that the sign on the regres-
sion coefficients is quite often in the opposite direction from that expected.
For example, for breath in St. Louis, "does anyone else in household smoke"
has a positive regression coefficient (i.e., lower breath levels in households
with smokers, see Table 41) and "anyone have hobby using benzene products"
also has a positive regression coefficient (i.e.,  low breath levels in
households with benzene hobbies).   Therefore, in general, the stepwise
regressions do not indicate that the questionnaire variables explain the
variation in the air and breath levels.  Again, this is undoubtedly due to
low benzene levels and limited sample sizes.

     Finally, Table 43 gives answers to the questionnaire variables for four
selected individuals with relatively high air or breath levels (two individ-
uals in Houston and two in St. Louis).   The individuals'  air and breath
benzene levels are given in the table and Appendix Figures E-l through E-4
indicate why these four individuals were selected.  It is not obvious from
Table 43 why these four individuals had high air or breath benzene levels.
                                    107

-------
           Table 42.  RESULTS OF STEPWISE REGRESSIONS  BY SITE WITH AIR AND BREATH BENZENE LEVELS
                  AS THE DEPENDENT VARIABLE (ALSO LOG [AIR] AND LOG [BREATH] BENZENE LEVELS)
              Air
        [+]  Years in Area (8)
        [-]  Other Smoke (12)
        [-]  Years at Address (20)
              Houston

      Log (air)


[+]  Years in Area (8)
[+]  Grow Food (10)
       Breath
    Log (breath)
[-]  Other Hobby (14)
[-]  Other Smoke (12)
[+]  Race (2)
[-]  Other Hobby (14)
[+]  Race (2)
N2
R =(corr)
R=corr.
47
.27
.52
47
.19
.44
41
.33
.57
41
.25
.50
o
oo
                                                  St. Louis
[-] Eggs




N
R =(corr)
R=corr.
(19)




o 50
2 .08
.28
[-] Eggs (19)




50
.09
.30
[-] Grow Food (10)
[+] Other Hobby (14)
[+] Other Smoke (12)
[-] Other Job (13)
[-] Weight (17)
53
.34
.58
[-] Grow Food (10)
[+] Other Hobby (14)
[+] Other Smoke (12)
[+] Sex (1)

53
.35
.59
         Stepwise procedure set up so that all variables in final equation are significant at 0.10 level.

         Air column indicates that Air Benzene level = linear function (years in area, other smoke, years
         at address).   The correlation coefficient for this regression is 0.52.

        [ ] sign on regression coefficient (e.g., [+] = years lived in this area has a positive regression
            coefficient).

        ( ) question number in Table 39 (e.g., [8] = years lived in this area).

-------
Table 43.  ANSWERS TO QUESTIONNAIRE FOR SELECTED INDIVIDUALS WITH HIGH AIR OR BREATH BENZENE LEVELS

1.

2.



3.
4.
5.
6.
7,
Variable
Race

Race



Avg. Hrs . Outside
Each Day
Hours Away From
Home on Weekday
Hours Away From
Home on Weekend
Current Health
Status
Ever Treated for
Anemia
Houston St. Louis
Indiv. 1 Indiv. 2 Indiv. 1 Indiv. 2
air = 3.1 pg/ni = 45 Air = 125 = 118
Categories Breath =14 =1 Breath =26 =8.8
Male Male Male Female Male
Female
White White Other White White
Black
Asian
Hispanic
Am. Indian/ Alaska
Other
0-2 > 6 > 6 0-2 0-2
3-5
>6
0-7 > 8 > 8 0-7 > 8
>8
0-4 > 5 0-4 > 5 > 5
>5
Excellent Good or Good or Excellent Good or
Good Fair Fair Fair
Fair
Yes No No No Yes
No
                                                                              (continued)

-------
Table 43 (continued)
Variable
8.

9.


10.

11.


12.


13.


14.


15.


Years Lived in
This Area
Way Cool Homes

•
Grow Own Food
in Home Garden
Source of Drink-
king Water

Does Anyone Else
in Household
Smoke
Anyone Have Job
With Benzene
Prod.
Anyone Have Hob-
by Using Benzene
Prod.
Others in House
Treated for
Anemia
Categories
0-4
5-15
Air Cond. or
Evap. Cooler
Fan Only
Yes
No
Tap Municipal
Supply
Other
Yes
No

Yes
No

Yes
No

Yes
No
Do Not Know
Houston St. Louis
Indiv. 1 Indiv. 2 Indiv. 1 Indiv. 2
air = 3.1 [ig/m = 45 Air = 125 = 118
Breath =14 =_1 Breath =26 =8.8
5-15 0-4 5-15 5-15

AC or AC or AC or AC or
Cooler Cooler Cooler Cooler

Yes No No Yes

Tap. Mun. Other Tap Mun. Tap Mun


Yes No No No


No No No No


Yes No No No


No or No or No or No or
don't don't don't don't
know know know know
                                          (continued)

-------
Table 43 (continued)

16.
17.
18.
19.
20.
Variable
Age
Weight
Height
Numbers Eggs
Eaten Past 48
hrs .
Yrs Lived At
This Address
Categories
21-30
31-40
100-125
126-150
151-175
176-200
>201
<60
61-66
67-72
>72
0
1
2
>3
0-4
5-15
Houston St. Louis
Indiv. 1 Indiv. 2 Indiv. 1 Indiv. 2
air = 3.1 |Jg/m = 45 Air = 125 = 118
Breath =14 =1 Breath =26 =8.8
31-40 31-40 31-40 >41
176-200 176-200 100-125 >201
67-72 67-72 61-66 67-72
> 3 Missing None None
5-15 0-4 5-15 5-15

-------
                                 SECTION 10

                 DISCUSSION OF THE HEALTH EFFECTS OF BENZENE
     Benzene exposure has been implicated in the development of hemato-
toxicity in humans and animals (35,36,37).  In humans, benzene exposure may
induce chromosomal aberration, pancytopenia, and leukemia (36).  However,
information about the human health effects associated with benzene exposure
remain incomplete, and animal studies have been important in assessing and
predicting the toxicological effects of such exposure on humans.  It should
be noted that no animal model has yet been found for the leukemogenic effect
of benzene found in humans.  Furthermore, the exposures that produced
leukemia in some occupationally exposed individuals were varioualy estimated
between 10 and 1000 ppm (3.1 x 104 to 3.1 x 106 pg/m3).  The health effects
of benzene have been reviewed recently (6).  Some effects such as chromosome
breakage have been noted at chronic exposures as low as 2 to 3 ppm (time-
weighted average) (38).   When compared to the highest air benzene levels
found in the two sites,  0.04 ppm, the question of the potential health
effect revolves around the representative nature df the limited air sampling
in this study.and the safety factors that should be applied.

     The toxicology (6,7,12) and metabolism (3,12) of benzene have been
reviewed recently.  Benzene is efficiently (^30-50 percent) absorbed by the
lungs, which represents the major route of exposure, both occupational and
environmental.   Percutaneous absorption appears to be an unlikely contributor
to environmental exposure.  While oral ingestion of benzene may be significant,
it would be expected to appear equally in the control populations.

METABOLISM OF BENZENE IN ANIMALS

     The metabolism of benzene in animals appears to proceed along similar
pathways (39,40).  Of the routes of exposure, inhalation of benzene in
animals is the route that most closely approximates the primary means of
human exposure (40).  By this route of administration, benzene is absorbed
into the respiratory system where the reagent quickly reaches equilibrium
with alveolar air.  A portion of the benzene is eliminated unchanged in the
expired air, and the actual rate of removal is probably determined by the
vapor pressure of benzene in the lung and its concentration in the blood,
which in turn depends on the rate of absorption and metabolism (39,40).

     Of the remainder, another small portion of benzene, due to its high
lipid solubility, is partitioned and absorped in the adipose tissues and
bone marrow, where it may undergo further metabolism.  The majority of the
benzene is transported via blood to the liver, the principal organ of benzene
                                    1.12

-------
metabolism (39).   The rate of benzene metabolism and its tissue distribution
patterns depend on the route of administration in the animal and probably
reflect the circulatory sequence from the point of administration to the
liver (39).

     Once absorbed, benzene is rapidly distributed in the blood, where it is
eliminated or metabolized.  According to most reports, 30 to 50 percent of
the absorbed benzene is exhaled through the lungs.  The remainder is fixed
in the bone marrow, fatty tissues, and liver where it is metabolized.  Only
0.1 to 0.2 percent of the benzene is eliminated unchanged in the urine.
About 20 to 30 percent of the absorbed benzene is oxidized to phenol and
eliminated in the urine as the glycinate, sulfate, or glucuronide ester.  A
wide variety of minor and trace (<5 percent) metabolites have been detected
in humans and laboratory animals including pyrocatechol, hydroquinone,
hydroxyquinol and sundry ring-cleavage products.  It has been reported (41)
that up to 8 percent of the urinary phenols are excreted in the free form.
Excretion in the feces (in rabbits) appears to be negligible (42) with only
0.5 percent of a radioactive dose detected.  Benzene accumulation in the
body apparently occurs in the fat deposits (43), although only 2.6 percent
of a benzene dose was found in rabbit fat 2 days after exposure (42).  Thus,
the major excretion routes are exhalation of benzene and as a conjugated
phenol in the urine.

     Animal studies with labelled benzene indicate that the elimination of
the reagent via the lung and urine accounts for the principle detoxification
pathways  (39).  In expired air, a large, percentage is eliminated as unchanged
benzene.  In the urine, the principal oxidative products are the alkaline
salts of phenylsulfuric and phenylglucuronic acids.  The predominant form of
the final urine metabolites seems to be species-dependent in animals.  In
mice, the glucuronide form predominates while in rats, dogs, cats, and
humans, the sulfate formation is preferred (39).

     The kinetics of metabolism and excretion are not well-defined.  Benzene
is eliminated rapidly through the lungs, but apparently exhibits two rate
constants; one which applies over the first 60 min after exposure and one
which applies over longer periods (8).  The biological half-life of benzene
in humans has been estimated to be over 22 hr (44); however, blood levels
decay much faster than this after exposure.  Sato cited blood level half-
lives of 200 to 350 min for the latter portion of the decay curve (180-300
min) and much faster decay for the first 180 min.

     Handy and Schindler  (45) have derived a model for total body-burden of
various pollutants using the pollutant concentrations, volume of air inspired
 (tidal  respiratory volume times respiration frequency), absorption factor,
and body weight.  They assumed a continuous constant exposure, a biological
half-life  of 0.126 days, and, for lack of data at the time, 100 percent
absorption.  Using this model, they predicted a total body-burden of 2.02
mg/kg for  continuous exposure to 25 ppm.  Sato  (8) found ^0.2 mg/kg after a
2-hr exposure.  These numbers are reconcilable if the Handy-Schindler model
uses 30 to 50 percent as the absorption rate and the 2-hr exposure used by
Sato is extrapolated to steady state  for which the model was derived.  If
these relations are extrapolated to very low benzene exposure and blood


                                    113

-------
levels correspond to the body-burden, some estimate of potential blood
levels may be made.  Blood levels of 0.2-0.4 ng/mL would be anticipated  for
continuous exposure to 10 ppb, a high ambient air level.

MECHANISM OF BENZENE BIOTRANSFORMATION

     The proposed scheme for the hepatic oxidative metabolism is depicted in
Figure 15.  The metabolic sequence begins with the hydroxylation of benzene
by the heptatic mixed function oxidase to the benzene oxide.  This epoxide
may then undergo a number of metabolic conversions.  The epoxide may be
converted by the glutathione-S-epoxide transferase to yield phenylmercap-
turic acid.  Alternatively, the epoxide hydrase may catalyze the enzymatic
conversion of the epoxide to the diol, which then may be dehydrogenated to
catechol.  However, the predominant pathway for the metabolism of the epo-
xide may be its spontaneous rearrangement to the phenol.  Both the phenol
and catechol may be converted to the corresponding quinol, followed by the
conjugation with a number of reactants to form either phenylsulfate or
phenylglucuronides.

     Supportive evidence for the proposed oxidative metabolism of benzene
have come from both in vivo and in vitro studies (41,44,46,47).  Microsomal
preparations from mice, rats, rabbits, and dogs are capable of catalyzing
both the hydroxylation of benzene and the subsequent conjugation reactions
to yield either phenylsulfate or phenylglucuronide (44,47).  In addition,
Gerina et al. (46) have reported that benzene oxide can rearrange spon-
taneously to yield the phenol in the presence of mammalian hepatic micro-
somes.  However, attempts to trap the benzene oxide intermediates after
incubation of 14C benzene with rabbit liver microsome in vitro have failed
and may reflect the transient nature of the intermediate, which has a half-
life of about 2 min in aqueous solution (48).

     Moreover, the efficiency of the hepatic mediated oxidative metabolism
of benzene can be markedly altered by chemical inducers and inhibitors.
Pretreatment of animals with SKF-525A, an inhibitor of microsomal enzymes,
prior to benzene administration can depress both phenol and glucuronide
formation in rats.  Alternatively, treatment of animals with phenobarbital
induces the content of cytochrome P450, the dominant heme protein in bio-
logical hydroxylation reactions, and a number of hepatic microsomal enzymes
(45).  Liver microsomes from pretreated animals are capable of metabolizing
benzene more efficiently, increasing severalfold the concentration of
conjugated and unconjugated phenols over control animals (47).  More impor-
tantly, these researchers indicate that phenobarbital-treated rats exhibit a
greater resistance to leukopenic action of benzene.

     The molecular site of benzene hematotoxicity is still uncertain.
Although benzene is found to disrupt both DNA and RNA synthesis, the interrup-
tion appears after toxicity is manifest (48).  At the translational level,
benzene metabolism in animals has been associated with the disaggregation of
polysomes to smaller aggregates in rabbit reticulocytes.  This effect may be
reversed by addition of hemin (37,48).  Freedman (49) has proposed that the
primary toxic effect of benzene at the cellular level is probably the
                                     114

-------
              Benzene (100%)
                                       Expired unchonged (40%)
                               JAHH

phenyl mercopturic              benzene
   ocid (0.5%)                  oxide
                   glutothione
       NHAc
       I      tronsferase
 SCHj-CH

       COOH

hydroquinol (5%)

        fiOH

HO'
              epoxide
              hydrase

       spontaneous


phenol (23-50%)

        nOH     •>
                                                            Trcns-Trans
                                             Cis-Cis     muconic acid (1.3%)
                    benzene glycol (0.3%)      muconic ocid       COOH
                               H
                               OH 	»
                               H
                                                         OH
                                                                      COOH
                                                                            COOH

-------
inhibition of reticulocyte heme synthesis at or before the aminolevulinic
acid (ALA) synthesis step.  Heme synthesis appears to be maximal in the
earliest precursor cells and to decrease with cell maturity, while globin
synthesis increases with cell maturity.  These researchers suggest that heme
and globin syntheses are interdependent and that heme synthesis may be
necessary to increase intracellular hemin concentration to a level to allow
erythropoiesis to proceed.  Benzene, because of its lipid solubility, may
prevent heme synthesis in early bone marrow cells, leading to development of
aplastic anemia in treated animals.  Induction of ALA synthesis is related
to several hormones and the inhibition of this reaction by benzene may
explain the higher susceptibility to benzene-induced hematotoxicity in the
female as compared to male animals (37).  These findings in rabbit reticulo-
cytes, although important in deducing the hematotoxic mechanism, need to be
interpreted and extrapolated with care to human clinical situations.  More
work with animals is required to determine the toxicity level of benzene in
bone marrow, especially the mechanism of concentration, and its subsequent
development to the pathological hematological conditions.

