^r
-------
Chlorinated Dioxins and Furans
in the General U.S. Population:
NHATS FY87 Results
Final Report
Prepared by:
John S. Stanley
Midwest Research Institute
425 Volker Boulevard
Kansas City, MO 64110
EPA Contract No. 68-DO-0137, Work Assignment 27
and
John Orban
Battelle
Columbus Division
505 King Ave.
Columbus, OH 43201
EPA Contract 68-02-4292
for the:
Field Studies Branch
and
Design and Development Branch
Exposure Evaluation Division
Office of Toxic Substances
U.S. Environmental Protection Agency
401 M Street, S.W.
Washington, D.C. 20460
U.S. Environmental Protection Agency
Region 5.Library CPM?><
77 West Jackson D .'' .., .12Ji Floor
Chicago, IL 6060s-, j
-------
DISCLAIMER
This document has been reviewed and approved for publication by the Office of
Toxic Substances, Office of Pesticides and Toxic Substances, U.S. Environmental
Protection Agency. The use of trade names or commercial products does not
constitute Agency endorsement or recommendation for use.
-------
AUTHORS AND CONTRIBUTORS
The determination of the levels of the polychlorinated dibenzo-p-dioxins (PCDDs) and
dibenzofurans (PCDFs) in the general population of the United States described in this report
was achieved through cooperative efforts of many EPA and contract support staff. EPA staff
participating in the program included principal investigators from the Field Studies Branch
(FSB) and Design and Development Branch (DDE) of the Exposure Evaluation Division (BED)
of the Office of Toxic Substances (OTS). Contract support to OTS was provided by Midwest
Research Institute (MRI) under EPA Contract Nos. 68-02-4252 and 68-DO-0137, and Battelle
Columbus Division under Contract No. 68-02-4292. The roles and responsibilities of each of
these organizations and contributing authors to this report are presented below.
Midwest Research Institute (MRI)
MRI was responsible for the coordination of the collection of the FY87 NHATS
specimens, preparation of the NHATS composites and quality control (QC) samples,
conducting the HRGC/HRMS analysis of the composites, reporting the results and
coordination of the preparation of this final report. Individuals contributing to this final
report included: Dr. John Stanley, Ms. Karin Bauer, Mr. Paul Cramer, and Dr. Jerry Flora.
Battelle Columbus Division
Battelle was responsible for developing the FY87 NHATS specimen collection design,
creating and maintaining the data bases on the Patient Summary Reports (PSRs), designing
the specimen compositing plan and the statistical methodology for data analysis, and
conducting the statistical analysis to develop estimates of the PCDD and PCDF residue levels
in the general U.S. population based on demographic factors. Individuals contributing to this
final report included: Dr. John Orban, Dr. Robert Lordo, Dr. Al Unger, Dr. Ron Menton,
Ms. Barbara Leczynski, Ms. Tamara Collins, Ms. Pam Harford, and Ms. Claire Matthews.
Westat
Mr. John Rogers of Westat provided technical support in processing all analysis reports
and conducting outlier analyses.
EPA Exposure Evaluation Division (BED)
EPA was responsible for oversight in the development of the study plan, managing
and coordinating the conduct of the overall study, and reviewing, editing and finalizing this
report. Key staff included: Ms. Janet Remmers, Mr. John Schwemberger as Work Assignment
Managers and Dr. Joseph Breen and Ms. Edith Sterrett as Project Officers.
November 27,1991
111
-------
CONTENTS
Preface iii
Figures vii
Tables viii
Glossary xi
Executive Summary 1
1.0 Introduction 7
1.1 Background 7
1.2 Emphasis on PCDDs and PCDFs 7
1.3 Related studies 9
1.4 Objectives of the FY87 study 9
1.5 Report organization 12
2.0 Summary and Recommendations 13
2.1 Summary of FY87 NHATS results 13
2.2 Comparison of FY87 NHATS with other studies 18
2.3 Recommendations for future studies 21
3.0 Overall Data Quality 23
3.1 Sampling 23
3.2 Compositing 25
3.3 Demonstrated laboratory quality control 26
3.4 Statistical analysis 27
4.0 NHATS Sample Design 29
4.1 Sampling design 29
4.2 Sample collection procedures 35
4.3 Sample collection summary 37
5.0 Composite Design 41
5.1 Design goals and compositing criteria 41
5.2 Laboratory compositing procedures 43
5.3 Summary of composite samples 45
6.0 Chemical Analysis Procedures and Quality Control Data 49
6.1 Sample preparation 49
6.2 HRGC/HRMS analysis 52
6.3 QA/QC for chemical analysis 59
6.4 Synopsis of analytical results 72
6.5 Statistical analysis of the quality control data 76
7.0 Statistical Methodology 85
7.1 Statistical model 85
7.2 Statistical analysis 89
8.0 Results 93
8.1 Data restrictions and descriptive statistics 93
8.2 Population estimates 95
8.3 Hypothesis testing 98
8.4 Estimated rates of change of selected PCDD and PCDF
concentrations 100
-------
CONTENTS (continued)
8.5 Outlier detection 102
8.6 Model validation . . 104
9.0 Comparison of FY87 Data with FY82 and VA/EPA Data Bases . 107
9.1 Overview of the analytical programs 108
9.2 Comparison of study designs 108
9.3 Comparison of analytical procedures 110
9.4 Comparison of results 115
10.0 Bibliography 143
Appendices
A FY87 NHATS Composite Data Listed by Analyte for Each Composite A-l
B Analytical Protocol B-l
C Method for Estimating Measures of Uncertainty C-l
D Plots of Estimated Concentrations Versus Spiked Level with Tolerance
Bounds for FY87 NHATS QC Samples D-l
E Supplementary Descriptive Statistics for FY87 NHATS PCDDs and
PCDFs E-l
VI
-------
FIGURES
Number Page
ES-1. National average concentration for PCDDs and PCDFs from FY87 NHATS 4