BENZENE TOXICOLOGY

Benzene Toxicity

     The general effects of benzene toxicity are shown in Table 44.  Clinical
toxicity due to benzene exposure may be divided into two categories:  acute
toxicity and chronic toxicity.  Acute benzene toxicity is generally associated
with CNS stimulation, followed by depression and respiratory failure (50).
These toxic effects in animals appear to be independent of the route of
administration and require concentration^ ) many times in excess of that to
induce chronic toxic effects.  Chronic benzene toxicity, however, leads
essentially to a disorder of the hematopoietic system with the primary
effect in animals localized to the bone marrow (37,50).  Macrocytosis,
characterized as appearance of giant abnormal red blood cells, may serve as
an indicator of benzene-induced hematotoxicity.  Another common clinical
sign of chronic benzene toxicity is pancytopenia, defined as a decrease in
circulating erythrocytes, granulocytes, and platelets (see Table 45).  The
marked leukopenia is probably due to the shorter lives of the peripheral
leukocytes in benzene-treated animals.  These leukocytes may also exhibit
altered function such as a decrease in phagocytic function and in alkaline
phosphatase activity.  Other chronic effects noted in animals include
changes in levels of urine and blood porphyrin and aminolevulinic acids,
suggestive of altered heme synthesis.  Severe pancytopenia is generally
associated with aplastic anemia, a condition characterized by a decrease in
the number of hematopoietic precursor cells within the bone marrow.  The
putative metabolite(s) of benzene responsible for the hematotoxic effects is
still uncertain.  Young animals and female animals appear to be more suscep-
tible to benzene toxicity (see Table 46), but these findings need to be
verified.  In general, available data on dose-responsive hematotoxic effects
in animals are difficult to assess because of variation in dose and route of
administration, age and species of animals, and in the parameter of toxicity
studied.
                                    116

-------
Table 44.   GENERAL EFFECTS OF BENZENE INTOXICATION
Authors
Carpenter
et al.
Gerarde and
Ahlstrom
Hough and
Freeman
Jenkins et al.
Nahum and
Hoff

Sidorov
Withey and
Hall
Year Treatment
1944 Inhalation
35,000-45,000
ppm
1966 Sc injection,
1 mL/kg x 14
1944 Inhalation
600-1,000 ppm,
1970 Inhalation, 817
ppm x 30 days
1934 Inhalation, air
saturated with
benzene vapor

1972 Inhalation,
periodic or
continuous
1975 Intubation
Species
Rabbits
Rats
Dogs
Rats,
Guinea
pigs, Dogs
Cats,
Monkeys

Mice, Rats,
Guinea
pigs,
Rabbits
Rats
Comment
Slight anesthesia in approxi-
mately 4 minutes; deaths
occurred in 22 to 71 minutes,
indicating wide differences in
tolerance to benzene narcosis
Leukopenia
Leukopenia, shorter survival
times
Slight drop in WBC count
Ventricular extrasystole or
periods of ventricular tachy-
cardia which occasionally
terminated in ventricular
fibrillation
LD50 studies on joint action
affected more by continuous
exposure than by periodic
LD50 studies on joint action
of benzene and perchloro-
                                             ethylene.   Effect was additive,




                                                          (continued)

-------
                                             Table  44  (continued)
Authors
Wolf et al.
Yakushevich
Year Treatment
1956 Inhalation, 80-
88 ppm for 6
months
Oral, 1 mg/kg/
day for 6
months
1973 Inhalation, con-
tinuous and
periodic
Species
Rats,
Guinea
Pigs,
Rabbits
Rats
Comment
Milk leukopenia, splenic and
testicular degeneration
Continuous inhalation caused
more pronounced effects than
periodic inhalation
oo
        Source:  Leong, B. K. J., Experimental  Benzene  Intoxication,  J.  Tox.  Environ.  Health Supplement, 2_,
               (1977), pp. 46.                                                                             ~

        Sc = subcutaneous.

-------
                   Table  45.   BONE MARROW CHANGES DUE TO BENZENE INTOXICATION
Authors
Das et al.
Das et al.
Year
1969
1969
Treatment
Sc injection
Repeated sc
injection
Species
Guinea
Pigs
Guinea
pigs
Comment
Leukocytosis followed by
leukopenia and granulocyto-
penia in 6-8 days.
Removal bone marrow hypo-
plasia.
 Gerarde
 Koike et al.
1956
1959
Moeschlin and
Soeck
Steinberg
1967
1949
Sc injection
1 mL/kg x 14
days
Rats
1 mL/kg daily
x 5 weeks, or
2 mL/kg daily
x 3 weeks

Sc injection
2 mL/kg x 1
-1 mL benzene
per rabbit
Rats
Rabbits
Rabbits
Leukopenia, involution of
the spleen and thymus, and
a decrease in femoral
marrow nucleated cell count
and nucleic acid.  The injury
was reversible.

A rapid fall in femoral
marrow nucleated cell count
and in DNA-P content per
dry weight of bone marrow.

Severe inhibition of DNA-
synthesis in bone marrow
cells.

Only primitive reticular
cells during regeneration.
Benzene inhibits cell divi-
sion and maturation.
Source:  Leong, B. K. J., Experimental Benzene Intoxication, J. Tox.  Environ.  Health Supplement, 2,
         (1977), pp. 51.
Sc = subcutaneous.

-------
                             Table 46.   EFFECTS OF AGE  AND SEX ON BENZENE TOXICITY
          Authors
Year
    Treatment
Species
            Comment
        Avilova
        Desoille et al.
1971
1961
N3
O
        Desoille et al.
        Gadaskina et al.
1962
1973
        Ito
        Ito
        Ito
1962


1962


1962
Technical details
not available
Sc injection of
benzene, 0.1 g/kg
Sc injection of
benzene, 0.1 g/kg

Technical details
not available
Inhalation
Inhalation
Inhalation ex-
posure, daily
for 2 months
Rats
Rabbits
Rabbits
Rabbits
Mice
Rats
Rats
Young rats were affected
more severely and for a
longer period than older rats

Female rabbits or male rabbits
previously feminized by castra-
tion or injection of estrogen
were more sensitive to the
leukopenic effect of benzene

Gestation has no effect on the
leukopenic effect of benzene

Young rabbits are more sensi-
tive to benzene poisoning,
probably because of less
efficient conjugation with
salfuric and glucuronic acids

1 hour LC50 =14,500 ppm for
both male and female mice

Hematological changes appeared
faster in females than in males

Ovariectomized or testosterone-
treated female rats showed early
development of hematological
changes
                                                                                       (continued)

-------
                                     Table 46 (continued)
   Authors
Year
    Treatment
Species
            Comment
 Kimura  et  al.
1971
Manyashin et al,
1968
Oral
Intubation
Rats
Rabbits
Benzene is more toxic orally to
14-day-old rats than young adult
and older adult rats.  The LD5Q
values were 3.4, 3.7, and 4.4
mg/kg, respectively.

Young rabbits were more sensi-
tive than older rabbits.  Young
rabbits excreted more phenol
than older rabbits
Minai
1967      Sc injection
          for 5  days
                       Rabbits       Rats  are  more  sensitive and
                       Hamsters,     hamsters  are least sensitive to
                       Rats,  Mice    benzene poisoning
Source:  Leong, B. K. J., Experimental Benzene Intoxication,  J.  Tox.  Environ.  Health, Supplement, 2,
         (1977), pp.  60.
Sc = subcutaneous.

-------
     The contribution of these chronic toxic effects to development  of
leukemia, defined as a neoplastic condition arising from proliferation  of
white blood cells or precursors in blood and/or in bone marrow,  remains
uncertain (6,50).  (See Table 47).  The changes in leukocyte numbers and
functions after benzene exposure in animals are likely to alter  the  immune
system and may increase their susceptibility to development of leukemia.
However, whether benzene, its metabolite product(s), and/or severe damage to
the bone marrow cells act as the initiator(s) or promotor(s) of  leukemia is
unknown.  While acute myelogenous leukemia and related disorders have been
strongly linked to benzene exposure in humans by a number of epidemiologic
studies (6,36,50), attempts to induce leukemia in experimental animals have
not been very successful (50).  Table 47 presents a representative sample of
such attempts to induce tumors in experimental animals.

BENZENE-INDUCED MUTAGENICITY AND CHROMOSOMAL ABNORMALITIES

     The mutagenicity of benzene has been tested with the Ames/Salmonella
mutagenesis bioassay which detects carcinogens as mutagens (40).  The
specific Salmonella strains employed include TA98 (which detects frame shift
mutagens) and TA100 (which detects base-substitution mutagens).  No mutagenic
activity was detected.  Attempts to increase the sensitivity of the bioassay
by addition of epoxide hydrase inhibitor, use of pre-incubation procedure,
and the addition of S-9 microsome from bone marrow of methylcholanthrene-
treated rats were all unsuccessful in detecting mutagenic activity.

     Atypical cell nuclei, altered DNA metabolism and cell division, and
increased chromosome abnormalities are common cytologic features found in
cultured cells and in animals exposed to benzene (40,50,51).  Wolman (51)
notes that giant nuclei, especially in erythroid precursor cells, have been
found in newts, rabbits, and humans.  Moreover, dividing erythroblasts from
newts treated with benzene exhibit higher incidence of amitotic arrests and
chromosomal anomalies at various stages of mitosis than control animals.
Such chromosomal abnormalities may lead subsequently to unequal nuclear
division, polynucleated cells, and atypical nuclei.

     Apart from these nuclear structural changes, other chromosomal aberra-
tions and altered DNA synthesis have been demonstrated in cultured human
lymphocytes and HeLa cells after benzene treatment in vitro and in cultured
rat and rabbit bone marrow cells after treatment in vivo (40,51) (see Table
48).  For example, bone marrow cells of rats exposed to benzene over a
period of 12 days exhibit a higher incidence of chromosomal breaks and gaps
(51).  Similarly, cultured human leukocytes treated with 1-2 x 10 3M benzene
are shown to have a higher incidence of chromosomal breaks and gaps.  Moreover,
the addition of radiation (100 rad) and benzene (1-2 x 10 3M) to these cells
has a synergistic effect, enhancing the number of chromosomal aberrations;
the additive effect is presumably attributed to benzene-induced inhibition
of DNA repair of radiation-induced chromosome breaks.  At higher benzene
concentration, these cultural cells become more susceptible to inhibition of
DNA synthesis (manifest as a decrease in tritiated thymidine uptake) (40).
The reduction in DNA synthesis in these cells at higher benzene concentra-
tions complicates the interpretation of dose-dependency to incidence of
chromosomal breaks and whether these chromosomal changes should be classified
                                     122

-------
Table 47.   POSSIBLE LEUKEMOGENIC  AND  TUMORIGENIC EFFECTS OF BENZENE
Authors
Ami el
Hamaguti
and Yosida
Hiraki et al.

Kirschbaum
and Strong

Laerum
Lignac
Ward et al.
Year Treatment
1960 Sc injections of
0.001 mL weekly
from age 1 month
until death
1938 0.001 mL weekly
1963 Sc injections

1942 Sc injections,
0.001 mL weekly

1973 Benzene was
painted on skin
twice weekly
for life
1932 0.001 mL benzene
in 0.1 mL olive oil
for 17-21 weeks
1975 Repeated six injec-
tions for 104 weeks
Species
Five
strains
of mice
Mice
Mice

High
leukemia
F strain
mice
Hairless
mice
Mice
C57BL/6N
Comment
No leukemogenic or aplastic
effects were observed
Developed "preleukemic"
state in lymph nodes, spleen,
and liver
Developed subcutaneous
sarcomas
Six of 20 treated mice
developed leukemia; 29 of 212
control also developed leukemia

Frequency of spontaneous
reticulum cells neoplasma was
the same as in unpainted
control
Eight of 33 mice developed
lymphoblastoma
No evidence of carcinogenic
effects
Source:  Leong, B. K. J., Experimental Benzene Intoxication, J.
         (1977), pp. 55.

Sc = subcutaneous.
                                                 Tox.  Environ.  Health,,  Supplement,  2,

-------
               Table 48.  POSSIBLE MUTAGENIC AND TERATOGENIC EFFECTS OF BENZENE
Authors
Dobrokhotov
Year
1972
Treatment
Sc, 200 mg/kg/
Species
Rats
Comment
Chromosomal change

s in bone
Kissling and
Speck
Matsushita
1972
1966
N>       Phillip and
*"       Jensen
Pollini et al.
                     1970
1965
Watanabe et al,
1970
day x 12 days

Sc, 0.2 mL/kg/
day for 18 wks
          Sc, 2 mL/kg
Inoculated with
0.005, 0.003,
0.0015 mL.
Examined after
60, 64, and 68
hours of incuba-
tion

Sc injection, 3
mL/kg on llth to
15th gestation day
                                                              Rabbits
                       Rats
                       Rats
                                                              Chick
                                                              embryo
                                                              Mice
marrow cells

Chromosomal aberrations (breaks
and gaps) were present in bone
marrow cells

Mitotic inhibition in metaphase
stage and chromosomal damage,
probably due to inhibition of
the nucleoprotein metabolism
of the nucleic acid synthesis

Chromatid aberrations of
chromosome of bone marrow at
12 and 24 hours postinjection

All doses inhibited mitosis of
embryo hemopoietic cells
Cleft palate, agnathia, and
micrognathia were observed
more frequently in pregnant
mice injected with benzene on
13th day

-------
as a mutational or toxic event (51).  In addition, the inclusion of both
breaks and gaps as signs of chromosome aberration by many researchers may
inflate the aberration rates since the formation of chromosomal gaps may be
inducible during preparation of the chromosomal sample (51).  Thus, while
the association between benzene exposure and development of chromosomal
abnormalities has been demonstrated, in many animal species, the dose-res-
ponse relationship remains unclear.

PROBLEMS ASSOCIATED WITH EXTRAPOLATION OF ANIMAL DATA TO HUMANS

     The use of similar animal systems to predict the toxicological behavior
of xenobiotics in humans has been the cornerstone of most toxicological
studies; but there are problems when laboratory toxicity data are extrapolated
to man.  A basic question is whether high dose testing with animals (for
both acute and chronic benzene exposure) is a good measure of human health
effects expected from either intermittent acute exposure or chronic low dose
exposure.  The findings that the intervals between intermittent benzene
exposures enable the animal to recover from the toxic effects further compli-
cates the estimation of human risk (52).  Another problem has been the lack
of information quantifying the relationship between benzene dose and incidence
of toxic effects.  Some of the difficulties in assessing the dose-response
relationship have already been discussed in this report.   Robert Dixon, Head
of the Environmental Toxicology Division at NIEHS, has suggested that toxico-
logical testing with animals should meet certain guidelines (53) .   Foremost
among the recommendations is that the maximum tolerable dose should exhibit
no overt toxicity; should not shorten the life span of the test animal(s);
and should not decrease the expected weight gain during the animal's lifetime.
Another is that the test dose should not saturate the pharmacokinetic proces-
ses nor overwhelm the animal's adaptive mechanisms.   Whether many of the
experimental studies investigating benzene toxicity in animals fulfill the
guidelines is uncertain.  Thus, extrapolation of laboratory benzene toxicity
data to humans must be interpreted with caution.
                                    125

-------
                                 REFERENCES


 1.   E.  V.  Anderson,  Chem.  Eng.  News,  54 (19),  34 (1976).

 2.   Anon., Chem.  Eng.  News,  54 (47),  17 (1976).

 3.   P.  H.  Howard  and P.  R.  Durkin,  Sources of  Contamination,  Ambient  Levels,
     and Fate of Benzene  in the Environment,  EPA  560/5-75-005, Dec.  1974.

 4.   SRI, Chemical Origins  and Markets,  Stanford  Research  Institute, Menlo
     Park,  CA, 1967,  p.  18-19.

 5.   S.  J.  Mara and S.  S. Lee, Human Exposures  to Atmospheric  Benzene, EPA
     Contract 68-01-4314, Final Report,  October 1977-

 6.   Assessment of Health Effects of Benzene  Germane  to  Low-Level Exposure,
     EPA 600/1-78-061,  U. S.  Environmental Protection Agency,  Washington,
     D.C. 1978, 120 pp.

 7.   J.  J.  Kocsis  and R.  Songder, "Current Concepts of Benzene Toxicity,"
     CRC Crit. Rev. Toxicol., 3, 265 (1975).

 8.   A.  Sato, T. Nakajima,  Y. Fujiwara,  and N.  Murayama, "Kinetic Studies on
     Sex Differences  in Susceptibility to Chronic Benzene  Intoxication-With
     Special Reference to Body Fat Content",  Brit. J.  of Ind.  Med.,  32,  321-
     328 (1975).

 9.   E.  D.  Pellizzari,  Quantification  of Benzene  in 150  Ambient Air  Samples,
     Final  Report, EPA P. 0.  No. DA-7-43205,  August 1977-

10.   M.  G.  Erskine, "Benzene," Chem. Econ. Handbook,  Stanford  Research
     Institute, Menlo Park,  CA,  618.5021A-618.5024G (1972).

11.   Ambient Concentrations  of Benzene in the Houston Area,  Air Quality
     Surveillance  Group Region 7, Texas  Air Control Board, March  (1977).

12.   "Criteria for a  Recommended Standard...Occupational Exposure to Benzene",
     NIOSH/74-137  (1974).

13.   G.  S.  Parkinson, Benzene in Motor Gasoline—An Investigation into
     Possible Health  Hazards  In and Around Filling Stations  And In Normal
     Transport Operations,  Ann.  Occup. Hyg.,  14,  145  (1971).
                                    126

-------
14.   Battelle-Columbus Laboratories,  Letter to Richard J. Johnson (Office of
     Air Quality Planning and Standards,  EPA,  Research Triangle Park), from
     C.  W.  Townley,  Concerning "Results of Self-Service Exposure Samples,"
     May (1977).

15.   R.  J.  Newsome,  V. Normal, and C. H.  Keith, "Vapor Phase Analysis of
     Tobacco Smoke," Tobacco Sci., 9. 102-110  (1965).

16.   M.  Gordon and P. C.  Goodley,  "Isolation and Characterization of Indus-
     trial Organic Pollutants in Water,"  Amer. Chem. Soc., Div. Water, Air
     Waste Chem., Gen. Pap., 11(1), 91 (1971).

17.   U.S. Environmental Protection Agency, "Industrial Pollution of the
     Lower Mississippi River in Louisiana," Region VI, Dallas, Texas, Sur-
     veillance and Analysis Division (1972).

18.   J.  J. Friloux,  "Petrochemical Wastes as a Water Pollution Problem in
     the Lower Mississippi River," Paper submitted to the Senate Subcommittee
     on Air and Water Pollution, April 5  (1971).