2-1. Estimated national average concentrations for the target PCDDs and PCDFs
from the FY87 NHATS 15
2-2. Estimated averaged concentrations by age group for selected PCDDs and
PCDFs from the FY87 NHATS 17
3-1. Program steps designed to ensure overall program quality 24
6-1. Example of the separation of 2,3,7,8-TCDD from other TCDD isomers on a 60-m
DBS column 61
8-1. Measured concentrations of 2,3,7,8-TCDD versus average age with estimated
regression lines 101
9-1. Comparison of analytical methods for FY82, VA/EPA, and FY87 studies 112
9-2. Comparison of analytical standards from the FY82, VA/EPA, and FY87 studies 114
9-3. Example of profile plots 122
9-4. Profile plots of average 2,3,7,8-TCDD concentrations with 95% confidence limits
for FY82 and FY87 adipose tissue samples 125
9-5. Profile plots of average 2,3,4,7,8-PeCDF concentrations with 95% confidence
limits for FY82 and FY87 adipose tissue samples 126
9-6. Profile plots of average 1,2,3,7,8-PeCDD concentrations with 95% confidence
limits for FY82 and FY87 adipose tissue samples 127
9-7. Profile plots of average HxCDD concentrations with 95% confidence limits for
FY82 and FY87 adipose tissue samples 128
9-8. Profile plots of average 1,2,3,4,6,7,8-HpCDD concentrations with 95% confidence
limits for FY82 and FY87 adipose tissue samples 129
9-9. Profile plots of average OCDD concentrations with 95% confidence limits for
FY82 and FY87 adipose tissue samples 130
vu
-------
TABLES
Number Page
1-1. Summary of Studies Conducted to Determine Levels of PCDDs and PCDFs in
Human Tissues 10
2-1. Estimated National Average Concentrations of PCDDs and PCDFs from the
FY82 and FY87 NHATS 19
3-1. Summary of DQOs for PCDD/PCDF Analysis Effort 26
4-1. Summary of Sampling Strategy 31
4-2. Sample MSAs for Fiscal Year 1987 National Human Adipose Tissue Survey 33
4-3. Age, Race, and Sex Subquotas for each NHATS Collection Site within a
Stratum-FY87 Design 36
4-4. FY87 NHATS Collection Summary 38
4-5. FY87 NHATS Sample Sizes by Categories 39
5-1. Distribution of FY87 NHATS Composite Samples by Census Division and Age
Group 46
5-2. Demographic Makeup of FY87 NHATS Composite Samples 47
6-1. Internal Standard Spiking Solution for chlorinated Species 50
6-2. Concentration Calibration Solutions for PCDD/PCDF 53
6-3. Initial Calibration Data—Relative Response Factors 54
6-4. Typical Daily Sequence for PCDD/PCDF Analysis 55
6-5. Ions Monitored for HRGC/HRMS of PCDD/PCDF 56
6-6. HRGC/HRMS Operating Conditions for PCDD/PCDF Analysis 58
6-7. Quality Control Samples 62
6-8. PCDD and PCDF Native Spiking Solution 64
6-9. Control Sample Spike Levels 65
6-10. PCDD and PCDF Spike Check Results 66
V1U
-------
TABLES (continued)
Number Page
6-11. OCDD Spike Check Results 66
6-12. Results of Performance Audit Samples Analyzed With the FY87 NHATS
Composites 67
6-13. Interlaboratory Comparison—Spike Check Solution 68
6-14. Interlaboratory Comparison—Control Lipid Results (pg/g) 69
6-15. Interlaboratory Comparison—Low Level Spiked Lipid (% Recovery) 70
6-16. Interlaboratory Comparison—High Level Spiked Lipid (% Recovery) 71
6-17. Summary of QC Data for the FY87 NHATS PCDDs and PCDFs 73
6-18. Regression Models Used to Analyze FY87 NHATS QC Data 79
6-19. Estimated Recoveries for Each Batch and Estimated Average Recovery Over All
Five Batches 81
6-20. Predicted Concentration (C, pg/g) and Coefficient of Variation at Each Spike
Level 82
7-1. NHATS Analysis Factors and Categories 86
8-1. Frequency Counts of Data Restrictions and Data Qualifiers in the 48 Samples
for FY87 NHATS Composite Samples 94
8-2. Summary of Unrestricted Data for FY87 NHATS Composite Samples 96
8-3. Estimated Average Concentrations (pg/g) with Relative Standard Errors (%) for
Selected PCDDs and PCDFs from FY87 NHATS Composite Samples 97
8-4. Significance Levels from Hypothesis Tests for Differences Between
Demographic Groups for NHATS FY87 PCDDs and PCDFs 99
8-5. Estimated Rates of Change of Selected Dioxin and Furan Concentrations in
Human Adipose Tissue 103
8-6. R-Squared Correlation Between Predicted and Observed Concentrations for
FY87 Dioxins and Furans 105
rx
-------
TABLES (continued)
Number Page
9-1. Marginal Comparisons of FY82 and FY87 NHATS Individual Specimens Used
for PCDD and PCDF Analysis 109
9-2. Marginal Comparisons of FY82 AND FY87 NHATS Composite Designs Ill
9-3. Number of Composite Samples and Types of Statistical Comparisons Made
Between FY82 and FY87 NHATS Results 117
9-4. Comparisons of LODs and Percent Detected for PCDDs and PCDFs Between
NHATS FY82 and FY87 118
9-5. Comparisons of NHATS FY82 and FY87 Predicted National Average
Concentrations of PCDDs and PCDFs in Human Adipose Tissue 120
9-6. Comparisons of NHATS FY82 and FY87 Significance Levels from Hypothesis
Tests for Differences Between Demographic Groups for Selected PCDD and
PCDF Levels 123
9-7. Results of Profile Analysis Comparing FY82 and FY87 Predicted Concentrations
of Selected PCDDs and PCDFs in Human Adipose Tissue - By Census Region .... 131
9-8. Results of Profile Analysis Comparing FY82 and FY87 Predicted Concentrations
of Selected PCDDs and PCDFs in Human Adipose Tissue - By Age Group 132
9-9. Results of Profile Analysis Comparing FY82 and FY87 Predicted Concentrations
of Selected PCDDs and PCDFs in Human Adipose Tissue - By Race Group 133
9-10. Results of Profile Analysis Comparing FY82 and FY87 Predicted Concentrations
of Selected PCDDs and PCDFs in Human Adipose Tissue - By Sex 134
9-11. Weighted Average Concentrations and Standard Errors (SE) of Selected PCDFs
from FY82 and FY87 NHATS 138
9-12. Arithmetic Averages (pg/g), Standard Errors, and Number of Samples for
Selected PCDDs Obtained from the VA/EPA, FY82 NHATS (15-44 Age Group),
and FY87 NHATS (15-44 Age group) Studies by Collection Year Category 139
9-13. Arithmetic Averages (pg/g), Standard Errors, and Sample Sizes for Selected
Analytes Obtained from the VA/EPA, FY82 NHATS (15-44 Age Group), and
FY87 NHATS (15-44 Age Group) Studies 141
-------
GLOSSARY
DDE Design and Development Branch
BED Exposure Evaluation Division
EPA Environmental Protection Agency
FSB Field Studies Branch
FY Fiscal Year
HpCDD Heptachlorodibenzo-p-dioxin
HpCDF Heptachlorodibenzofuran
HpCDPE Heptachlorodiphenylether
HxCDD Hexachlorodibenzo-p-dioxin
HxCDF Hexachlorodibenzofuran
HxCDPE Hexachlorodiphenylether
IQS Internal Quantitation Standard
MRI Midwest Research Institute
MSA Metropolitan Statistical Area
NHATS National Human Adipose Tissue Survey
OCDD Octachlorodibenzo-p-dioxin
OCDF Octachlorodibenzofuran
OTS Office of Toxic Substances
PCBs Polychlorinated Biphenyls
PCDD Polychlorodibenzo-p-dioxin
PCDF Polychlorodibenzofuran
PCDF Polychlorinated dibenzofuran
PeCDD Pentachlorodibenzo-p-dioxin
PeCDF Pentachlorodibenzofuran
RS Recovery Standard
TCDD Tetrachlorodibenzo-p-dioxin
TCDF Tetrachlorodibenzofuran
TSCA Toxic Substances Control Act
VA Veterans Administration/U.S. Department of Veteran Affairs
XI
-------
EXECUTIVE SUMMARY
BACKGROUND
The National Human Monitoring Program (NHMP), operated by the United States
Environmental Protection Agency's Office of Toxic Substances (USEPA/OTS), is an ongoing
national program to monitor the human body burden of selected chemicals. The main
operative program of the NHMP is the National Human Adipose Tissue Survey (NHATS).
The NHATS is an annual survey to collect and analyze a nationwide sample of adipose tissue
specimens from autopsied cadavers and surgical patients. The purpose of the NHATS is to
identify and quantify the prevalence and levels of selected compounds in human adipose
tissue. The analysis results are used to establish an exposure-based chemicals list and to
estimate baseline levels and trends of the selected chemicals.
In the past, NHATS data have been used to monitor levels of selected organochlorine
pesticides and polychlorinated biphenyls (PCBs) in the United States population. However, in
Fiscal Year 1982 (FY82) the specimens collected under NHATS were used in the Broad Scan
Analysis Study, in which composited NHATS specimens were analyzed using high resolution
gas chromatography/mass spectrometry (HRGC/MS) in order to expand the list of target
analytes. The Broad Scan Analysis Study identified 17 volatile organic compounds, 30 semi-
volatile organic compounds, 5 polychlorinated dibenzo-p-dioxins (dioxins or PCDDs), and 5
polychlorinated dibenzofurans (furans or PCDFs). The Broad Scan Analysis Study
demonstrated that the PCDDs and PCDFs could be detected in the general U.S. population
across all geographic regions and age groups. Using a similar study design, EPA analyzed the
FY87 NHATS samples for PCDDs and PCDFs.
This report presents the objectives, methodology, and results of the FY87 NHATS
samples and compares the FY87 NHATS results with results from the analysis of the FY82
NHATS samples and from a related study of PCDDs and PCDFs conducted jointly by the U.S.
Veterans Administration and EPA's Office of Toxic Substances (VA/EPA).
OBJECTIVES
The specific objectives of the FY87 NHATS study were to:
• Identify the PCDD and PCDF isomers (specifically those with chlorine
substitution in the 2,3,7,8-position) that are present in human adipose tissue,
• Estimate the average concentrations of PCDDs and PCDFs in the adipose tissue
of humans in the U.S. population and various demographic subpopulations,
-------
Determine if any of the demographic factors (geographic region, age, race, and
sex) are associated with the average concentrations of PCDDs and PCDFs in
human adipose tissue, and
Compare the estimated average concentration levels of PCDDs and PCDFs
found in the FY87 NHATS with estimates from the FY82 NHATS and the
VA/EPA studies.
APPROACH
Population estimates were based on data obtained from the chemical analyses of 48
composite samples. The composite samples were prepared from 865 individual specimens
collected in a statistically designed survey of autopsied cadavers and surgical patients. A
statistical design was also used in creating the composite samples. HRGC/HRMS was used to
measure the concentrations of seven PCDDs and 10 PCDFs in the composite samples. The
resulting data were statistically analyzed using a model developed specifically for the
composite samples to estimate average concentration levels in the U.S. population and to
determine if any of the four demographic factors (geographic region, age, race, and sex) are
associated with the average concentration levels.
Traditionally, one of the objectives of NHATS has been to estimate the prevalence of
chemicals in the U.S. population as well as in various demographic subpopulations. More
generally this involves characterizing the distribution (i.e., variability of body burden levels
among individual donors. While these are still important objectives for NHATS, it was
decided that the FY87 survey would focus available resources on estimating average
concentration levels in the U.S. population and characterizing differences among various
demographic subpopulations. With the more focused objectives it became more efficient to
combine individual specimens into composite samples prior to chemical analysis.