19.   J.  Novak, J. Zluticky, V. Kubelka, and J. Mostecky, "Analysis of Organic
     Constituents Present in Drinking Water,"  J. Chromatogr. 7J)(1),  45
     (1973).

20.   I.  H. Williams, "Gas Chromatographic Techniques for the Identification
     of Low Concentrations of Atmospheric Pollutants," Anal. Chem.,  37, 1723
     (1965).

21.   J.  C. Smoyer, D. E.  Shaffer,  and I.  L. DeWitt, "Program to Sample and
     Analyze Air Pollution in the Vicinity of  a Chemical Reclamation Plant,"
     Inst. Environ.  Sci., Tech. Meet., Proc.,  1_7, 339 (1971).

22.   R.  E. Neligan,  M. J. Leonard, and R. J. Bryan, "The Gas Chromatographic
     Determination of Aromatic Hydrocarbons in the Atmosphere," ACS Nat.
     Meeting, September 12-17, Atlantic City,  NJ (1965).

23.   A.  P. Altschuller and T. A. Bellar,  "Gas  Chromatographic Analysis of
     Hydrocarbons in the Los Angeles Atmosphere," J. Air Pollution Control
     Assoc. , 13(2),  81 (1963).

24.   W.  A. Lonneman, T. A. Bellar, and A. P. Altschuller, "Aromatic Hydrocar-
     bon in the Atmosphere of the Los Angeles  Basin," Environ. Sci  Technol
     2(11), 1017 (1968).

25.   K.  Grob and G.  Grob, "Gas-liquid Chromatographic-Mass Spectrometric
     Investigation of C6-C20 Organic Compounds in an Urban Atmosphere.
     Application of Ultratrace Analysis on Capillary Columns," J. Chromatogr.,
     o *-_ \ J-) > -L v -L-* ' J- J •

26.   E.  R. Stephens, "Hydrocarbons in Polluted Air," Summary Report  Coordinat-
     ing Research Council Project CAPA-5-68, Statewide Air Pollution Research
     Center, Univ. of CA., Riverside, U.S. Nat. Tech. Inform  Serv   PB
     Rept. 230-993 (1973).

                                     127

-------
 27.   S.  Pilar and W.  F.  Graydon,  "Benzene and Toluene Distribution in Toronto
      Atmosphere," Environ.  Sci.  Technol., 7^(7),  628 (1973).

 28.   Preliminary Assessment of Suspected Carcinogens in Drinking Water,
      Appendices, EPA 560/14-75-003A (June 1975).

.29.   Preliminary Assessment of Suspected Carcinogens in Drinking Water,  EPA
      560/4-75-003 (June  1975).

 30.   A.  F.  Fentiman,  M.  B.  Neher,  G.  W.  Kinzer,  P.  R.  Sticksel,  R.  W.
      Coutant, G. A.  Jungclaus,  N.  A.  Edie,  J.  McNutty,  and C.  W.  Townley,
      Environmental Monitoring-Benzene,  EPA-560/6-79-006,  U.S.  Environmental
      Protection Agency,  Washington, DC,  1979,  pp.  110.

 31.   National Cancer Institute,  On Occurrence, Metabolism,  and Toxicity
      Including Reported  Carcinogenicity of  Benzene, Summary Report, Washing-
      ton, DC (1977).

 32.   E.  D.  Pellizzari, M.  D.  Erickson,  and  R.  A.  Zweidinger.   Formulation
      of  a Preliminary Assessment of Halogenated  Organic Compounds in  Man
      and Environmental Media, EPA 560/13-79-006,  U.S.  Environmental
      Protection Agency,  Washington, DC.

 33.   E.  D.  Pellizzari, letter to Joseph Breen, Office of Toxic Substances,
      U.S. Environmental  Protection Agency,  Concerning the Pilot  Study at
      Love Canal under EPA Contract No.  68-01-4731,  September 1978.

 34.   A.  Sato, T. Nakajima,  and Y.  Fujiwawa, Determination of Benzene  and
      Toluene in Blood By Means of a Syringe-Equilibration Method Using a
      Small  Amount of Blood, Brit.  J.  of Ind. Med.,  32,  210-214 (1975).

 35.   B.  D.  Goldstein, Benzene Toxicity.   A  Critical Review.   Introduction,
      J.  Tox. Environ. Health, Supplement 2, 1-4  (1977).

 36.   B.  D.  Goldstein, Hematotoxicity in Humans,  J.  Tox. Environ.  Health,
      Supplement 2, 69-106 (1977).

 37.   H.  S.  Cohen, M.  L.  Freedman,  and B. D. Goldstein,  "The Problem of
      Benzene in Our  Environment:   Clinical  and Molecular Considerations,"
      Am. J. Med. Sci., 275, 125-138 (1978).

 38.   D.  J.  Kielian,  and  Daniel,  R. C.  "A Cytogenetic Study of Workers
      Exposed to Benzene  in the Texas Division of Dow Chemicals," USA, February
      28, 1978.

 39.   G.  M.  Rusch, B.  K.  J.  Leong,  and S. Laskin, "Benzene Metabolism," J.
      Tox. Environ. Health.  Supplement 2, 23-36 (1977).

 40.   G.  M.  Dean, "Genetic Toxicology of Benzene, Toluene, Xylenes, and
      Phenols," Mut.  Res. 47,  75 (1978).
                                     128

-------
41.   C.  G.  Hunter,  Aromatic Solvents,  Ann.  Occup. Hyg., 9, 191 (1966).

42.   D.  V.  Parke and R.  T.  Williams,  "Studies in Detoxicification 49.  The
     Metabolism of  Benzene  Containing  14C Benzene," Biochem. J., 54, 231,
     (1953a).

43.   C.  G.  Hunter and D.  Blair,  "Benzene.  Pharmacokinetic Studies in Man,"
     Ann.  Occup. Hyg., 15,  2-4,  193-201.

44.   H.  S.  Posner,  C. Mitoma,  S.  Udenfriend,  "Enzymatic Hydroxylation of
     Aromatic Compounds.   II.   Further Studies of the Properties of the
     Microsomal Hydroxylating  System," Arch.  Biochem. Biophys., 94, 269-279
     (1961).

45.   H.  H.  Cornish and R. R. Ryan,  "Metabolism of Benzene in Nonfasted,
     Fasted, and Arylhydroxylate Inhibited Rats," Toxicol. Appl. Toxicol.,
     7,  767-771 (1965).

46.   D.  Gerina, J.  W. Daly, B. Witkop, P. Zaltzman-Nirenberg, and S. Uden-
     friend, "Role of Arene-Oxide-Oxepin System in the Metabolism of Aromatic
     Substrates.  I.  In Vitro Conversion of Benzene Oxide to a Permercapturic
     Acid and a Dihydrodiol, Arch.  Biochem. Biophys., 128, 176-183 (1968).

47.   R.  Snyder, F.  Uzuki, L. Gonasun,  E.  Bromfeld, and A. Wells, "The Metabo-
     lism of Benzene In Vitro," Toxicol.  Appl. Pharmacol., 2, 346-360 (1967).

48.   S.  Undenfriend, "Arene Oxide Intermediates in Enzymatic Hydroxylation
     and Significance to Drug  Toxicity,"  Ann. N.  Y. Acad. Sci., 179, 295
     (1971).

49.  M.  L.  Freedman, "The Molecular Site of Benzene Toxicity," J. Tox.
     Environ. Health Supplement, 2, 37-44 (1977).

50.  B.  K.  J. Leong, "Experimental Benzene Intoxication," J. Tox. Environ.
     Health Supplement,  2,  45-62 (1977).

51.  S.  R.  Wolan, "Cytologic and Cytogenetic Effects of Benzene," J. Tox.
     Environ. Health Supplement, 2, 63-68 (1977).

52.   D.  L.  Coffin,  D. E.  Gardner, G.  I. Sidorenko, and M. A. Pirrigin, "Role
     of Time as a Factor in the Toxicity of Chemical Compounds in Intermittent
     and Continuous Exposures.  Part II.   Effects of Intermittent Exposure,"
     J.  Tox. Environ. Health,  3, 821-828 (1977).

53.   R.  L.  Dixon, "Extrapolation of Laboratory Toxicity Data to Man:  Factors
     Influencing the Dose-Response Relationship," Fed. Proc., 39, 53 (1980).
                                    129

-------
                             APPENDIX A




SAMPLING AND ANALYTICAL PROTOCOLS FOR BENZENE IN AIR, WATER, BLOOD




                         BREATH, AND URINE
                                130

-------
             SAMPLING AND ANALYSIS OF BENZENE IN BREATH AND AIR
PRINCIPLE OF METHOD

     Recovery of volatile organics from Tenax GC is accomplished by thermal
desorption and purging with helium into a liquid-nitrogen-cooled nickel
capillary trap (1-3).   The vapors are then introduced into a high resolution
glass capillary gas chromatographic column where the constituents are
separated from each other (2,4).  Quantification of the benzene in the
sample is accomplished by gas chromatography/flame ionization detection
(GC/FID).  Confirmation of a portion of the samples selected on the basis of
availability of replicates and dispersal throughout the set is made by gas
chromatography/mass spectrometry (GC/MS).  The thermal desorption inlet-
manifold used with both GC/FID analysis and GC/MS analysis is shown in
Figure A-l.  The overall analytical system for the GC/MS analysis is shown
in Figure A-2.

RANGE AND SENSITIVITY

     The linear range for this analysis is defined on the lower extreme by
the background of the Tenax cartridges.  The upper limit is defined by the
capacity of the capillary column, a somewhat variable factor depending upon
individual column characteristics.  Typically, support coated open tubular
(SCOT) columns have a wider dynamic range, about 104, but lower resolution.
The wide bore (0.55 mm i.d.) wall-coated open tubular (WCOT) columns have
about a  103 linear dynamic range.

     The detection limit for ambient air is determined by the Tenax cartridge
background, which is a function of the original cleanup of the cartridge and
its storage time and conditions.  Freshly cleaned cartridges typically have
<20 ng each of benzene (0.6 |Jg/m3 for a 30-L air sample).  The control
samples  from the two study sites contained 50 ± 40 ng (1.7 (Jg/m3 for 30-L
air sample) showing some storage effect.  Breath samples have an additional
background due to the benzene in the breathing air used in the experiment.
Depending upon the air purity, sampling conditions, and storage background,
values of 0.3 to 4.2 |Jg/m3 may be obtained.

INTERFERENCES

     Extremely high levels of other hydrocarbons can degrade the separation
of benzene from other constituents with GC/FID.  This is not a problem with
GC/MS where specific ions can be monitored.

PRECISION AND ACCURACY

     Replication of standard cartridges is generally ±10%.  Air control
samples  that have been subjected to the rigors of transportation and storage
show recoveries averaging 75% with a standard deviation of 19% (26% relative
standard deviation).  Breath controls averaged 82 ± 22% recovery (27% relative
standard deviation).
                                    131

-------
                                                              COMPRESSION SPRING
                                                                   TEMPERATURE
                                                                   CONTROLLER
ro
                                                                   HEATING CARTRIDGE

                                                               IS'    ^.CARRIER GAS
                              ALUMINUM
                                HEATING
                                   BATH
                                  SIX -PORT
                               TWO POSITION
                                  VALVE
                                                                          TO GLC CAPILLARY
                                                           HEATING AND COOLING BATH

                                                           Nl CAPILLARY TRAP
   VALVE POSITION A
   (SAMPLE DESORPTION)
    VALVE  POSITION B
    (SAMPLE INJECTION)
CARRIER     f  "\
  GAS  »	-f-p  CM,—»>TO GLC
                                                                                         PURGE
                                                                                          GAS
                                                                                                                VENT
                                                                                        I
                                                                                        I	
                                         Figure  A-l.   Thermal  desorption  inlet-manifold.

-------
                                                        PURGE
                                                         GAS
                GLASS
                 JET
              SEPARATOR
  TWO
POSITION
 VALVE
                                                               THERMAL
                                                               DESORPTION
                                                               CHAMBER
                CARRIER
                 GAS
CAPILLARY
GAS
CHROMATOGRAPH


. (
A
CARRIER
GAS
i

§
"N
rAPn i AP
HEATED
BLOCKS
                                                                EXHAUST
                                                       TRAP
                        ANALYTICAL   SYSTEM
Figure A-2.   Analytical  system for analysis of organic vapors
              in ambient  air.
                              133

-------
APPARATUS AND REAGENTS

Collection of Samples

Personal Monitor Pump—
     A personal monitor pump  (MSA Co. -- Model C-200) is used  for  air  sample
collection.  Flow rates are adjusted to M).05 L/min  for. an 8-hr  collection
period.  Flows are adjusted such that a total volume of "^0.024 m  air  is
sampled for a given collection period.

Spirometer—(See Appendix B.)

Sampling Cartridges--
     The sampling tubes are prepared by packing a  10-cm long by  1.5-cm
i.d. glass tube containing 8  cm of 35/60 mesh Tenax  GC with glass  wool in
the ends to provide support (2,9).  Virgin Tenax  (or material  to be  recycled)
is extracted in a Soxhlet apparatus for a minimum  of 18 hrs each time  with
methanol and n-pentane prior  to preparation  of cartridge samplers  (2,9).
After purification of the Tenax GC sorbent and drying in a vacuum  oven at
120° C for 3 to 5 hrs at 28 in of water, all the  sorbent material  is meshed
to provide a 35/60 particle size range.  Meshing  and all further cartridge
preparation is conducted in a "clean" room.  Cartridge  samplers  are  then
prepared and conditioned at 270° C with helium flow  at 30 mL/min for 30 min.
The conditioned cartridges are transferred to Kimax® (2.5 cm x 150 cm)
culture tubes, immediately sealed using Teflon -lined caps and cooled. This
procedure is performed to avoid recontamination of the  sorbent bed (2,10).

Inlet Manifold

     An inlet manifold for thermally recovering vapors  trapped on  Tenax
 sampling cartridges is used and is shown in  Figure A-l  (1-4).

Gas Chromatograph

     A Varian 3700 gas chromatograph with a  glass  capillary column is  inter-
 faced  to the inlet manifold above.  This analytical  system  is  presented
 schematically in Figure A-l.

Mass Spectrometer/Computer

     A Varian MAT CH-7 mass spectrometer capable  of  a resolution of  2,000
 equipped with single  ion monitoring  capability  is used  in tandem with  a
Varian 1700 gas chromatograph and  interfaced to a Varian 620/L computer
 (Figure A-2).

     A glass jet separator is employed  to interface  the glass  capillary
 column to  the mass spectrometer on the  Varian MAT CH-7  GC/MS/COMP  system.
 The separator is maintained at 240°  C  (2).
                                     134

-------
Reagents and Materials

     All reagents used are analytical reagent grade.  All solvents  are
"distilled in glass" (Burdick & Jackson) or are redistilled before  their
use.

     Solvents -- methanol, pentane, acetone.

     Sorbent -- Tenax GC  (35/60 mesh) is obtained from Alltech Associates.
The same pretested lot is used throughout.

PROCEDURE

Collection of Benzene in Ambient Air

     The volume of air which can be sampled for benzene is limited  by the
breakthrough volume of benzene on the Tenax cartridge.  At normal ambient
temperatures, the sample volume is limited to 30 to 35 L of air.  This
volume  limitation must be observed for both air and breath samples.

Collection of Benzene in Breath

     The method for collection of benzene in breath and the specialized
equipment required is described in Appendix B.

Desiccation of Tenax Cartridges from Breath Samples

     Breath is an especially humid air and as such considerable water accumu-
lates on the cartridges.  Since water frequently interferes with the transfer
of  sample from the cartridge to a gas chromatograph (GC) capillary  trap, a
desiccation step is highly desirable.  To establish that desiccation can be
performed without loss of sample, Tenax cartridges were first loaded with
high humidity air and spiked with benzene.  They were then desiccated over
calcium sulfate  (precleaned in a muffle furnace at 400° C for 1 to  2 hrs.
Thermal desorption of these cartridges indicated recovery after desiccation
was 93.4%.   (Relative standard deviation for 4 determinations was 3.8%.)

Analysis of Samples

     The instrumental conditions for the analysis of benzene on the sorbent
Tenax GC sampling cartridge is shown in Table A-l.  The thermal desorption
chamber and the  six-port  Valco valve are maintained at 250° C and 220°  C,
respectively.  The helium purge gas through the desorption chamber  is adjusted
to  15-20 mL/min.  The nickel capillary trap on the inlet manifold is cooled
with liquid nitrogen.  In a typical thermal desorption cycle, a sampling
cartridge  is placed in the preheated desorption chamber and the helium  gas
is  channeled through  the  cartridge to purge the vapors into the liquid
nitrogen capillary  trap  [the inert activity of the trap has been shown  in  a
previous study  (4,8)].  After the  desorption has been completed, the six-port
valve is rotated and  the  temperature on the capillary loop is rapidly raised
 (greater than 100°/min);  the carrier gas then introduces the vapors onto  the
high resolution  GC  column.  The glass capillary column is temperature-


                                    135

-------
 Table A-l.  OPERATING PARAMETERS FOR THERMAL DESORPTION ,'GC/FID
                       ANALYSIS OF BENZENE
   Parameter
  Setting
Inlet-manifold

     desorption chamber
     valve
     capillary trap - minimum
                    - maximum
     thermal desorption time
     purge rate (He)
GC.
     108 M glass SCOT SE-30
     Nitrogen carrier gas flow rate
     47 M glass WCOT-SE-30/BaC03

     Helium carrier gas flow rate
 250°C
 250°C
-196°C
 250°C
 7 min
 30 mL/min
45°C initial l°/min
4.0 mL/min
50°C initial for 3 min
then 4°/min to 200°C
2.1 mL/min
                               136

-------
programmed as indicated.   After all the components have eluted, the column
is cooled to ambient temperature and the next sample is processed  (2).