The principal advantage of compositing is that it reduces the number of samples that
need to be chemically analyzed. However, this does not necessarily result in a corresponding
reduction of precision for the population estimates. With composite samples, the effects of
sampling error on the precision of average concentration estimates are substantially reduced
because the concentration in a composite sample is the average of individual specimen
concentrations.
The main disadvantage of using data from composite samples is that the prevalence
and distribution of chemical concentrations among individuals in the population cannot be
accurately estimated. However, it was possible to achieve all of the study objectives for the
FY87 NHATS as outlined above by using appropriate design criteria for combining specimens
into composites.
-------
The overall quality of results was ensured by using statistical sampling and
compositing designs, implementing quality control procedures, validating analytical and
statistical methodologies, and adhering to data quality objectives. Data quality objectives
(DQOs) were established for target ion ratios, percent recovery of target analytes, and the
presence of chemical interferences. Only data that met all DQOs were included in the
statistical summaries and analyses.
SUMMARY OF THE FY87 NHATS RESULTS
National Weighted Average Concentrations
The estimated national average concentrations of the target analytes are presented
graphically in Figure ES-1. The 1,2,3,4,7,8- and 1,2,3,6,7,8-HxCDD isomers have been reported
together due to incomplete chromatographic separation. Also, the percent of composite
samples in which the compound was detected is listed for each target compound. Data from
all of the target PCDDs and three of the 10 PCDFs were analyzed using a statistical modeling
approach. Population estimates were calculated using weights determined from the 1980
census counts. Based on the 95% confidence intervals the maximum uncertainty of the
estimated average concentrations is 12% for the six PCDDs and 26% for the three PCDFs. The
estimates for the remaining PCDFs were calculated using weighted averages of the measured
composite concentrations. Weights were determined from the 1980 U.S. Census counts. The
seven remaining chemicals (PCDFs) did not meet minimum criteria to perform a more detailed
statistical analysis. Several of these compounds were not detected in a sufficient number of
composite samples to allow meaningful statistical analysis.
Age Effects
Each composite was prepared with specimens from one of the three age groups (0-14,
15-44 and 45+ years). The average concentrations of the nine PCDDs and PCDFs that were
statistically analyzed were found to increase significantly with the age of the donor. For
example, the average concentration of 2,3,7,8-TCDD increased from 1.98 pg/g in the youngest
age group (0-14 years) to 9.40 pg/g in the oldest age group (45+ years) — a 375% increase.
Increases attributed to age effects in the other chemicals ranged from 24% for 2,3,7,8-TCDF to
863% for 2,3,4,7,8-PECDF.
-------
CD <
O) ^
CO .£
05 c CO
^ 2 ^
•si *
c c
O CD
B 8
2- O
o
3] 5]
OOi-OCM
c co r^ Tr
-^-
in
51 5] Jo]
COCO
co fv» c\i
T- r^ T- ••- CM
p
o
o
o
o
p
o
o
o
o
: o 2 —
c
g
03
C
CD
O
C
o
O
O
o
o
o
O5COO5^—
CO
TJ
C
2
O
Q.
c
O
O
Q
O
o
1
CO
jC
en"
CM
g
O
I
1
CO
i-C
en"
CM
^
g
O
£
i
CO
rC
co"
en
CM"
,_"
CO
N."
g
O
5
i
CD
CO
N."
en
c\T
,_"
g g
0 g
Q. U
I O
1
CO
rv."
CD"
TT
cn
CM"
ft fe
Q o
O co
H Q.
1 1
CO CO
iC K
en cn_
CM" CM
*^
LL
O
o
1
CO
rC
^•"
cn"
CM"
fe
0
X
T
CO
rC
^•"
en"
CM"
^_"
fe
O
1
CO
tvT
to"
en"
CM"
^_*
^
O
X
1
CO
rC
co"
•qr
en"
CM"
LL
Q
0
1
O5
co"
rs."
en"
CM"
^_"
^
O
X"
1
00
rC
CO*
rr"
en"
CM"
LL
o
£•
1
OJ
co"
(C
^
eo"
CM
CO
LL
O
O
o
2
TJ
0)
to
O
CO
_QJ
75
175
UJ
CO
^J
Q.
ro
CO
a>
o
O
0)
e
0)
Q.
<
u-
o
Q
U
c-
13
c
re
U2
D
Q
U
c
o
x:
re
u.
^-*
I
O
u
cu
tc
re
re
I
re
2
3
bC
-------
Geographic Effects
There were significant differences in the estimated average concentrations of 2,3,4,7,8-
PeCDF among the four geographic regions (north central, northeast, west, and south). The
estimated average concentration in the west region was 4.49 pg/g compared to an average of
13.7 pg/g in the northeast and a national average of 9.70 pg/g. The data combined with the
FY82 NHATS data also suggest the possibility of regional effects on the average concentrations
of 2,3,7,8-TCDD and 1,2,3,4,6,7,8-HPCDD, but the estimated regional differences were not
statistically significant.
Race and Sex Effects
The differences in estimated average concentrations between Caucasian and non-
caucasian donors and between male and female donors were not statistically significant for
any of the modeled compounds.
COMPARISON WITH FY82 NHATS
The FY87 NHATS estimated national average concentrations of 2,3,7,8-TCDD,
1,2,3,4,6,7,8-HPCDD, and OCDD were consistent with the estimates established in the FY82
NHATS. However, the FY87 estimates for many of the other PCDDs and PCDFs were
significantly lower than the estimates obtained in the FY82 NHATS. It is likely that these
differences are due in part to advances in the analytical method between FY82 and FY87.
COMPARISON WITH THE VA/EPA STUDY
The VA/EPA study was a retrospective study which used surplus specimens from the
NHATS repository to compare PCDD and PCDF concentrations in the adipose tissue of
Vietnam veterans with non-Vietnam veterans and civilians. A total of 197 specimens collected
between 1971 and 1982 were analyzed individually using analytical methodology equivalent to
the FY87 NHATS study. All specimens in the VA/EPA study are within the middle age group
(15-44 years) and hence allowed comparison to the same age group evaluated in the FY82 and
FY87 NHATS studies. The estimates obtained in the VA/EPA study were two to three times
higher than the FY87 NHATS estimates. The average values of the compounds across
collection years (1971 to 1987), particularly 2,3,7,8-TCDD, indicate a decrease in adipose tissue
residue levels. These differences may be due to a decrease in exposure levels through
environmental pathways or consumer products over time or as a result of the differences in
storage times between the VA/EPA (up to 16 years) and FY87 NHATS (up to 2 years) studies.
Further studies or analyses are needed to resolve these differences.
-------
1.0 INTRODUCTION
1.1 BACKGROUND
The National Human Adipose Tissue Survey (NHATS) is the main operating program of
EPA's National Human Monitoring Program (NHMP). Under the NHATS program, human
adipose tissues are analyzed to monitor human exposure to potentially toxic compounds. The
adipose tissue specimens were collected according to statistically designed sampling plans. The
tissues are collected by cooperating pathologists and medical examiners during routine
examination of tissues that have been excised during post mortem examinations or for therapeutic
reasons during elective surgeries. The tissues are collected based on a statistically developed
design of Metropolitan Statistical Areas (MSAs) to provide chemical residue information that can
be correlated to demographic data (geographic region, age, sex, and race).
The NHATS program was established in the late 1960's. The chemical residues of interest
during that time frame were organochlorine pesticides and PCBs. Recognizing the need to extend
the capabilities of the NHATS program, the Office of Toxic Substances (OTS) initiated a series of
programs in 1984 to expand the utility of the tissue repository. Foremost in these analysis efforts
was the conduct of a study termed the "Broad Scan Analysis Study." The broad scan analysis
included the determination of volatile and semivolatile organic compounds based on methods
requiring mass spectrometry as a detection device. Included in the semivolatile organic
compounds were the poly chlorinated dibenzo-p-dioxins (PCDDs) and the dibenzofurans (PCDFs).
The data from the broad scan analysis effort demonstrated that the PCDDs and PCDFs were
prevalent in the adipose tissues of the general U.S. population from all census divisions and
within each age group.
1.2 EMPHASIS ON PCDDs AND PCDFs
The ubiquity of the PCDDs and PCDFs in adipose tissues provides evidence of widespread
exposure to a class of potentially toxic compounds. On the basis of animal studies, the U.S. EPA
considers 2,3,7,8-TCDD to be one of the most potent known carcinogens studied (U.S. EPA 1987,
U.S. EPA 1989). Studies on the toxicity of other PCDDs and PCDFs have demonstrated the
compounds with chlorine substitutions in the 2,3,7,8-ring positions results in the greatest toxicity.
In addition, it has been demonstrated that toxicity varies with the degree of chlorination.
Potential for exposure to PCDDs and PCDFs exists from multiple sources including:
• Diet, particularly fish, poultry, meats, and dairy products that are impacted by
environmental releases of PCDDs and PDCFs.
-------
• Incinerator emissions (municipal, hospital and hazardous waste incinerators,
automobiles, etc.),
• Halogenated commercial products such as specific herbicides (2,4-D, Agent Orange)
or wood preservatives (pentachlorophenol), and
• Bleached paper products and the effluents from pulp and paper bleaching
operations.
• Combustion of halogenated aromatics, e.g., PCB transformer fires.
• Metal smelting and reclamation processes.
The PCDDs and PCDFs are environmentally persistent, and the most toxic isomers, the
2,3,7,8-substituted compounds, bioaccumulate in the food chain. EPA has studied the prevalence
of these compounds extensively in the general environment through the various programs
conducted as part of the National Dioxin Study, and in support of investigations at potential
hazardous waste sites. The current studies involving the pulp and paper industry resulted from
data generated from fish residue data from.the National Dioxin Study. In addition, several states
(e.g., California, Vermont, Connecticut, and Mississippi) have conducted monitoring programs to
determine background levels of PCDDs and PCDFs in ambient air, soil, sediment, and fish in
areas that are potentially impacted by emissions from incineration sources or effluents from the
pulp and paper industry (Stanley et al. 1989a, Heil and Fitzgerald 1987, State of Connecticut 1986,
personal contacts).
EPA has promulgated regulations to specifically address the releases of PCDDs and PCDFs
to the environment. Under Sections 4 and 8 of the Toxic Substances Control Act (TSCA), the
manufacturers of specific halogenated products are required to determine and report the levels
of halogenated dioxins and furans in those products (U.S. EPA 1987). EPA has also recently
developed guidelines for establishing test procedures for release of halogenated dioxins and furans
in industrial effluents (USEPA 1991), particularly from pulp and paper bleaching facilities.
The NHATS program provides EPA with a unique capability to monitor the impact of
changes in environmentally persistent chemicals. As an example, the NHATS program has
demonstrated a decline in the adipose tissue levels of PCBs in the general U.S. population since
the restriction and regulation of use and disposal of PCBs in the mid 1970s (Robinson et al. 1990).
The NHATS samples should also provide data on the result of efforts to reduce human exposure
to PCDDs and PCDFs.
-------
1.3 RELATED STUDIES
Over the past 10 years, considerable attention has been given to the determination of
PCDDs and PCDF residues in human tissues. Many of these studies are identified and
summarized in Table 1-1. Initial studies focused on body burden levels of individuals who were
potentially exposed through occupational exposures (for example, Vietnam veterans to the
herbicide Agent Orange) (VA/EPA1991, Kang et al. 1991, CDC 1987, CDC 1988, CDC 1988a, Pirkle
et al. 1989, Schecter 1987, Kahn 1988, Kahn et al. 1988, Kahn et al. 1990, Gross et al. 1986),
improper handling of contaminated wastes (Times Beach, Missouri) (Patterson et al. 1986,
Andrews et al. 1989, Patterson et al. 1987a, Patterson et al. 1989), contamination of food products
(Yusho and Yucheng) (Ryan et al. 1987), and accidental releases from chemical production
facilities (Seveso, Italy) (Fachetti et al. 1981, CDC 1988c).