Quantitation

     Quantitation of benzene is accomplished by comparing peak areas at the
benzene retention time to peak areas from cartridges loaded with known
amounts of benzene from a permeation system (5,6).

                                     A   / s
                           ,   ,o,    unk/5std
     Benzene concentration C|Jg/m ) - 7	j-	
                                      std  unk

where     A  ,  is the area of the benzene peak in the sample,
          AUn, is the area of the benzene peak in the standard(s),
          gs , is the amount of benzene added to the standard (fjg) ,
          vs ,  is the volume of air or breath collected (m3).
           unk
QUALITY ASSURANCE PROGRAM

Reagent and Glassware Control

     Reagent and glassware control is required to minimize contamination.
Sample containers, glassware, etc., are cleaned with Isoclean§, rinsed with
distilled/deionized water and heat treated at 450°-500° C to insure the
removal of all traces of organic compounds.

Quality Control Samples

     Blank sampling cartridges and control cartridges loaded with known
amounts of benzene are prepared for each sampling trip.  The total number of
cartridges dedicated to blanks and controls is greater than 10% of the
maximum number of field samples to be collected.  A portion of the samples
is designated "lab blanks/controls" and remains in the laboratory; another
portion is designated "field blanks/controls" and is carried to the field in
the same  containers as the sample cartridges.  This procedure not only
provides  a check on possible contamination during transport and storage, but
also allow calculation of overall recoveries during the storage and analysis
phases.

     Blank samples for the breath sampling and analysis are generated in the
field and in the laboratory by pumping 0.07 m3 of air through the spirometer
with the mouthpiece plugged and collecting two parallel blank cartridges.
Controls were generated in the same way except that the cartridges were
spiked with a solution of benzene in methanol before sampling.

REFERENCES

1.   E. D. Pellizzari, Development of Method for Carcinogenic Vapor Analysis
     in Ambient Atmospheres, Publication No. EPA-560/2-74-121, Contract No
     68-02-1228, 148 pp., July 1974.
                                     137

-------
2.    E.  D.  Pellizzari, Development of Analytical Techniques for Measuring
     Ambient Atmospheric Carcinogenic Vapors,  Publication No.  EPA-60Q/2-75-
     075, Contract No. 68-02-1228, 187 pp.,  November 1975.

3.    E.  D.  Pellizzari, J.  E.  Bunch, B. H.  Carpenter, and E. Sawicki,  Environ.
     Sci. Technol., 9, 552 (1975).

4.    E.  D.  Pellizzari, The Measurement of  Carcinogenic Vapors  in Ambient
     Atmospheres.  Publication No. EPA-600-7-77-055, Contract  No.  68-02-
     1228,  288 p. June 1977-

5.    E.  D.  Pellizzari, Analysis of Organic Air Pollutants by Gas Chromatog-
     raphy and Mass Spectroscopy, EPA-600/2-79-057,  243 pp March 1979.

6.    E.  D.  Pellizzari, "Ambient Air Carcinogenic Vapors Improved Sampling
     and Analytical Techniques and Field Studies, EPA-600/2-79-081, 340  pp.
     May 1979.
                                    138

-------
             SAMPLING AND ANALYSIS OF BENZENE IN DRINKING WATER
PRINCIPLE OF THE METHOD

     Two samples of drinking water are collected at each participant household.
The first is the initial water flow and the second is the water collected
after 3 rain of maximum purge of the water lines.  As a screen, water samples
are composited such that benzene present at >1 |Jg/L would be detected in the
composite.  Any elevated values can then be evaluated individually.

SAMPLE COLLECTION

     At each participant household, two samples of cold tapwater are collected
in 120-mL precleaned glass bottles from the kitchen tap or source commonly
used for drinking and/or cooking.  The first sample is taken immediately
upon turning on the water, without flushing; the second sample is taken
after the water has been allowed to run for 3 minutes.  Time is measured
using a stopwatch.

COMPOSITING OF WATER SAMPLES

     Aliquots of two or three water samples are mixed and an aliquot of the
composite analyzed.

SAMPLE ANALYSIS

     The water samples are analyzed by the Guidelines Establishing Test
Procedures for the Analysis of Pollutants, Purgeables-Method (624) [1] using
GC/FID for screening purposes.

QUALITY ASSURANCE PROGRAM

Reagent and Glassware Control

     Reagent and glassware control is required to minimize contamination.
Sample containers, glassware, etc., are cleaned with Isoclean®, rinsed with
distilled-deionized water, and heat treated at 450°-500° C to insure removal
of all traces of organic compounds.

Quality Control Samples

     Blank water and contro'l water spiked with known amounts of benzene are
prepared for each sampling trip.  The total number of cartridges dedicated
to blanks and controls is greater than 10% of the maximum number of field
samples to be collected.  A portion of the samples is designated "lab
blanks/controls" and remains in the laboratory; another portion is designa-
ted "field blanks/controls" and is carried to the field in the same containers
as the samples.  This procedure not only provides a check on possible contami-
nation during transport and storage, but also allows calculation of overall
recoveries during the storage and analysis phases.
                                    139

-------
REFERENCES




1.   Federal Register, No.  233,  Monday,  December 3,  1979,  p.  69532.
                                    140

-------
                  SAMPLING AND ANALYSIS OF BLOOD AND URINE
PRINCIPLE OF THE METHOD

     A blood or urine sample is equilibrated at 37°C with an air space of
determined volume until equilibrium is attained.  The entire headspace is
then purged into a cryogenic trap which can be placed in line with a GC as a
sample loop and heated.  In this manner, the recovery is determined by the
partition between fluid and air and avoids the many artifacts and other
problems introduced by purging (i.e., foaming, precipitation occlusion, and
sorbent background).

RANGE AND SENSITIVITY

     The range is limited by the limit of detection on one extreme and by
the chromatographic capacity of the capillary on the other or ^104.  Minimum
detectable concentration for,the method is estimated to be 1.6 JJg/L (95%
confidence level).

INTERFERENCES

     No interferences have been observed; however, high levels of other
hydrocarbons in the sample could cause the benzene peak to be obscured.

PRECISION AND ACCURACY

     Precision at 500 [Jg/L is 8% relative standard deviation increasing to
33% at 1.8 pg/L.  Recovery of control samples spiked at 5 MS/L was 98%.

APPARATUS AND MATERIALS

1.   A thermostated two-position, six-port valve with nickel capillary trap
     as indicated in Figure A-3.

2.   A gas chromatograph with flame ionization detector.

3.   Thermostated oven (37°C).

4.   Glass capillary GC column, phase SE30.

5.   Glass hypodermic syringes (10 mL) and needles.

6.   Silicone rubber septum material.

7.   Liquid nitrogen.

8.   Ultrapure air (<0.1 ppm total hydrocarbon).
                                    141

-------
                                    VftLVE POSITION A
                                    (SAMPLE PURGE)
                                     VALVE POSITION B
                                     (SAMPLE INJECTION)
                                 CARRIER    X"V
                                   GAS  »   / O  q-X—»-TO GLC
                                 PURGE
                                   GAS
                                                      VENT
                                         (HUIID
                                 U.
Figure A-3.   Six-port,  two-position valve for  the introduction of headspace  samples,

-------
PROCEDURE

Collection of Blood and Urine Sample

     Blood samples are collected from selected participants from a brachial
vein by venipuncture using a 10-mL Venoject tube.  These blood samples are
collected by experienced medical personnel using accepted medical procedures.

     Urine samples are collected by the participants in a precleaned 120-mL
wide-mouth bottle.

Analysis of Samples

     Pre-equilibrate a 10-mL glass syringe at 37°C (>30 min), remove the
needle from the syringe and inject 1.0 mL of blood or urine (sample, standard,
or blank) into the syringe which has been sealed around the plunger with
saturated lithium chloride.  Adjust the volume to 10 mL by filling the
syringe with "ultrapure" air, replace the needle on the syringe, and seal it
by inserting it into a piece of silicone septum material.  Incubate the
entire syringe assembly at 37°C for 20 min.  After the incubation, the
needle is removed and the syringe is connected to the cryogenic trap via an
18-gauge needle.  The total air space in the syringe is purged through the
trap.  An additional 1 mL of air is purged through the trap from another
syringe.  The latter step is to prevent sample holdup in the transfer lines
to the trap.

     At this point, the coolant (liquid nitrogen) is removed from the trap,
the valve rotated, and the trap rapidly heated to 175°C.  The GC operating
parameters are given in Table A-2.

     Calibration of the GC is obtained by analyzing blood spiked with known
amounts of benzene and blood blanks under identical conditions to the sample.

Quantitation

     Peak areas of benzene in unknown samples are compared to calibration
curves generated with known amounts of added benzene.  This results in the
following relationship:

                                        A   /g
      Concentration of benzene (|Jg/L) = -~	—
                                           std

where     A    is the peak area of the sample,
          g .  , is the amount of benzene added to the standard
          A   , is the peak area of the standard.

VARIATIONS IN THE BLOOD ANALYSIS PROTOCOL

     Pretest blood samples were analyzed using a glass GC column 285 x 0.2
cm packed with 2% OV-101 on Gas Chrom Q (100/120 mesh).  The blood samples
from Houston were analyzed on an SP-1000 (0.1%) on Carbopack C (80/100 mesh)


                                    143

-------
 Table A-2.   OPERATING PARAMETERS FOR GC/FID ANALYSIS OF BENZENE
                       IN BLOOD AND URINE
   Parameter
        Value
Column

     47 M glass WCOT SE-30,  BaC03

     Helium carrier gas flow rate
50°C initial for 3 min
then 4%nin to 200°C
2.1 mL/min
                               144

-------
column, 200 x 0.2 cm.   A photoionization detector (HNU, Inc.) was used as a
detector; however, the sensitivity was less than that obtained with the
method described above O20 |Jg/L) .

QUALITY ASSURANCE PROGRAM

Reagent and Glassware Control

     Reagent and glassware control is required to minimize contamination.
Sample containers, glassware, etc., are cleaned with Isoclean®, rinsed with
distilled-deionized water, and heat treated at 450-500° C to insure the
removal of all traces of organic compounds.

Quality Control Samples

     Blank blood was obtained from several individuals who have low benzene
exposure potential and the blood was pooled.   Aliquots were placed in the
same type of vials used for sample blood storage.  Some of these aliquots
were spiked with known quantities of benzene  as controls.   In total the
number of aliquots is 10-20% of the expected  number of blood samples.   A
portion of the samples is designated as "lab  blanks/controls" and remains in
the laboratory; another portion is  designated "field blanks/controls"  and
is carried to the field in the same containers as the sample cartridges.
This procedure not only provides a  check on possible contamination during
transport and storage, but also allows calculation of overall recoveries
during the storage and analysis phases.
                                    145

-------
                     APPENDIX B




SPIROMETER FOR THE DETERMINATION OF BENZENE IN BREATH
                         146

-------
            SPIROMETER FOR THE DETERMINATION OF BENZENE IN BREATH

INTRODUCTION AND PRINCIPLE OF THE METHOD
     Conkle et al.  (1) have evaluated human breath for "normal" levels of
various organics including benzene.   They needed a multistage cryogenic trap
for sample collection which was too  cumbersome for field sampling so the
spirometer described below has been  designed.

     There are several design criteria which must be met.  Due to the low
levels of benzene anticipated, the sample must be accumulated over a period
of time.  Since the subject breathes into the device for a period of time
(up to 15 rain), he/she must not be unduly discomforted while maintaining
sample integrity.  This comfort factor requires minimal back pressure (<2 mm
of Hg) .  A Tenax cartridge used as an accumulator directly coupled to the
exhaust valve exhibited too great a  back pressure at normal respiratory flow
rates  of 7 L/min.  Since transient flow rates may reach 10 times the average,
an alternative design was considered.  A collapsible reservoir such as a
Tedlar bag was used between the subject and the cartridge(s) and the expired
air drawn through the cartridge(s) by a metered pump.  With an adequate
volume in the bag,  matching the pumping rate and the respiratory flow was
relatively simple.   The same low pressure differential had to be maintained
for the inspired air, hence a similar bag arrangement was used to provide
the subject with ultrapure air.  The sampling apparatus is shown schemati-
cally  in Figure B-l.

MATERIALS

1.   Air cylinder-Zero 0.1  Air (Airco, Inc. Cat. No. 331-09926) with two-
     stage regular valve [all metal  seals for low hydrocarbon background
     e.g., Airco Model 18-75 (CGA 590)].

2.   Activated Carbon Filter (28.8 cm i.d. x 16 cm) with 1/2-in fittings.

3-   Air Humidifier-Midget impinger  (Lab Glass LG-6819-122) with I 24/40
     joint and 12/5 ball joints at entrance and exit; 250-mL Erlenmeyer with
     24/40 joint; two 12/5 ground ball sockets (Lab Glass:  LB-1045-110),
     two Size 12 pinch clamp for ball and socket joints (Lab Glass:  LG-
     1045-102).

4.   Bulkhead Quick Connects (2)  [Body Assembly, Viton o-ring, 1/4 in
     stainless steel  (Swagelok:  SS-QC4-B1-400)].

5-   Body and Stand for the Spirometer--A metal box 30 x 45 x 60 cm with a
     door in the top and openings for tubing and four sturdy legs to produce
     an overall height of 90 cm.  The following fixtures are added to support
     the various components Flexframe® foot plates (2), Flexframe® rods (2 -
     12 cm, 1-60 cm), Clamp holders (3), and a three-prong clamp.

6.   Tedlar bags, 50-L capacity (2 minimum).  The dimensions and configura-
     tion in the Tedlar bags are given in Figure B-2.  Fittings necessary
     for each bag, Teflon® nut 7/16 x 20 straight thread (1), o-seal straight
     thread adapter  [Swagelok:  SS-401-A-OR], 1/4 in nut and ferrules,


                                    147

-------
           TO PUMP

           A  db
  TEFLON
CONNECTORS
                                                             ULTRAPURE
                                                             AIR TANK
                                         DOUGLAS
                                       VALVE AND
                                      MOUTHPIECE
       Figure  B-l.   Schematic diagram of the spirometer.
                                148

-------
                           7O CM
  r
i
  2
  U
      TEFLON
      NUT
                                                 /    \    *
                                                       DUCT
Figure B-2.   Specifications  for  50-L  Tedlar bag with  duct.   Opening  to
             fit 2.5-cm pipe,  tapering  for better  fit;  standard  Teflon
             nut for 7/16-2  straight  thread sealed inside.
                               149

-------
     stainless  steel,  1/4-in stainless steel quick connect with shut-off
     [Swagelok:   SS-QC4-D-400].

 7-   Clamp,  pinch type (1).
                                *
 8.   Douglas Mouthpiece assembly  with noseclip as follows:  Tedlar flap
     valves  cut  to replace the  rubber valves supplied, 2 sections of 1 1/4
     in o.d. copper tubing with 90° elbow (see Figure B-3), all joints were
     soldered and the  entire piece nickel-plated.

 9.   Spring  or wire spiral (1)  3 cm diameter and 6-7 cm in length, chrome or
     nickel-plated.  (This device keeps the intake end of Tedlar Bag A from
     collapsing  during inhalations).

10.   Teflon® tubing,  1/4-in o.d. 2-3  M.

11.   Glass "Y",  1/4 in x 8 mm x 8 mm.

12.   Teflon   straight union reducers  (4) (Beckman, 830511).

                                                 ®
13.   Straight unions,  stainless steel with Teflon  ferrules 1/4 in (5)
     [Swagelok:   SS-400-6, T-404-1, T-403-1].

14.   Glass or plastic tubing 8-mm i.d., 2 cm in length (2).

15.   "Y" connector, polypropylene (1) [Fisher:  15-320-10D].

16.   Forceps, 200 mm (2) [Fisher 10-316A].

17.   Cast iron  ring stand, 10 x 15 cm base (1).

18.   Drying  tubes, polyethylene, 152  mm (1 or more).

19.   Rubber  tubing 1/4-in i.d.,  3/32-in wall thickness.
                              (S)
20.   Quick Connect with Teflon  or nylon ferrules [open] (1)  [Swagelok:
     brass or stainless steel - QC4-S-400].

21.   Pump, Nutech Model 220 (Nutech Inc., Durham,  NC) or equivalent (1).

22.   Stopwatch  (1).

23.   Glovebag with clips [I2R:   x-24-17]; (Optional) Helium tank and regula-
     tor; Teflon® tubing to connect tank and glove bag; ring stand and
     clamp.