Several studies have been conducted to determine the levels of PCDDs and PCDFs in the
tissues of the general population both within the United States and internationally. Table 1-1
provides a summary of the studies that have been reported to date on the residue levels of these
compounds. The studies conducted in the United States include a limited number of samples
from Binghamton, New York (Schecter et al. 1986); Atlanta, Georgia (Patterson et al. 1986); Salt
Lake City, Utah (Patterson et al. 1986); St. Louis, Missouri (Graham et al. 1986a, 1986b); the State
of Missouri (Patterson et al.); and the State of California (Stanley et al. 1989). The studies
conducted in Binghamton, St. Louis, and the State of California focused on the determination of
total PCDDs and PCDFs. The other studies focused on 2,3,7,8-TCDD only.
1.4 DETECTIVES OF THE FY87 STUDY
The specific objectives of the FY87 NHATS study were to:
• Identify the PCDD and PCDF isomers (specifically those with chlorine substitution
in the 2,3,7,8 position) that are present in human adipose tissue,
• Estimate the average concentrations of PCDDs and PCDFs in the adipose tissue
of humans in the U.S. population and various demographic subpopulations,
• Determine if any of the demographic factors (geographic region, age, race, and sex)
are associated with the average concentrations of PCDDs and PCDFs in human
adipose tissue, and
• Compare the estimated average concentration levels of PCDDs and PCDFs found
in the FY87 NHATS with estimates from the FY82 NHATS and the VA/EPA
studies.
-------
c
"S
P^
S
» -S
n ^J
"^ s
g
n W
a J
£
W
Cu
H!
§
UM
O
"«3
& i
u x:
r! frl
es Conducted to Determil
Exposure classific
•3
i2 "B,
^ S
fc> 8
I ^
G o
a 2
en
1 fg
H 8-5
"o J
•S "8
(0
g
J-
d.
i
(U
0
5 ol
*3 c? '*"**'
^* S ^^ «J
xg; ^ "5
•^ > 3 WJ
"^ >3
0)
8 S
1 1
W M
BH BH
D D
Q D
£ kl
a
J
1
2
General population comf
763 individual specimens
Civilian
Non- Vietnam veteran
Vietnam veteran
*> g <* >8
^
ii "o ii ij a
£%££ o
eg §
1* 1
UH
D
O O
D B
y fc
1 1
s 60 e-
General population. Che
production work handlin
recreational residential e:
General population
3 " m
1-1
S
in g
s
CJ «• TT? U5
§ to "3 ^ 3 n
5 j2 *i ,&- § £
i«5 v — C/5 t/J •<
I
I
en
a
c
o
•
3
O.
»
y **
-S EH
5 2 s ;z
u u
II
1* o
t/5 t/5
OD
Q Q
&•
«
p
|£^ e
D
U
O D Q Q
Q D n D
o o o
B, B, pt,
£ (0 03 td
C X X J3
c tu v w
0) t>o 60 60
t> c e -
> S E
goo
O
10
-------
f
a
c
•a
c
a
*i*
^
2
Reference
•1
(d
a
V
i
.S
« -g
ll
1 B
c
o
•J3
5
vC
"S
(d
"u
3
1
S.
(A
"S.
6
^
c>
2
OJ
"5.
B c
a 0
« •&
•s JS
i"5
to
(0
"s
IE
a.
i
OJO
U
OO O
5?
t-H ri
(d td
•S t!
c c
.c .c
**
rt.
S E
cL §
^ §
< cr»
in «
SS
u u
Q Q
OD
££
yl_^
>~,
~
c
S
9) us
> C
B il
| |
^ £
l>-
t>. m
a
Q
c
u
O
5 i
0) O
$ &
p *JS
u «
o> q» a»
JG 5fi s
CL UH UH
III
UH
D
p n D
O O Q
¥¥X
If
s€ &>
c a |
E o <
u «. .
•3 .b «?
> «? D
"* m >S
18 s "
1
c
(U
" ^
"is tl
1? n -jrj u _f a
I * B S "*
^ I 8,11-1:
«s I" lllJ-
•
!•
5 «« 5
mi ii m
ss
u u
1
§ §
JS
a.
ts
c
O E
H ffi S
E^crb^
B B B B
"u "u u *u
(0 « (0 W
tu u. u. u.
01
O
is
0)
"5.
B
T3
c
•a -a
ta ta
c c
•c
v
>
11
-------
1.5 REPORT ORGANIZATION
This report presents all pertinent information with respect to the assessment of data
quality, the statistical design, the conduct and results of the analytical procedures, the statistical
analysis procedures and extrapolations to population estimates, and the comparability of the FY87
data set to the previous PCDD and PCDF analysis programs conducted using the NHATS adipose
tissue samples.
Section 2.0 (Summary and Recommendations) presents a synopsis of the significant
findings from the statistical analysis of the FY87 data set in comparison to the previous studies
conducted through the NHATS program. Overall data quality issues with respect to analytical
and statistical criteria are discussed in Section 3.0. The study design for collection of individual
specimens in the FY87 collection program is described in Section 4.0, and the compositing design
and procedures for preparing the composite samples are discussed in Section 5.0. Section 6.0
describe the analytical procedures used to determine of PCDDs and PCDFs and summarizes the
supporting analytical quality control (QC) data. The statistical methodologies are described in
Section 7.0, and the results of the statistical analysis are presented in Section 8.0. Section 9.0
focuses on the comparisons of the FY87 study with the FY82 and the VA/EPA collaborative study.
Study designs, analytical and statistical methodologies, and graphical comparisons of the FY82 and
FY87 data are covered.
Supporting information on the individual sample data, the detailed analytical protocol,
measures of uncertainty, and plots of QC data are presented in Appendices A through E.
12
-------
2.0 SUMMARY AND RECOMMENDATIONS
The human adipose tissue specimens analyzed in the FY87 NHATS were collected from
October 1986 through September 1987, following a statistically based survey design. A statistical
design was also used to composite the specimens prior to chemical analysis. Section 2.1 presents
a summary of the FY87 NHATS results. The results include estimated national average
concentrations of PCDDs and PCDFs in human adipose tissue and the effects of the geographic
region, age, race, and sex on the average concentrations. The statistical comparison of the FY82
and FY87 NHATS results and a descriptive comparison with the EPA/VA results are summarized
in Section 2.2. Recommendations for future studies are presented in Section 2.3.
2.1 SUMMARY OF FY87 NHATS RESULTS
Forty eight (48) adipose tissue samples analyzed in the FY87 NHATS were prepared using
865 individual specimens collected in a statistically designed survey of autopsied cadavers and
surgical patients. The composite samples were analyzed for PCDDs and PCDFs using high
resolution gas chromatography/mass spectrometry (HRGC/HRMS). The resulting data were
analyzed using a statistical model developed for composite samples to estimate average
concentration levels and to determine if any of the four demographic factors (geographic region,
age, race, and sex) are associated with the average concentration levels.
Traditionally, one of the objectives of NHATS has been to estimate the prevalence of
chemicals in the U.S. population as well as in various demographic subpopulations. More
generally this involves characterizing the distribution (i.e., variability) of body burden levels
among individual donors. While these are still important objectives for NHATS, it was decided
that the FY87 survey would focus available resources on estimating average concentration levels
in the U.S. population and characterizing differences among various demographic subpopulations.
With the more focused objectives it became more efficient to combine individual specimens into
composite samples prior to chemical analysis.
The principal advantage of compositing is that it reduces the number of samples that need
to be chemically analyzed. However, this does not necessarily result in a corresponding reduction
of precision for the population estimates. With composite samples the effects of sampling error
on the precision of average concentration estimates are substantially reduced because the
concentration in a composite sample is the average of individual specimen concentrations.
The main disadvantage of using data from composite samples, however, is that the
prevalence and distribution of chemical concentrations among individuals in the population
cannot be accurately estimated. But, it was possible to achieve all of the study objectives for the
13
-------
FY87 NHATS as outlined above by using appropriate design criteria for combining specimens into
composites.
Because of criteria based on pre-determined data quality objectives (DQOs) the number
of composite samples used in the data analysis varied among the target compounds. Several of
the compounds, particularly the PCDFs, were not detected in a sufficient number of composite
samples to perform a reliable statistical analysis. This generally occurred for the compounds that
were detected at levels near or below the method detection limits. Interferences which precluded
accurate measurement, were encountered for two of the HxCDF isomers. Thus, only nine of the
target analytes met minimum criteria for applying the statistical model. However, summary
statistics are provided for all target chemicals. The criteria for applying the statistical model were:
(1) At least 30 composites must meet all DQOs for chemical analysis, and
(2) The chemical must be detected in at least 50% of the composites meeting the
DQOs.
2.1.1 Population Estimates
Figure 2-1 presents the estimated national average concentrations for the target chemicals.
Because of wide range of concentration levels among the target chemicals, the estimates are
displayed on a log scale. The estimates were calculated using either a statistical modeling
approach or by taking weighted averages of the composite sample concentrations. The weights
used in the weighted average estimates are based on 1980 U.S. Census population counts.
Further discussion of this approach is presented in Section 9.4.
For the nine chemicals that were statistically analyzed, the error in estimating the average
concentration levels, based on 95% confidence limits, was less than 12% for the PCDDs and less
than 26% for the PCDFs. The quality of these estimates is supported by validation of both the
chemical analysis method and the statistical approach to handling the data and the performance
of the method as demonstrated from the results of the Quality Control Samples.
The estimates for the chemicals, particularly PCDFs, that were detected in less than 50%
of the samples may be significantly affected by the detection limits of the analytical method.
Whenever the chemical was not detected in the sample, it was assumed that the concentration
was below the limit of detection (LOD) of the analytical method. Thus, the data value was
"censored" and no measured concentration was reported. To calculate an average concentration
using censored data, the value of LOD/2 was used in place of a measured concentration
whenever the chemical was not detected. This approach minimizes the potential error in the
estimated average concentration. But when more than 50% of the data are censored there is
14
-------
CD <•.
cn o
CO £;
C CO
O CO
co
o CM co
SI
3] JQ| 51
coco
(A
JO
fo
to LO
co .b
c c
3 §
Z8
O Tf
T- CVJ
1- h-
q
o
o
o
o
q
CD
o
o
o
o
o
T3
JD
"o
Q
T3
C
O
Q.
O
O
o
o
1^ CO
O
1 § g
1
a
eo"
CM
1
ao
co"
^J"
1
00
N.
ec"
co
CM
1
ffl
00
NT
WI
1
eo
-------
much more uncertainty in the estimated average. The maximum systematic error due to
censuring is the average LOD divided by two.
2.1.2 Demographic Comparisons
Statistical hypothesis tests were performed for each of the nine modeled compounds to
determine if there were statistically significant differences in average concentration levels among
individuals from different census regions (north central, northeast, west, and south), age groups
(0-14 years, 15-44 years, and 45+ years), race groups (Caucasian, non-caucasian), and sexes (male,
female). Statistical conclusions derived from these analyses are summarized below.
Age Effects. Concentration of PCDDs and PCDFs increase with age. The differences in
the average concentrations among age groups was statistically significant at the 0.05 level for all
nine of the modeled chemicals. Figure 2-2 displays the average concentrations by age group for
each of the modeled chemicals. The highest average concentration was always found in the
oldest age group (45+ years) and, except for 2,3,7,8-TCDF, the lowest was found in the youngest
age group (0-14 years). The increases from the lower to the upper age groups range from 24%
for 2,3,7,8-TCDF to 863% for 2,3,4,7,8-PeCDF.