24.   Binder  clips, large (1 or more)  to close ducts of Tedlar bags.

25.   Drierite (>20% indicating)  (500  g or more).
*
 Sargent Welch Cat.  No.  S-7695 (not currently in stock)
                                    150

-------
26.   Helium tank and regulator with quick-connect body.  Use activated
     charcoal filter in-line.

27.   Stopper ties with paper clips (2) (Fisher, 14-632).

28.   Distilled water.

29.   Tenax cartridges  and storage containers--!.6 x 10 cm glass tubes
     filled with 6 cm of 35/60 mesh Tenax held in place with glass wool
     plugs (see Appendix A for details).

30.   Ethanol and swabs for sterilizing equipment.

PROCEDURE

     Fill the humidifier with 100 mL distilled water.   Evacuate both Tedlar
bags using the pump with the mouthpiece plug in place.  Place cleaned Tenax
cartridges in their fittings between Tedlar Bag B and the pump.

     The valve from the ultrapure air is  opened and Bag A partially inflated
with a plug in the mouthpiece.  The subject is then attached to the apparatus
with the nosepiece in place and allowed to breath,  inhaling from Bag A and
exhaling into Bag B.  When Bag B is partially inflated, the pump is started
at a nominal 7 L/min.   The flow rate is then adjusted to approximate the
subject's breathing rate.  The test is continued until 60 L of breath have
been collected or 15 min pass whichever occurs first.   The subject is then
removed from the apparatus and the plug replaced in the mouthpiece and Bag B
is exhausted through the two parallel cartridges.

     Data concerning the breath collection are recorded.  Subject code
number, time at start  and end, pump meter readings  beginning and end, and
ambient temperature are all recorded to permit quantitation later.

     See Appendix A for the Tenax cartridge analysis.

Reference

1.   J. P. Conkle, B.  J. Camp, and B. E.  Welch, Trace Composition of Human
     Respiratory Gas,  Arch Environ. Health, 30, 290-95 (1975).
                                    151

-------
                APPENDIX C




SEQUENTIAL WITHOUT REPLACEMENT SAMPLING FOR




    ATTRIBUTES AND SUBPOPULATION MEANS
                    152

-------
      SEQUENTIAL WITHOUT REPLACEMENT SAMPLING FOR ATTRIBUTES



                      AND SUBPOPULATION MEANS




     Consider a finite population with N units u(i); i = 1(1)N.  We will



denote this universe by U = (u(i); i = 1(1)N}.  This population  contains



some number M of units belonging to a particular subpopulation or domain of



interest D, where M is unknown.   Let PQ = M/N denote the unknown popula-



tion proportion belonging to domain D.  If Y(i) denotes a variate value of



interest associated with the i-th population member, then we define



          Yn(+) =  I  Y(i)                             (1)

           U      i£D


and



          YD(-) = YD(+)/M                              (2)




Now, we examine a without replacement sampling scheme analogous  to the



familiar with replacement method that leads to the negative binomal distri-



bution.  We state the following theorem:



     Theorem 1.  If units are selected with equal probabilities  and without



replacement from a finite universe of size N until m members of  a particular



domain D have been selected, then n (the number of draws required to yield



m members of D) has the following distribution:
                                                   r = m,...,N-M+m
                          -^f  -y-t. in/   y^i. ^J




which can also be written




                        /NPn-l\/rNQn>/^N-l'

                                            }  for r = m,...,N-M=M





where QD = (1-PD).   Notice that PD can take the values 0,(1/N),  (2/N).  .  .,




(K/N),.  .  .1.  This distribution will be referred to as the  Negative




Hypergeometric distribution.
                              153

-------
     Proof.  For m members of D to have been selected after the n-th

draw requires that m-1 had been selected after the  (n-l)-th, draw, the

final member of D being selected on draw n.  Therefore,
Recall the factorial equation for combinations is of the form:

           C    n!
          n r =
                (n-r)!r!

and write
                       M! _  (N-M) ! _  M-m+1 "\
     Pr(n = r/m) =   (M-m+1) » (m-1) 1   (N-M-r+m) ! (r-m) !   N-r+U

                       M!  (M-m+1)
                     (M-m+1)!(m-1)!
                           N!  (N-r+1)
                      M!
     (N-r+1)!(r-l)!


                   /N-M\
                   \r-m/
                     (M-m)!(m-1)!
                             N!
/M-l\  f$-tt
\m-lj  \r-m
                  M
which is the result stated in Theorem 1.

     The'following corollary is an immediate consequence of Theroem 1.

     Corollary 1.  If X = n-m is number of units in the sample of

Theorem 1 that do not belong to D, then
                              154

-------
          Pr(X =  x)  = Pn
                      \J  \ILL- J-t  X " f I   \

                                         for x = 0,1,.  .  .,N-M


implying that

                         -v        ,  /I, ,\ -1
         N  )(M-l)!
         M  \(M-l-m+1)!(m-1)! |

                       1-1
               HM-m) !m!

          N  /M\ "I
          m  | m J


where M!, for example,  is M factorial  or M(M-l) (M-2) .  .  .  1



     Applying the results of Theorem 1 or  more specifically, its corollary,

we can show  that  (m-l)/(n-l)  is an unbiased estimate of P_.  Notice, that
Now,
      EUn-1)'1}  =
EUm+x-1)-1}  =  P,    '»-*   *   (^  I          f ^
                            /M-l\  ^  /N-M
                            \m-V  x=o  \x
                    -   PD    /M-l\  N;M   /N-M\// N-2
                        N-I   v--v  xi0   \x y  \m+x-
Letting  W= N-l, M = M-l and m = m-1, we  see  from corollary 1 that the sum

above is
                    .   N-l    M-l
                                    155

-------
where
                       m-1
           yn(m-l)  =   Z
            U          k=l
is the simple average of the variate values associated with the first (m-1)

units drawn that belong to domain D.  Notice, that if the m-th such unit is

selected at the n-th draw, then the (m-1) units selected up to and including

the (n-l)-th draw constitute a simple random sample from domain D of unknown

size M.  Therefore,  given that the m-th member of D is selected on draw n,

we have


      E{yD(m-l)| n}  =  YD(-)

and

                                     ,2
    Var{yD(m-l) | n}  =   ^ - -£-   Sj,
                             - JL-)
Therefore ,
                        NPD yD(m-l)  | n=r}  =  NPD YD(-)   and
                        «•?(*-+)<
Having shown  that P_  is  an  unbiased  estimate  of PD>  it  is  clear that Y^(+)

is an unbiased estimator for YD(+).  The variance of YD(+) is
     Var{Yn(+)}  =  N* Y?(O  Var{Pj + N2 (-^r - = 1 sj  E{Pn}           (5)
          Ij             D          1J       \ ID" X   H /  U     U



The variance expression in (5) can be recast as follows:
                                  156

-------
    Var{YD(+)}  =
Given that the m-th  sample member from domain D is selected on the n-th draw,


we see that


  E{y2(m-l) In}   =  Var{y_(m-l) | n}  +  E2{yD(m-l) \ n}
Also, with
        s2(m-l)  =   I  [Y(k)  - yn(m-l)]2/(m-2)
         "          k=l
                 =   SSD(m-l)/(m-2)

denoting the standard mean square among the first (m-1) variates associated

with domain D, we see that


  E{s2(m-l) | n)  =  S2


This allows us to write


     Var{YD(+)}  =  N2  |E{var(PD)}  E{y2(m-l) | n}



                 t  ) ^ .  _!|_ (  .t^,,, „•


Finally, noting that
                  D }   -       (m-2)       f    (n-1) 1
                  N )   ~     (n-l)(n-2)    [      N~ J
                                 157

-------
 we see that

                                                   /- n<>    SSn(m-l)
 Recalling  that
  we find that


                                                          SS_(m-l)
                  •>       /-«.i^    I          -9             _
  var{YD(+)}   =  N2  [1 -
  The quantity in curly brackets above can be expanded, noting that


   SSn(m-l)/(n-l)  =   Z  z2(i)/(n-l)  -  PD fifa-l)
     u
where      z(i)  =  ID(i) Y(i)

with

                    j 1  if u(i) 6 D

                    i 0  o'therwise  .

Letting

                    n-l
        zD(n-l)  =   2   z(i)/(n-l)  =  P- yD(m-l)
                    isl
and define
              2     n"1                  2
              ^  =   Z   [z(i) - zD(n-l)]2/(n-2)  ,
                                  158

-------
we can write var{¥..(+)} as


      var{YD(+)  =  N z(n-l)}  =  N2 [ ^y - J ]  s* .                   (6)
Again, we  notice  that the form of var{YD(+)}  is  identical to the form of

the unbiased variance estimator for a domain  total  when a simple random

sample of  fixed size  (n-1) is drawn without replacement.

     To extend  these  results to a stratified sequential scheme,  we define


           Y_(h+) =    I.   Y  (hi)
           D        i£D(h) D


as the population total for domain D in stratum h = 1(1)H.  The corresponding

stratum average  is  denoted by YD(h«) = Y_(h+)/M(h).  We are  interested  in

estimating the population average


                      H            H
             Y   =    X  Y.(h+) /  I  M(h)
                    h=l  U       h=l
 for domain D.  We propose the ratio estimator


             2       H                     H
             YD  =   Z  N(h) PD(h) yD(h) / I  N(h) PD(h)                   (7)
                    h=l                   h=l


 where

     N(h)      is the population size of stratum  (h)

     A

     PD(h)  =  fm(h) - ll/[n(h) - 1] is the stratum (h) estimate  for
               *DUJ = Mth^/NCh) the proportion of domain D members in
               stratum (h) .

                                  159

-------
     yQ(h)     is the simple average of the [m(h) - 1] variate values
               associated with the first [m(h) - 1] members of domain
               D(h) selected for the sample.

                                  A
     To estimate the  variance of YD in  (7),  we will use the approximate

variance formula for a ratio
         Var{YD  =
                                      YDVar{M(+)}
                    -2 YD Cov{YD(-H-);
to suggest the estimator

                                       o     A
           {§D}  =   [var{YD(++)}  •«•  YD var{M(+)}
                    -2 Y  cov{YD0~O; M(+)}     M (+) .                  (8)
The covariance between
                    H
         YDOHO  =  Z  N(h) PD(h) yD(h)
                   h=l
•and
                    H
                  = Z N(h)  Pn(h)
                    h=l
 can be written as
     ~                  H   ,
 Cov{YD(-H-');  M(+)}  =   Z  IT(h) Cov{PD(h)yD(h);
                       h=l
 where
                                                                            (9)
Cov{PD(h)yD(h); PD(h)}  =  YD(h) Var{PD(h)}
                                   160

-------
 where Yn(h) = YD(h+)/M(h) is the domain D(h) average.  The  obvious  estimator

 for the covariance expression in (9) is


      cov{PD(h)yD(h); PD(h)}  =  9D(h) var{PD(h)}   .                       (10)


 Using the covariance estimator in (10), we can expand the variance  estimator

 in (8) to
         |r                             iy

{YDJ   =  I  N^h) |var[PD(h)yD(h)] + YD var[PD(h)J
         var
                  "  2 YI


 Defining


                          1   if sampling unit u(hi) e D(h)

                          0   otherwise

 and let

          2L(hi)  =  ID(hi) Y(hi)



then, we have shown that
          2D(h)  =   Z   zD(hi) / [n(h) -
and
           2
          82(h)  =       ^(hi) ' ^(h))2 / ln(h) - 2]
                                   161

-------
can be used to calculate
    \

var{Pn(h)y_(h)}  =
     DD
It is also clear that
         !D(hO  =  PD(h)
and
                 =   Z   IID(hi) - ID(h-)]2 / [n(h) - 2]
leading to
  It is also possible to show that
       cov{D(h)yD(h);
  where
                   =   Z   [ZChi) - 5(h)] [I(hi) - I(h-)] / [n(h) - 2]
                                      162

-------
Finally, we have shown that
        var{YD}  =
                              - 2  „
 If we  let
                                          A

         tjj(hi)  =  N(h)  ID(hi)  [Y(hi)  -  YD]


 or
                                          A      »^

         tpChi)  =  W(h)  ID(hi)  [Y(hi)  -  YD]  / PD


 where


            W(h)  =
              PD  =
 then
 If [n(h)  -  1]  « N(h)  for  all h =  1(1)H,  then equation (11) is approximately


                      H
        var{Y  }  =   I  s2(h)/[n(h)  -  1]   .                               (12)
             U      h=l  C



      It is  intriguing to notice that by discarding the n(h)-th unit  in the

 -sequential  draw from stratum  (h)  we  can  treate the previous  [n(h)  -  1]

 units as a_ simple  random  sample  of fixed size  drawn without  replacement.

 Faced with the results  outlined  in the previous paragraphs,  one  would be
                                   163

-------
tempted to conjecture that given the value of n(h), these n(h) - 1  observa-


tions are, in fact, equivalent to a SRS  of fixed size.  This contention can


be easily disproved by using the 'Negative-Hypergeometric1 distribution for


n(h) in Theorem 1 to show for a particular example (say  N = 5, M = 3 and


m =2) that




     Var
                             \
                             L
             N  l   P (1 - P )
                   ^rkVA   *nJ
(m-1)   Nl  VN-1/   "Dvi   *D
                              (M-m+1) M(N-M)

                              (m-1) N2 (N-l)
Or
                   9              9

              (m-D ]   -j    /M\   ["(M-m-H)  (N-M)      ]               (13)


              (n^l?)        \*'   I (m"1)  M(N-1)       J  '
 For the example mentioned above with N=5,  M=3, m=2
                 =  (m-1)2  E {(m+x-1)"2}
                         9     /M  i\ ^~^   /\s u
                    *•»*'»  C-!)±   (N;M
                 -   (»  2 Jo  0 A + 1)
                     (f) I1/4 + 2/4x6 + 1/9x4}
                                   164

-------
                •              •  13/3°
The right-hand side  of  (13)  is





    (!)'  [Hi  *   ']    •
                               If
This small  counter-example is sufficient to  show  that  the first n(h)-l




units do  not  constitute a SRS of fixed size  conditional on n(h).  This




is one example  (counter-example) used to demonstrate that the technique




of n associated with m+1 is needed to obtain unbiased  estimates for the




m values  obtained from this selection process.
                                  165

-------
                       APPENDIX D




DATA COLLECTION INSTRUMENTS FOR THE HOUSEHOLD SURVEY AND




                MATERIAL SUBMITTED TO OMB
                          166

-------
                                                                                              OMB No. 1S8-S78Q1Q
                                                                                     Approval Expires September 1980
                                RESEARCH TRIANGLE INSTITUTE
                              STUDY OF BENZENE BODY-BURDEN

                              PARTICIPANT CONSENT FORM
                                    (HIGH EXPOSURE AREA)


  I understand that the Research Triangle Institute is engaged in a study of the exposure and absorption of benzene
by persons living in areas having various levels of benzene in the environment. I understand that the survey is being
conducted in order to help measure the levels of exposure and absorption of benzene in populations environmentally
exposed to benzene, and is limited to the purpose stated. I further understand that the survey is being conducted
under the auspices of the United States Environmental Protection  Agency in cooperation with the

Texas and City of Houston  Health Departments.	


  I do hereby freely consent to participate in this study of benzene exposure and absorption and understand that my
participation will consist of providing answers to a questionnaire related to environmental exposure and the follow-
ing environmental and biological samples: (II  two four ounce samples of cold tap water from a source commonly
used for drinking and cooking, (2) a sample of environmental exposure collected by a small device which I will keep
with me for a short time. (31  a small (approximately  10 cc) blood sample to  be taken  from an arm vein, and (4) a
breath sample. I understand that an agent of the Research Triangle Institute will administer the questionnaire in my
home and at the same time collect the tap water samples, instruct me regarding the exposure monitoring  device,
and make arrangements regarding collection of the breath and blood samples. I understand that after the  collection
of the breath and blood samples I will receive an incentive of ten dollars for my full  participation in the study. I
understand that a small number of households  and individuals will be selected for the collection of  duplicate tap
water and blood samples and reinterview, but that such selection would not entitle me to further compensation.


   I understand that my name will not be voluntarily disclosed, and that my name will not be referred to in any way
when compiling and evaluating the results of the study. I understand that participation in this study may result in no
direct benefits to me, other than those described herein, and that I am free to withdraw from this study at any time.
It has been explained to me that there are no significant risks to me from participation in this study. I further under-
stand that while participating in the study I will be free to ask any questions concerning the study; if I  have any fur-
ther questions about the project. I know that I am free to contact

Dr.  Richard  K. Donelson, Texas  Department  of Health  tfliephone number  512-458-7328  0-


Dr.  Robert  A. MacLean,  City  of  Houston  Health Dept.                   713-222-4295
	v     telephone number 	
 or Mr. Benjamin S. H.  Harris, III, Survey Operations Center,  Research Triangle Institute, Research Triangle Park,
 North Carolina 27709, telephone number 919-541-6055.

Daw:    I   I   I - |   I  I  -  |  j   |         Panicio«nf» Name: 	
        IMomhl     IDayl     lYarl                                        (Printl



Site Number:   |	|     Segment Number: |  |    |    Houiehold Number:   |   |   |   |    Participant Number:   |   |   |
 SIGNATURES:

 Participant:	_	_	__  WjtneM:
                                                                           iterviewer Number:   |  |    |   [
                                                  167

-------
                                                                                                            OMB No. 158-S78010
                                                                                                  Approval Expires September 1980
                                    RESEARCH TRIANGLE INSTITUTE

                                      STUDY OF BENZENE BODY-BURDEN
   NOTICE: The information recorded on this questionnaire will be held in nrict confidence, and will be u*ed toMy for reeureh into
   the effects of environmental fectora on public heelth. All results will be summarized for group! of people; no information about
   individual persem will be released without the consent of the individual. This questionnaire it authorized by law (P.L.94-469). While
   you are not required to respond, your cooperation is needed to make the results of the survey eomprehensiva. accurate, and timely.
1.  Site number:
D
HOUSEHOLD SCREENING QUESTIONNAIRE

   2.  Segment number:
                                                                        3.  Household/housing unit number:
4.  Interviewer number:
                                               S.  Dan:
                                                          (Mondil     IDtyl
6.  e.  What is the exact address of this housing unit?
              IScrnt Numotr md Nuntl
                                                                           tAptrmmt Numbtrt
              (Civ)                                ISaal


   b. Is this an eligible housing unit?     jjj Yes (Go to Quanta! 7)    |» [ No (Go to Qumtion 6el

   E. If no. indicate reason and STOP I   | 1 [Vacant    | 2 | Nonexistent   | 3  \ Business   |  * [ Group quarars

                                    |*| Vacation quarters       [ * | Other ISptcifyt  -

7. •. Do you have a responsible screening respondent?    | 1 | Yes (Go to Qutaioa 7bl

                                                  I*] No  (Go to Quasi/an 7el
   b. If vet, indicate whether respondent is  [ * | Household respondent  [ * | Neighbor respondent

   c. If no, indicate reason and STOP I     [JjJ Refusal      [  a | No response after | _ | calls

                                      \3\ Other (Spfeifyl
                                                                                      (Zip Cod*)
8.  a.  Do you have a telephone?   |  ' | Yes (Go to Outaion 8bl       [_fJNo (Go to QiMttion tel

   b.  If yes, whet is the number?  [  1 | Unlisted

                              Pi Refused
   c.  If no, what is the number of the neerest telephone?
                                                    lArn Codfl
9.  How many persons reside in this household?
                                                                             (Go to Quation SI
                                                                                                     Do not know
                                                      168

-------
10.  For each person in your household, including yourself, please indicate:
    a.   Age (in ytirs K of list birthdtyl,
    b.   Sex IM for milt, F for fimtlt).
    c.   Whether or  not each person ii • smoker II - Yu, 2* No, 3 "Do not knowl,
    a.   Whether or  not each person presently suffers respiratory distress II * Yts. 2 - No. 3 ' Do not know),
    e.   Whether or  not each person a presently under medical care II - Ya, 2 - No. 3 * Do not knowl,
    f.   Whether or  not each person is presently taking prescription medication II • Yts. 2 * No, 3 - Do not knowl.
    g.   Each person's primary hobby Isuch a samps, coins, ptinrino, gardening, building models, refinishing furniture. nc.1.
    h.   The nature  of the business where each person works Isuch ft tt horn*, school, servict sntion. tank, ete.1, and
    I.   Relationship to you*
    beginning with the oldest and proceeding to the youngest (Enter appropriate codes or responses in matrix talowl:
Household
Member
Number
01
02
03
04
05
06
07
OB
09
10
0.
Age
(Yuri)










b.
Sex
(M or Ft










t
Smoker










d.
Respiratory
Dlltreu










S.
Medical
Care










t
Preecrlptlon
Medication










g.
Hobby










h.
Nature of
Business










i. Relationship
to
Respondent










Participant
Number




















 11.  a    Does anyone in your household have as a hobby painting, building models, gardening, or refinishing furniture?

         I 1  I Yes ICo m Question llbl      \  t \ No       | 3 |  Do not know (Go to Question 121
     b.   If yes, indicate relationship to respondent
         and household member number(s) from Question 10
 12-  a.   Is anyone in your household employed at a painter or in a service station, garage, furniture repair shop, or chemical plant?

          I ' I  Yes (Go to Question 12bl     \ 3 \  No        | 3 |  Do not know (Go to Question 131
    b.   If yes, indicate relationship to respondent .
         end household member number(s) from Question 10
 13.  Is anyone in the household eligible to participate in the survey?    |jlj Yes (Go to Question 141       FT] No ISTOPH

       If it is apparent that the household contain no persons eligible to participate in this study, thank the respondent and proceed to the
       next household. However. If persona  in the household appear to be eligible to participate « this study, continue to question 14.
 14. Would you participate in a health study a* a paid volunteer
                                                              	           '••••—           •-
 15. In your opinion, would other members of your household participate in a health survey as a paid volunteer?
                            Q|   I               I	1        i	1                  .	.
                  Yes. all    I a I  Yes. some
     r?    QYM      [J]No        [j] Donotknow

     participi

N°     I 4 I Do "*" know    | a |  Inrtigible
4   Ineligible
                                                              169

-------
Site Number
                            RESEARCH TRIANGLE INSTITUTE




                           STUDY OF BENZENE BODY-BURDEN




                           HOUSEHOLD SCREENING LOG







                               Segment Number
                                                                Interviewer Number
                  Date
                       IMonth)
                                 (D,y)
i-m
    fYurl
                                                       Day of Week
n
Household/
Hauling Unit
Number





























































































Street Addren
or Description
























«






Eligible
Household
Member(t)
Yes































No































Member(s)
Agree(s) to
Participate
Yet































No































Number
of
Participants






























































Reason(s) for Nonparticipation, Nonresponse,
or Ineligibilitv































                                        170

-------
COMMENTS
       171

-------
                                                                            OMB No. 158-S78010
                                                                     Approval Expires September 1980
                         STUDY OF BENZENE BODY-BURDEN
  Sponsored by:
    Office of Toxic Substances
    Environmental Protection Agency
    Washington. D.C. 20460
                                           Conducted by:
                                             Research Triangle Institute
                                             P.O. Box 12194
                                             Research Triangle Park, North Carolina 27709
                     QUESTIONNAIRE
      THE RESEARCH TRIANGLE INSTITUTE OF RESEARCH TRIANGLE PARK. NORTH CAROLINA, IS
      UNDERTAKING A RESEARCH STUDY FOR THE U.S. ENVIRONMENTAL PROTECTION AGENCY
      OF LEVELS OF BENZENE ABSORPTION BY PERSONS LIVING IN COMMUNITIES EXPOSED TO
      BENZENE. THE INFORMATION RECORDED IN THIS QUESTIONNAIRE WILL BE HELD IN STRICT
      CONFIDENCE AND WILL BE USED SOLELY FOR RESEARCH INTO THE EFFECTS OF ENVIRON-
      MENTAL FACTORS ON PUBLIC HEALTH. ALL RESULTS WILL BE SUMMARIZED FOR GROUPS OF
      PEOPLE; NO  INFORMATION ABOUT INDIVIDUAL PERSONS WILL BE RELEASED WITHOUT THE
      CONSENT OF THE INDIVIDUAL. THIS QUESTIONNAIRE IS AUTHORIZED BY LAW (P.L 94-469).
      WHILE YOU  ARE NOT REQUIRED TO RESPOND, YOUR COOPERATION  IS NEEDED TO MAKE
      THE RESULTS OF THIS SURVEY COMPREHENSIVE, ACCURATE, AND TIMELY.
                          Study number:
Site number:
-    D
                  Segment number:
                                       Household number:
Participant number:
                                        172

-------
NOTES
     173

-------
First, I  would like to aik tome general question! about you.

 1. Sex  (by obttnetion):    |_jj Male   | 2 |  Female         5.  What is your birthdate?
                                                                            IMonthl      IDiyl
                                                                                                    (Yuri
2.  Race:
               Hlv.nlc
                                   m Am.rlc.n Indian/
                                   |jj Alaskan N.tlv.
                                                                                       I   T~l    I   I    [
                                                                                     - I   I   I - I.  I    I
            "7] Black, not of        H"] Ailan/Paelf te        g   What is your approximate weight in pounds?
               I Hispanic origin      I   I Islander
              1 Whit*, not of
              J Hispanic origin
rTlOtllar
 6 I (Specify).
                                                                                       Ibl.
                                                                                             m
                                                                                                  Do not know
 3.  Household member number /from HSQI:     \   \   \


 4.  What was your age in years at last birthday?   [   |   |
                                                   Years
                                                          7.  What is your approximate height in feet and inches?

                                                                            [_| Feet       [   |[ Inches
Next, I would like to ask some questions about your occupation.


 8.  Are you presently employed in any capacity?  | 1 |  Yes IContinutl     \ 2 \  No (Go to Q. 121


 9.  How long have you been employed by your present employer? [    |   [  Units  | * |  Days  |  2 [  Months   [ 3 |  Years
10.   Does your occupation usually take you away from home?     |  1  |  Yes (Continutl

11.   What is the nature and location (street address) of the company for which you work?

     (Specify! 	
                                                                                       |  2 I  No (Go to Q  131
                                                                                                     (Zip Code)
12.   If not presently employed, which of the following best describes your status?
                    Housewife ~\
                                (Go to Q 161
                    Student  J
                                                                                         (Continual
13.  What is/was your usual occupation? (Specify I -


14.  Are you presently employed in this occupation?     | 1 |  Yes      | 2 [  No

15.  If yes to above question, how long have you been employed in that occupation?

     (Questions 14 and 1 5 miy btikipptd for unemployed,         i - 1 — i        i - 1        r— i           i - 1
     retired, end disabled pmoro.)                              \   \   \ Units  |_JJ  Days  |_*J Months    | 3 |  Years

16.  Have you worked at any of the following occupations/businesses at any time during the pest week?


                                            Q]  Yes (STOP!)        [_*_] No (Continue)


                                            [_jj  Yes (STOPI)        | 2 | No (Continutl


                                            Qj  Yes (STOPI)        [T] No (Continutl


                                            Qj  Yes (STOPI)        |jj No (Continue!
     ..    Painting


     b.    Service station or garage


     t.    Chemical plant


     d.    Petroleum plant
          Furniture refi
                     inishing or repair       | 1 J Yes (STOPI)        | 2 | No (Continue!
                                                        174

-------
Next I would like to ask some questions regarding your health and personal habits.


17.   Do you smoke?       [7] Yes  (STOPII    [7] No  (Continue!
18.  What is the average number of hours that you spend out of doors each day?
                                                                             |    |   |  Hours
19.  How many hours of the day. on the average, do you normally spend away from home?
                                        Hours                           Houn

                          Weekdays   |   |   |              Weekends  |   |   |



20.  What do you consider the current status of your health?

                                      QJ  Excellent    [ 2 | Good     | 3 |  Fair   | 4 |  Poor



21.  Are you currently taking any prescription medication(s) on a regular daily basis?    | 1 |  Yes (STOPI)     | 2 |  No (Continue!
22.   Have you taken any non-prescription medications in the past 48 hours?           |  1  |  Yes      |  2 |  No
                                                                                 | 1 |
         If yes. specify
 23.   Are you presently under a doctor's care?      | ' J Yei       | 2 |  No
         If yes. specify reason
24.  Are you presently suffering from any respiratory problems {such as cold, cough, sore throat, flu. asthma, bronchitis, shortness of
     breath, laryngitis, pleurisy, etc.)?             ,	.                    •	.
                                               I ' |  Yes  (STOPI)      |  2 I  No /Continue)


25.  Have you ever been treated for anemia?      | 1 |  Yes                 2 |  No


26.  How many eggs have  you eaten in the past 48 hours?   [   |   |


27.  Do you pursue any of the following hobbies? (Check all thatappty.t     Ml Furniture refinishing    \ 3  [ Scale models


                                                                      I 2  [Painting               | 4  [Gardening

     Ufa positive response is obtained for any one of these, STOPtJ


Lastly. I  would like to ask some questions about your residence and household.


28.  How many years have you lived in this area?      |   [   |  Years


29.  How long have you lived at this address?         |   |   |  Units    | 1  |  Days      [ 2 | Months     [ 3 |  Years

30.  Do you cool your home with any of the following appliances? (Check all that apply.I

         \ ' [  Central air  conditioning       | 4 |  Window fan(s)               | 7  |  None of these

         I 2 |  Window air conditioner(s)    | 6 |  Ceiling exhaust fan(s)         | 8  |  Do not  know

         | 3 I  Evaporative cooler(s)         | 6 |  Circulating  fanls)             | 9  |  Other (Specify!	
31.  Does your household grow any of its own food in a home garden?   | 1  [ Yes   | 2  |  No    |3j  Do not know

     If yes, specify location of garden	
32.  Where does your household obtain fresh fruit and/or vegetables? (Specify/
33.  What is the primary source of your water for drinking?
                                                                              . cinern
         I  '  I  Bottled water            | 3 |  Tap - community well    |  5 [

         I  2  I  Tap - municipal supply    | 4 |  Tap - private well        |  8 |  Do not know

                                                                    [_£]  Other (Specify}
                                                            175

-------
34. Is that the same primary source of water for drink mixes such as coffee, tea, Kool-Aid, etc?

         |l| Yes    | 2 | NO        If no, how does it differ? (Specify) 	

35. What is the primary source of your water for cooking?

         | 1  | Bottled water             [  3 | Tap - community well      |  B  | Tap - cistern

          2  | Tap - municipal supply     [  4 | Tap • private well
                                                       0NO
                                                                       Do not know
                                                                   | 3 I Do not know
36. Does anyone else in your household smoke?  | 1 | Yes

    Ifyei. check til thtt tpply:  |jj Cigarettes |_2JCigan  |_^Jpipe    \ * [Other (Specify)

                                    lork at any of the following occupations/busi

                     _2J Service station/garage/engine repair  [ 3 [ Chemical/petroli

                                                                    ICtitck ell thtt tpply.t
                    tappiy:  |_-j ugarettes | « [Cigars  |  - |npe    | •» | otner ispecityl 	

37. Does anyone else in your household work at any of the following occupations/businesses? (Check ill thtt tpply.)

      I 1 [Painting   | 2 [ Service station/garage/engine repair [ 3 [ Chemical/petroleum plant  |  * | Furniture refinishing/repair

    Does anyone else in your household pursue eny of the following hobbies?

      | 1 [Painting   | 2 | Furniture refinishing | 3 [Scalemodels  [ * |Gi
38.
                                                                   ,«.,_« .

      |jj Painting   |_*J Furniture refinishing  |_*J Scale models  |  4 [Gardening
39. Has anyone else in your household ever been treated for anemia?  |  1  | Yes    | 2 | No    |  3 | Do not know

    If ..« ..•«.•&.*......*.h.**w ___L~_ _.._k__/_l *__~. U9S>.
    Has anyone else in your household ever been treated for anemia?  | 1 | Yes    |_
    If yes, specify household member number(s) from HSQ:

    Has anyone in your household, including yourself, ever bean treated for leukemia?
      i	1        i	1        i	1
40. Has anyone in you

      fl| Yes    \2\ No     LSJ Do not know

    If yes. specify household member number(s) from HSQ:
                               RESPONDENT/INTERVIEWER INFORMATION
41. Respondent:  [ 1 | Participant  | 2 [ Other (Specify)	


42. Interviewer number:  |    |   |   |                        «. Data of interview: |   |   |  -  |   |    |  -  |   |   |
                                                                                Household member number (from HSQ): I

                                                                               (Month)      (Day)       lYetr)
                                        ' SAMPLE INFORMATION
44. Personnel monitor number
On
Off
45. Br
Date
Month




Day


Mth sample



Collected
Yes
1

No
2
Year





Time
Houn



Minutes
:
:


Flow Rete
(ml/mini





Interviewer
Number





Date on
Date
Month
I
Dey
1
Yeer









Collected Sample
Time
Stert
Hi

xjri : Minutes
: 1
Stop
Hours

: MlnutM
t

Tamp.
C
F
1



Volume Meter
Reading
(cubic feet)
T



If not collected, reason:
lnt«rvi«w«r
Numbw


    Apparatus number |	j
46. Tap water sample
Original Sampled)
Collected
VM
1
No
2
If Collected, Oete
Month


Day


Year


If Not Collected.
Reason

Number



Duplicate Sample
Selected
Yet
1
No
2
Collected
Yes
1
No
2
If Not Collected.
Reason

    Source:
                                                        176

-------
47.  Blood sample (for high exposure area participants only).
Original Sample
Collected
Yes
1
No
2
If Collected, Date
Month


Day

Year


If Not Collected,
Reason

Interviewer
Number



Duplicate Sample
Selected
Yes
1
No
2
Collected
Yes
1
No
2
If Collected, Date
Month


Day


Year


If Not Collected,
Reason

                                                        COMMENTS
                                                             177

-------
                                                                                                  OMB No. 158-S78010
                                                                                        Approval Expires September 1930
                                   RESEARCH TRIANGLE INSTITUTE
                                 STUDY OF BENZENE BODY-BURDEN

                                 PARTICIPANT CONSENT FORM
    I understand that  the  Research  Triangle Institute is engaged in a study  of the exposure and absorption of
benzene by persons living in areas having various levels of benzene in the environment. I understand that the survey
is  being conducted  in order to help measure the levels of exposure and absorption of benzene in populations
environmentally exposed to benzene, and  is limited to the  purpose stated. I further understand that the survey is
being conducted under the auspices of the United  States Environmental Protection Agency in cooperation with the

 Texas and  City  of  Houston Health Departments.	

    I do hereby freely  consent to participate in this study of benzene exposure and absorption and understand that
my participation will  consist of providing  answers to a questionnaire related to environmental exposure and the
following  environmental and  biological samples:  (1)  two  four ounce samples of cold tap water from  a source
commonly used for drinking and cooking,  (2) a sample of environmental exposure collected by a small device which
I will keep with me for a short time, and (3) a breath sample. I understand that an agent of the Research Triangle
Institute will administer the questionnaire  in my home and at the same time collect the tap water samples, instruct
me regarding the exposure monitoring device, and make arrangements regarding collection of the breath sample.  I
understand that after  the  collection of the breath sample  I  will receive  an incentive of five dollars for  my full
participation in the study. I understand that a small number of households and individuals will be selected for the
collection of duplicate tap water samples and reinterview, but that such selection would not entitle me to further
compensation.

    I understand that  my name  will  not be voluntarily disclosed, and that my name will not be referred to in any
way when compiling and evaluating the  results of the study. I understand that participation in this study may result
in no direct  benefits to me, other than those described herein, and that I am free to withdraw from this study at any
time. It has  been explained to me that there are no significant risks to me from participation in this study.  I further
understand that while  participating in the study I will be free to ask any questions concerning the study; if I have
any further questions about the project, I know that I am free to contact

 Dr.  Richard K.  Donelson,  Texas  Department of Health                   512-458-7328
	1	 telephone number ^^^_________ or
 Dr.  Robert  A.  MacLean. City  of Houston  Health Dept.   tiltphon, number  713-222-4295

or Mr. Benjamin S. H. Harris,  III, Survey Operations Center, Research Triangle Institute, Research Triangle Park,
North Carolina 27709, telephone number 919-541-6055.


°UK    III- m -  I   I   I         Participant's Nama:
        IMonthl     IDtyl      lYnr)                                          (Prim)
Site Number:           Segment Number:
                                               Household Number:
                                                                              Participant Number:
SIGNATURES:
Participant:                                       -   Witness:
                                                                            Interviewer Number:
                                                  178

-------
                                                                                                OMB No. 158-578010
                                                                                      Approval Expires September 1980
                                 RESEARCH TRIANGLE INSTITUTE
                               STUDY OF BENZENE BODY-BURDEN

                               PARTICIPANT CONSENT  FORM
                                     (HIGH EXPOSURE AREA)


  I understand that the Research Triangle Institute is engaged in a study of the exposure and absorption of benzene
by persons living in areas having various levels of benzene in the environment. I understand that the survey is being
conducted in order to help measure the levels of exposure and absorption of benzene in populations environmentally
exposed to benzene, and is limited to the purpose stated. I further understand that the survey is being conducted
under the auspices of the United States Environmental Protection  Agency in cooperation with the

    Illinois  Department  of Public Health.
  I do hereby freely consent to participate in this study of benzene exposure and absorption and understand that my
participation will consist of providing answers to a questionnaire related to environmental exposure and the fallow-
ing environmental and biological samples: (1) two four ounce samples of cold tap water from a source commonly
used for drinking and cooking, (2) a sample of environmental exposure collected by a small device which I will keep
with me for a short time, 13) a small (approximately 10 cc) blood sample to be taken from an arm vein, and (4) a
breath sample. I understand that an agent of the Research Triangle Institute will administer the questionnaire in my
home and at the same time collect the tap water samples, instruct me regarding the exposure monitoring device,
and make arrangements regarding collection of the breath and blood samples. I understand that after the collection
of the breath  and blood samples I will receive an incentive of  ten dollars  for my full participation in the study.  I
understand that a small number of households and individuals  will be selected for the collection of  duplicate tap
water and blood samples and reinterview,  but that such selection would not entitle me to further compensation.


  I understand that my name will not be voluntarily disclosed, and that my name will not be referred to in any way
when compiling and evaluating the results of the  study. I understand that participation in this study may result in no
direct benefits to me, other than those described  herein, and that I am free to withdraw from this study at any time.
It has been explained to me that there  are no significant risks to  me from participation in this study. I further under-
stand that while participating in the study I  will be free to ask any questions concerning the study; if I  have any fur-
ther questions about the project, I know that I am free to contact

- Robert  L.  Wheatley _ telephone number 217-782-4674 „.


    Genelle Moore                                                             CIH 900  s.-j^e.
-r— - — --- telephone number  OiO-^gQ-3 /3O

or Mr. Harvey  S. Zelon. Survey Operations Center, Research Triangle Institute, Retearch Triangle Park. North Carolina 27709
telephone number 9 19-541-6054.
Date:   |   |   |     [_|   | - |   |   |         Participant'. Name:
       IMomhl           lYmrl
Site Number:   |	|    Segment Number:  [   |  |    Household Number:  -|  |   |    |    Participant Number-   |   |   |
SIGNATURES:

Participant:  	
                                                                          Interviewer Number:  [   j   |   |
                                                   179

-------
                                                                                                           OMB No. 158-S78010
Medium                                                                                          Approval Expires September 1930
St.  Louis


                                            RESEARCH TRIANGLE INSTITUTE
                                           STUDY OF BENZENE BODY-BURDEN

                                           PARTICIPANT CONSENT FORM


             I understand that the  Research Triangle  Institute Is engaged in a study of the exposure and absorption of
          benzene by persons living in areas having various levels of benzene in the environment. I understand that the survey
          is  being conducted  in order to help  measure the  levels of exposure and absorption  of benzene in populations
          environmentally exposed to benzene, and is limited to the purpose stated. I further understand that the survey is
          being conducted under the auspices of the United States Environmental Protection Agency in cooperation with the

          Missouri Department of Natural Resources  and St.  Louis Air  Pollution Control.

              I do hereby freely consent to participate in this study of benzene exposure and absorption and understand that
          my participation will consist of providing answers  to a questionnaire  related to environmental exposure and the
          following  environmental and  biological samples: (1) two  four  ounce samples of cold tap water from a source
          commonly used for drinking and cooking, (2) a sample of environmental exposure collected by a small device which
          I will keep with me  for a short time, and (3) a breath sample. I understand that an agent of the Research Triangle
          Institute will  administer the questionnaire in my home and at the same time collect the tap water samples, instruct
          me regarding the exposure monitoring device, and make arrangements regarding collection of  the breath sample. I
          understand that after the  collection of the breath  sample  I will receive an incentive  of five dollars for  my full
          participation  in the study.  I understand that a small number of households and  individuals will be selected for the
          collection  of duplicate tap water samples and  reinterview, but that such selection would not entitle me to further
          compensation.

              I understand that my name will not be voluntarily disclosed, and that my name will not  be referred to in any
          way when compiling and evaluating the results of the study. I understand that participation in this study may result
          in no direct benefits to me, other than those described herein, and that I am free to withdraw from this study at any
          time. It has been explained to me that there are no significant risks to me from participation in this study. I further
          understand that while participating in the study I will be  free to ask any questions concerning the study; if I have
          any further questions about the project, I know that I am free to contact

          Rick L.  Roberts,  Missouri Dept. of Nat. Resources                     314-751-3241
          	 telephone number	or

          Charles  M. Copley  or W. L.  Hagar,  St.  Louis	 telephone number 314-622-3334

          or Mr.  Harvey S. Zelon, Survey Operations Center, Research Triangle Institute, Research Triangle Park, North Carolina
          27709, telephone number 919-541-6054.
          DIM:
-ED
                                                      Participant's Name:
                  (Month)     (Dtyl      fYarl                                         IPrintl


          Sin Number:   |	|     Segment Number:     [   |     Household Number:  I   I   I   I   Participant Number:   |   |   |
          SIGNATURES:
          Participant:	   Witness:
                                                                                     Interviewer Number:  |   |   |   |
                                                                180

-------
                                                                                                             OMB No. 1SB-S78010
Low                                                                                               Approval Expires September 1930
St.  Louis  County


                                              RESEARCH TRIANGLE INSTITUTE
                                            STUDY OF BENZENE BODY-BURDEN

                                             PARTICIPANT CONSENT FORM


               I  understand that the Research Triangle Institute is engaged  in a study of  the exposure and absorption of
            benzene by persons living in areas having various levels of benzene in the environment. I understand that the survey
            is being conducted in order  to  help measure the levels of exposure and absorption  of  benzene  in populations
            environmentally exposed to benzene, and is limited to the purpose stated. I  further understand that the survey  is
            being conducted under the auspices of the United States Environmental Protection Agency in cooperation with the

            Missouri Department of Natural Resources and the  St.  Louis County  Health Department.

               I do hereby freely consent to participate in this study of benzene exposure and absorption and understand that
            my participation will consist  of  providing  answers to a questionnaire related to environmental exposure and the
            following  environmental and  biological  samples: (1) two four ounce samples of cold tap water from a source
            commonly used for drinking and cooking, (2) a sample of environmental exposure collected by a small device which
            I will keep with me for a short time, and (3) a breath sample. I understand that an agent of the Research Triangle
            Institute will administer the questionnaire in my home and at  the same time collect the tap water samples, instruct
            me regarding the exposure  monitoring device, and make arrangements regarding collection of the breath sample.  I
            understand that after the collection of the breath sample I will receive an incentive  of  five dollars  for my full
            participation in the study. I understand that a small number of households and individuals will be selected for the
            collection  of duplicate tap water samples and reinterview, but that such selection would not entitle me to  further
            compensation.

                I understand that my name will not be voluntarily disclosed, and that my name will not be referred to in any
            way  when  compiling and evaluating the  results of the  study. I understand that participation in this study may result
            in no direct benefits to me, other than those described herein, and that I am free to withdraw from this study at any
            time. It has been explained to me that there are no significant  risks to me from participation in this study. I  further
            understand that while participating  in the study I will be free  to ask any questions concerning the study; if I have
            any further questions about the project, I know that I am  free to contact

            Rick  L.  Roberts.  Missouri Dept. of Nat.  Resources     ,Biephone number 314-751-3241 or

            Clifford Mitchell.  St.  Louis  County Health Dept.       telephone number 314-726-1100

             or Mr. Harvey  S. Zelon, Survey Operations Center, Research Triangle Institute, Research Triangle Park, North Carolina
             27709, telephone number 919-541-6054.

            °"e:    I   I   I ~  I   I   I ~ I   I  I          Participant's Name:
                    (Month)     (Day)      lYear)                                         (Print)



            Site Number:   |	|    Segment Number:  [   |   |    Household Number:   |   |    |  |   Participant Number:



            SIGNATURES:
            Participant:	Witness:	
                                                                                       Interviewer Number:
                                                              181

-------
       MATERIAL SUBMITTED TO OMB WITH THE DATA COLLECTION INSTRUMENTS


1.   Justification

     Benzene is not only one of the most fundamental and well-known organic
chemicals, but is also of major industrial importance.  In the United States,
benzene ranks 13th in volume (1) with a projected production for 1976 of 1.5
x 109 gallons (2).  Approximately 88 percent of the domestic benzene produc-
tion is from petroleum sources with the remainder from coal (3).  The largest
benzene source is from catalytic reforming processes at oil refineries.
Other major production routes are dealkylation of toluene and as a co-product
with ethylene from steam crackers (2).

     The primary uses of benzene are as an additive in gasoline, chemical
manufacturing and solvent operations with chemical processing being the
major use.  While benzene is used in the commercial production of literally
hundreds of compounds (4), its major uses are as the starting material for
styrene (45%), cumene/phenol (20%) and cyclohexane (17%) (2).   These compounds,
in turn, are used in production of polystyrene plastics and rubbers and
other fabricated plastic products.

     Benzene is wide-spread in the environment, both in the air and water.
While the levels of environmental exposure are significantly less than the
industrial levels, there is no proof than these levels are inconsequential.

     The health effects of benzene have been extensively reviewed recently
(5), especially with respect to its potential carcinogenic effects.  Even
with the recognition of benzene toxicity for over 50 years much of its
action is still poorly understood.  The most serious effect of chronic
exposure is depression of the hematopoietic system ranging from milk rever-
sible depression of some of the formed elements to aplastic anemia and
leukemia.  The latter has been particularly difficult to study since no
animal model has been found in which benzene induces leukemia dispite epide-
miological evidence linking the two in humans.  Other toxic effects of
benzene are central nervous system depression and histochemical changes in
kidney, liver, small intestine, spinal cord and heart (6).

     In view of these serious consequences from chronic benzene exposure, an
evaluation of the exposure/body burden of benzene in the general human
population in areas of relatively high (industrial and urban), medium (urban)
and low (rural) benzene emissions is to be undertaken.

2.   Description of the Survey Plan

     This project is an epidemiologic study of exposure and absorption of
benzene among populations potentially exposed to benzene from urban environ-
ments, manufacturing or industrial users, or industrial storage facilities.
At each of two performance sites, a panel of respondents will be selected
and recruited; this panel will represent varying distances from emission
sources.  A questionnaire will be administered for each individual selected
for the study to obtain information on demographic variables,  residence


                                    182

-------
histories, and potential special exposure situations.  For each individual,
a sample of exposure (using a small personal monitoring device) and breath
(representing absorption or body burden) will be collected; tap water samples
will be collected at each residence.  Samples (air, breath, and water) will
be analyzed for benzene by gas chromatography/flame ionization detection.

     Two locations, St. Louis, MO, and Houston,  TX have been chosen as
performance sites based on exposure and emisson information (7).  Subject
selection will be by stratified area sampling; three areas at each site will
be geographically designated as high, medium, and low exposure area, on the
basis of emission sources and wind patterns.  To compensate for meteorological
variability, the areas will radiate in all directions from the emission
sources.  The exact boundaries of the target areas are subject to local
condition data which will be obtained by site visits of the person responsible
for drawing the sampling frame.  Since some of the low exposure areas will
be suburban and rural, census data may not be available for the total target
population.  The combined total target population in the St. Louis and
Houston areas is 1,857,377; from this target population, a total of 150
persons will be recruited.

     Delineation of exposure area boundaries will use dispersion modeling
and other mapping techniques.  Within each exposure area, sublevels will be
established based on city blocks and other physical features.   Household
interviewers will be assigned specific segments  to canvass and a specific
order in which to do the canvassing.  Persons contacted and meeting eligibi-
lity requirements will be asked to participate.   This process will continue
until either all segments have been exhausted or the target population has
been achieved.  A record of all household contacts will be made; nonrespon-
dents (those not able to be contacted) and nonparticipants (refusals) will
be recorded for each interviewer and at each site.  These records will be
compared to determine the likely effect of any bias in the final results.
To try to reduce the nonparticipation rate, and to reimburse the subject for
time spent on the study, volunteers will be offered a $5.00 incentive for
participating.

     Approximately 75 persons, evenly divided among the three (high, medium,
and low) exposure areas, will be selected at each site.  In order to partici-
pate in the study, an individual must meet certain criteria; potential
participants must reside in the target areas during the data collection
period, be 25-50 years of age, and be at their place of residence during the
time that exposure is monitored.  In addition, potential participants will
be carefully screened to eliminate individuals who smoke, suffer respiratory
distress, take prescription medications on a regular daily basis, and who
experience occupational or avocational exposure.

     The agency statistician who has reviewed this work plan and who has
been involved from the selection of the contractor is:
                                    183

-------
               David Svendsgaard
               Office of Statistics and Data Management
               Health Effects Research Laboratory
               Research Triangle Park, NC  27711
               FTS 625-2468

     The contractor for this study is:

               Research Triangle Institute
               P.  0. Box 12194
               Research Triangle Park, NC  27709

     The contractor is responsible for all phases of the study, including
study design, subject recruitment, chemical analysis of all samples, human
surveying and the statistical analysis and report writing.   RTI promises to
ensure the confidentiality of all personal data collected under this contract.
The only place a person's name and his study number will appear together is
on the consent form which will not be converted to machine-readable form and
will be sored in a secure area.  All material will be entered into the
computer by study identification number.  All publications  resulting from
this project will use statistical compilations of data.  No individual names
and associated data will be released.

     The contractor is presently testing and refining procedures for measuring
benzene exposure and absorption.  The data collection instruments are modifi-
cations of questionnaires which were used in another EPA-sponsored research
effort (OMB No. 158-S77006) which have been administered to some 1,115
respondents; the data collection instruments for the benzene study are more
conside, efficient, and relevant, having profited from the  earlier study.

3.   Tabulation and Publication Plans

     The results of the project will be summarized in a final report from
the contractor to EPA.  A draft of this report should be available approxima-
tely 9 months after approval of the questionnaire.

     In the analysis, RTI will examine the following relationships:

          a.   Analysis of the relationship between the levels of benzene in
               humans as measured by breath samples and various levels of
               exposure to benzene (£.£., low, moderate, and high or urban
               and rural).

          b.   Analysis of the relationship between the levels of benzene in
               human breath samples and environmental levels of benzene
               (i.£., levels in air, water, and possibly food).

The principal statistical techniques that will be used to examine these
relationships are the analysis of variance and multiple regression.  In some
instances, it may be worthwhile to employ the technique of stepwise regression;
this technique can be used to give insight into the relative strengths of
the various demographic and environmental variables in predicting toxic body


                                    184

-------
burdens in humans.   In addition to using these three techniques, other
techniques which will be employed to examine the relationships of interest
include computing correlations between pairs of variables; examining scatter
plots of body burden level versus benzene exposure levels and environmental
and demographic variables; and computing means of the environmental and
demographic variables for various body burden levels and then plotting these
means.

4.   Time Schedule for Data Collection and Publication

     Within six weeks of approval of the study RTI will be in the field at
the first study site.  Assuming a May approval date, data collection could
begin in June.  Five months will be required for data collection.  To complete
all analytic work and produce the draft report will take three months after
this.

5.   Consultations Outside the Agency

          Dr. Frank Johnson
          National Institute of Environmental Health Sciences
          Research Triangle Park, NC

          Dr. John M. Harrelson
          Assistant Professor of Orthopedic Surgery and Pathology
          Duke University Medical Center
          Durham, NC

          Dr. Stephen H. Gehlbach
          Adjunct Assistant Professor of Epidemiology
          School of Public Health
          University of North Carolina
          Chapel Hill, NC

          and

          Assistant Professor, Department of Community Health Sciences
          Duke University Medical Center
          Durham, NC

6.   Estimation of Respondent Reporting Burden

     The burden of this project on respondents covers:  the time necessary
to complete the questionnaire; the time and inconvenience of allowing the
field interviewer into the household to collect the tap water samples; and
the time and inconvenience of providing the breath and exposure samples.
All efforts will be undertaken to reduce to a minimum respondent burden, but
in order to complete all household data collection, approximately one hour
of time may be required of the participant, including the collection of all
relevant samples.  More specifically, we anticipate that up to 30 minutes
may be required to complete the questionnaires for a participant, 15 minutes
will be required to collect the tap water sample and explain the exposure
monitor, and 15 minutes will be required to obtain the breath sample.


                                    185

-------
7-   Sensitive Questions

     None of the questions is considered  to  be  particularly sensitive.

8.   Estimate of Cost to the Federal  Government

     The present estimated cost  of  the project  is  $150,000.

9.   References

     1-    E.  V.  Anderson,  Chem.  Eng.  News, 54 (19), 34  (1976).

     2.    Anon.,  Chem. Eng.  News, 54, (47),  17  (1976).

     3.    P.  H.  Howard and P. R. Durkin,  Sources of Contamination, Ambient
          Levels, and Fate of Benzene in  the Environment.  EPA 560/5-75-
          005, Dec., 1974.

     4.    SRI, Chemical  Origins  and Markets, Stanford Research Institute,
          Menlo  Park, CA,  1967,  p.  18-19.

     5.    Benzene Health Effects Assessment, U. S. Environmental Protection
          Agency, External Review Draft,  October 1977.

     6.    J.  J.  Kocsis and R. Songder, "Current Concepts of Benzene Toxicity",
          CRC Crit. Rev. Toxicol.,  3, 265  (1975).

     7.    S.  J.  Mara and S.  S. Lee, Human Exposures to Atmospheric Benzene,
          EPA Contract 68-01-4314, Final  Report, October, 1977.
                                   186

-------
                       APPENDIX E




DATA LISTING OF AIR AND BREATH BENZENE LEVEL DISTRIBUTIONS




             BY SITE AND BY EXPOSURE STRATA
                          187

-------
            Table E-l.   AIR AND BREATH BENZENE LEVELS AND SAMPLE WEIGHTS FOR HOUSTON AND ST. LOUIS
CO
CO
DBS
i
3
3
4


t
9
10
1 2


IS
16

• a
20
21
22
ad
-»Q
?b
27
28
30

Ji
33
34
36
37
30
39
40
l\

nil
as
46
17
.. 	 	 	 	 - 48
49
Sff
51
52
53
54
55
5*

Site
HOUSTON
HOUSTON
HOUSTON
	 HOUSTOU—
--HOUSTON

HOUSTON
HOUSTON
— HOUSTON—
HOUSTON

HOUSTON
HOUSTON



HOUSTON
HOUSTON
	 HOUSTON-
HOUSTON
HOUSTON
HOUSTON
HOUSTON
	 HOUSTON-
HOUSTON -
HOUSTON
HOU9TOM
HOUSTON
HOUSTON

HOUSTON
HOUSTON
HOUSTON
. HOUSTON

HOUSTON
HOUSTON
HOUSTON
- HOUSTON
si LOUIS
SI LOUIS
ST LOUIS
SI LOUIS
31 LOUIS
sr LOU'S

Expo-
sedi/
HIGH
HIGH
HIGH
HIGH
Hl^H
HIGH
HIGH
HIGH
HIGH
HIGH
HIGH
HIGH
• MJRH
HIGH
HIGH
	 MEDlUH-
-MCDIUM-

MtnniM
MEDIUM
MEDIUM
	 MEDJUH-
	 MEDIUM
HFDltlM
MEDIUM
MEDIUM
	 MEO.IUM.
MfDfllM
	 MEDIUM-
MEDIUM
MfcDIUM
MIX

1 (IH
1 (IN
LOH
LllW

L0»

LOH
. ... L0« 	
LOH
III*.
HIGH
HIGH
. man
HIGH
HIGH

Air*/
16.0
IP. n
8,6

14,0

13,0
35,0
h, |
1 1 ,o
fc,7

6,7
18,0
10,0
lo, o
"iff

30,0
25.0
	 LUO—
n,a

1,1
33.0
26,0
at*
11, n
4S,o
ln,n
20,0
25,0
O.B
a, l
18,0
18,0
13.0
17,0
l.fl
4.5
21.0
a.n
18,0
17.0
9.0
	 13,0
27,0
S,fc
17,0
25,0
	 86. 0
»6,0
17.0
lu.n

Breath
3.1
0,0
1.7
2.8
ft H
c h

e*l
3.0
o.o
"
0<0
. "„
1.4
3.5
-1.0

7 o
1.0
* •'
4,8
^ 8
2|S
^2)0 T"
-1,0 ...

1,8
4.4

y*5
n 4
o|o

0.3
»o
o!?
1.7
. 1.9
_. ... 4.0
	 8,2 	
9.2
. 9.'?
16.0
4.2

Weight-''
27.0
27,0
27.0
27.0
J7,n
2J,fr
23.6

23.6
29,0
",o

?s"n
?^ °
25.4
25,4
10^60 0
ikOBn'n
fcoan'o
6080,0
22775,0
A55& o
7973,0
797Sto
11387,5
8350,8
-8350^8 	
8350,8
Stra-
tum
l
i
i
,
i

i
i
3
a
f
p
2
i
1
1
1

1
2
j,
t
3
3
1 '
6825,0 4

6825,0
13650.0
5518.1
5518.3
5518.1
5518,1
3110.8
394.0
2482.2
2482.2
.2877.5
2877.5
2079,0
2079,0
2079.0
2079.0
2079.0
sisa.i
10.8
in. 8
10,8
10.8
10.8
9,0

a
4
4
i
1
1
i
1
2
_2 	
2
3
3
3
3
3 	
j
1
1
1 •
l
1
1
1

SEG2
n I
u )
u
it
n
)2
12 '
12
12
19
IS
15
17
17
17
17
26
26
27
27
27
29
24
24
25
25
20
21
21
21
22
22
22
23
3J
33
33
33
37
34
36
36
30
30
32
32
32
32 . ..
32
35
31
31
31
31
32 	 " •


-------
                                               Table E-l (cont'd.)
Co
063
57
59
60
	 _ 	 	 	 ti
. 6J
65
66
	 68
6?
i 71
72
	 	 	 Ta

77
78


01
62
83
ea
i •*

1, . *i
89
70



95
', 96



101
102
	 - — 	 	 	 I0«
- 	 ... 	 105
107
1, IOB
1" - 110
111
| l">

SHE
sr LUUIS
31 LUUI3
3f LOUI3
3f LUU13
-- sr LOUIS
5f LUUI3
31 LUUI3
	 8-T-fcOUlS
— sr- LUUIS

3t LOUIS
sr LUUIS
sr LUUIS

3t LOUIS
3F LUUI3



sr LOUIS
3r LUUIS



3f LOUIS
3f LUUIS


	 sr— t^uis-
Sf LUUIS
sr LOUIS
— sr LOUIS
	 81 LUUIS
sr LUUIS
sr LUUIS
- 31 LUUI3
sr LOUIS
sr LOUIS
SI LUUIS
SI LUUIS
sr LOUIS
sr LOUIS

EXPOSED
HIGH
HKH
HIGH
HIGH
--HIGH —

HIGH
HIGH
lllGH
	 HIGH 	
1(1 CH
H'liH
HIGH
	 HJCU 	
— HIGH 	
((IUH
MCpIMH
MEDIUM
MEDIUM
MEDIUM



MEDIUM
MEDIUM
MEDIUM


MEDIUM.
MEDIUM
MEDIUM
MEDIUM
MEDIUM

MEDIUM
MEDIUM
L0l"
LD*
{.0*
(_QH
LOW
L0«
(.OH
1 3H
- 1 OH
LOH 	
LOH
LOH
LON
LOW
LON
LOH
Lnj

*I«
1
| f f Q
6.2
23.0
	 52,0

3,7
17,0
	 £5.0

35,0
16.0
	 19,4-
	 t " -
7 1
i
t

	 4,0 -
•
3.5


^7*0
iB>0
20*0



J9.0
32,0
170
	 S2.0
9t |
«6,0
44,0
10 i 0

	 -t
iO , 0
10.0
118,0
'1 . "*
1.2
3,4

BREATH
N
8,9
3.8
----- 7./
3(0
0.1
5.7
	 4,7
. c
I
2.5
n C
- 6,2

19,'o

7*0
•
«.«
11,0



3,8
2,3
"


7,0
11.0
I i
2 2
r 5
6 ?
«,'
.

— i,«
•
8,8
17 0
5.0
.
1 l

WEIGHT
9,00
6.67
6.67
6,67

9.75
9.75
9 75
	 9.75 -
9 TC
9.75
9.75
Q 7C
--- 9,75 -
9 75
??t,7 ?s
2267,25
1763,«2



2016.00
1602.32


15«I "7
1583,47
1583,47
1 5 ** 1 '17
1S93 47

15^3 47
1583. 07
282,86
2"2 *fr
2S2,"(>
2«2,"6
fKf n*.
6611,71
66U.7J

66«f7| --
fciSd 7 1
664.71
413,60
413 60
413,60
413.60
413.60

STRATUM
1
1
1
1


2
2
p
	 2 	
C
J
i
2
j
. .. z . _

1
1



2
3
3


j
3
3



3
1
|
|
|

1
1
•

	 1
1
1
t
1
1

SEG2
32
T?
35
35
»•;
_ .. js 	 . . --.
\^
33
33
ll

\}
33
34
Ml
_ . JH . 	 _ . _ 	
34
1 \
11
12
1 2



14
15
(S
15
IS
Ih
16
16
Ik
1 6
16
1 **
16
21
?l
2(
21
?|
23
23
?*
23
	 23 . 	 _. 	 	
?\
23
24
?U
24
24
?U


-------
                                                          Table  E-l (cont'd.)
                                        DBS
                                                SITE
                                                         EXPUSED
                                                                  AIR
                                                                         BREATH    HEIGHT
                                                                                            STRATUM
                                                                                                      SEG2
Ill
115
116
i 1 7
1 1 A

ST LHUIs
ST LOUIS
ST LOUIS


LUH
1 llbl
LUH
LU«
- i OH
LUM

.
«J,0
J.9
1 9
, -
*
6.1
*:*
5 •>
16*0

4U.6UO
S?0,OOD .,
520,000
297, 1«1
297,1*3 	

1
?
2
2
.i.

21
PS
25
26
2f>
?h 	


                                         J^/   Sampling strata exposure level = high, medium or low


                                         2/   Air, breath = ug/m3


                                         _3/   Sampling weight used to weight benzene air and breath

                                              levels so that area estimates could be made.
VO
o

-------
                IS
                ?0
                10
                                                                               Spearman Corr. • +.06
                                            (IS)




                                            LOW
  (18)





MEDIUM
 (15)





HlRH
                                                       EXPnSFri
NUTtt      I nfi
-------
                             (ug/m1)
                             BPEATH I
VO
NS
                             to


                             1J


                             12


                             11


                             in
                                                                                  Spenrman Corr. • .23
(12)
                         (16)
                                                   (15)
                                                            i nv
                                                                                    MEntUH
                                                                                                              HIRH
              MUTE:      7 -Wt H»0 MISSINfi V4LIIES

                        ( ) - sample size
                                                 il  tins
                                  Figure E-2.   Site  = Houston.   Plot of breath  exposed =  *.

-------
CO
                                  (g g/..3)
                                   MR j
                                   ISO »
130


lln


120


110


100


 to


 HO


 TO


 bO


 SO


 UO


 in


 lo
                                                                                                 Spearman Corr. - -.02
                                                                                                                   (23)
                                                                                         (12)
                                                                (18)
                                                                LOW
                                                                                                                   HIr.H
                                                                           EXPnSFfl
                  lUTEt     IS nB« «»D HlS'iIwr, VALUES

                           ( ) sample size
                                                      ?« MRS
                                   Figure  E-3.    Site =  St.  Louis.   Plot of air exposed  =  *.

-------
              bRMTH




               27.'5
               25.1)
               22.5
               20.0
               17 .'5
               15.0
               12.5
               10.0
                7,5
                5.-0
                2.5
«>Ul     13 IB* H«0 MISSING VJLIJES



         ( ) sample size
                                                                                  Spearman Corr. • .09
                                              (17)





                                              LOW
  (15)





HErtlUM
 (23)




Hlr.H
                                     MPS HlOOtN
                Figure E-4.   Site =  St.  Louis.   Plot  of  breath  exposed =  *.

-------
TECHNICAL REPORT DATA .
(Please read Instructions on the reverse before completing!
1. REPORT NO. 2.
4. TITLE AND SUBTITLE
MEASUREMENT OF BENZENE BODY-BURDEN FOR POPULATIONS PO-
TENTIALLY ENVIRONMENTALLY EXPOSED TO BENZENE
IN THE ENVIRONMENT
7. AUTHORISIR. A. Zweidinger, S. D. Cooper, B.S.H. Harris,
III, T. D. Hartwell, R. E. Folsom, E. D. Pellizzari,
A. S. Sherdon, T. K. Wone and H. S. Zelon
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Research Triangle Institute
P. 0. Box 12194
Research Triangle Park, NC 27709
12. SPONSORING AGENCY NAME AND ADDRESS
Office of Pesticides and Toxic Substances
U. S. Environmental Protection Agency
Washington, DC 20460
3. RECIPIENT'S ACCESSION NO.
5. REPORT DATE
6. PERFORMING ORGANIZATION CODE
8. PERFORMING ORGANIZATION REPORT NC
Task I - Final Report
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
EPA No. 68-01-3849
13. TYPE OF REPORT AND PERIOD COVERED
Final - 12/7/77-6/10/80
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
16. ABSTRACT
      A pilot study was performed to assess the measurement of benzene  body-burden for
 populations potentially environmentally exposed to benzene.  Probability sampling was
 used to select the participants in the two study geographical sites, Harris  County,  TX
 and St. Louis, MO plus parts of Wood River, R.oxana, South Roxana and Hartford,  IL.
      Benzene levels were measured for the air and water environmental  exposure  for
 each participant and the benzene body-burden was measured through breath levels and,
 in  a subsample, blood levels.
      A pretest of occupationally exposed and non-exposed individuals was used to test
 analytical methodology and the concept of breath as an indicator of body-burden.  The
 blood tienzene levels expected and observed required analytical methods capable  of
 measuring  - 1 yg/L or below.  This methodology did not exist and had to  be developed
 for the pretest and pilot study.  Benzene levels for smokers and non-smokers were
 compared in the pretest.
      The range of air benzene levels found in the Harris County study  (49 participants)
 was 2 to 45 yg/m3 with a weighted means of 16.1 yg/m3; breath levels ranged  from 0 to
 14  yg/m3 with a weighted mean of 2.9 yg/m3.  In the St. Louis (68 participants) study
 the range  of air benzene levels was 3 to 125 yg/m3 with a weighted mean  of 26.8 yg/m3;
 breath levels ranged from 1 to 26 yg/m3 with a weighted mean of 8.5 yg/m3.
17.
                               KEY WORDS AND DOCUMENT ANALYSIS
                 DESCRIPTORS
                                             b.IDENTIFIERS/OPEN ENDED TERMS
c. COSATl Field/Group
 Benzene
 Body-burden
 Benzene  in Air
 Benzene  in Breath
 Benzene  in Blood
 Probability Sampling
RELEASE UNLIMITED
                                             19. SECURITY CLASS (This Report)'
                                                  UNCLASSIFIED
                                             20. SECURITY CLASS (This page)
                                                  UNCLASSIFIED
21. NO. OF PAGES
    206
22. PRICE
EPA Form 2220-1 (Rev. 4-77)
                      PREVIOUS EDITION IS OBSOLETE
                                              195

-------