Geographic Effects. One of the nine modeled chemicals was found to have statistically
significant differences among the estimated average concentrations from different census regions.
The estimated average concentration of 2,3,4,7,8-PeCDF in the Western census region was 4.49
pg/g compared to an average of 13.7 pg/g in the North East region. The national average was
9.7 pg/g. No other chemical for the FY87 composites was found to have significant geographic
effects at the 0.05 level of significance. This geographic effect for the 2,3,4,7,8-PeCDF was also
detected in the analysis of the data for the FY82 NHATS samples.
Race Effects. The difference in estimated average concentrations between Caucasian and
non-caucasian donors was not found to be statistically significant for any of the modeled
compounds.
Sex Effects. The difference in estimated average concentrations between male and female
donors was not found to be statistically significant for any of the modeled compounds.
2.1.3 Quality Control Results
Twenty (20) quality control (QC) samples were analyzed along with the 48 study samples.
Each of the five analysis batches contained eight to ten study samples, a method blank, an
unspiked control sample, and two spiked samples—one at a low spike level and one at a high
16
-------
Compound Age Group (
2,3,7,8-TCDD 0-14
15-44
45+
1,2,3,7,8-PeCDD 0-14
15-44
45+
1, 2,3,4,7,8/1 ,2,3,6,7,8-HxCDD 0 - 14
15-44
45+
1,2,3,7,8,9-HxCDD 0-14
15-44
45+
1,2,3,4,6,7,8-HpCDD 0-14
15-44
45+
OCDD 0-14
15-44
45+
0.
2,3,7,8-TCDF 0-14
15-44
45+
2,3,4,7,8-PeCDF 0-14
15-44
45+
1,2,3,6,7,8-HxCDF 0-14
15-44
45+
0.1
National Average
yrs.) Concentration (pg/g)
888888 1.98
4.37
£8888888888! 9-10
£8888888$ 3.30
£888888888^ 9.33
£88888888888$ 18.20
£8^88888888881 23.4
£8^8888888^8^88^ TO-S
£88888888^$$8$88I 12°-
£888888888 6.13
£8888888888 1"
£88888888888$ 17-1
£8888888888888^ 45.7
£888888888888888^ 993
£88888888888888888 174.
£88888888888888888< 215.
£88888888888888888888 592.
g8$££888$££8888$£8888888< nso.
10 1.00 10.00 100.00 1000.00 10000.0
Concentration (pg/g)
^8888883 1Q7
^88888^ 1.45
^8888888 2«
^888888^ 1.R7
£888888888? a.oo
£888888888888 IB.O
£$888883 ion
$8888888$ 459
^888888888$ 10^
K> 1.00 10.00 100.00 1000.00 10000.0
Concentration (pg/g)
Figure 2-2. Estimated averaged concentrations by age group for selected PCDDs and PCDFs
from the FY87 NHATS.
17
-------
spike level. The QC samples were chemically analyzed to characterize the accuracy and precision
of the analytical method and to determine if there were significant batch effects. The statistical
analysis of the QC data demonstrated that batch effects were relatively small and that the method
recovery criteria were met for 97.5% of the target analytes.
2.2 COMPARISON OF FY87 NHATS WITH OTHER STUDIES
The results from the FY87 NHATS were compared with results from two previous studies
involving specimens from the NHATS repository. Section 2.2.1 discusses the comparison with the
FY82 NHATS Broad Scan study and Section 2.2.2 discusses the comparison with a collaborative
study conducted between the Veterans Administration and the U.S. EPA (VA/EPA).
2.2.1 Comparison with FY82 NHATS
The study designs for FY82 and FY87 NHATS were very similar. Although there are minor
differences in the number of specimens collected and their distribution among demographic
populations, these differences do not significantly affect the comparison of results. On the other
hand, there were substantial improvements in the analytical method between the FY82 and FY87
NHATS. Additional internal quantitation standards were added at the sample extraction step and
carried throughout the analysis procedure. These additional internal quantitation standards
provide a better characterization of method performance and more accurate determination of the
PCDDs and PCDFs via the isotope dilution principle.
Statistical comparisons were performed to determine significant differences for the specific
compounds between the FY82 and FY87 NHATS studies. Differences noted in the measured
concentrations may be due to either actual changes in body burden levels or changes in the
analytical method. In addition to comparing national average concentrations, a more detailed
profile analysis was performed to determine if the demographic effects are consistent between
FY82 and FY87. The profile analysis provides meaningful results even if there are systematic
differences in measured concentration levels.
Table 2-1 compares the estimated national average concentration levels for selected
chemicals analyzed in FY82 and FY87. Also presented are the relative changes from FY82 to FY87.
For example, the estimated concentration of 2,3,7,8-TCDD is 9% lower for FY87 than determined
for FY82. Statistical comparisons were possible for the PCDDs and only one PCDF, the
2,3,4,7,8-PeCDF. The other PCDFs could not be compared statistically because one or both of the
estimates are based on data which did not meet the minimum criteria for statistical modeling.
18
-------
Table 2-1. Estimated National Average Concentrations of PCDDs and
PCDFs from the FY82 and FY87 NHATS
PCDDs
2,3,7,8-TCDD
1,2,3,7,8-PeCDD
HxCDD
1,2,3,4,6,7,8-HpCDD
OCDD
PCDFs
2,3,7,8-TCDF
2,3,4,7,8-PeCDF
HxCDF
1,2,3,4,6,7,8-HpCDF
OCDF
National
FY82
(Pg/g)
5.88
73.6 >b
122 >
142
768
39.7
35.4 >
20.9
20.6
56.0
average
FY87
(Pg/g)
5.38
10.7
86.8
110
724
1.86
9.70
14.2
15.3
2.28
Percent
change*
-9%
-85%
-29%
-23%
-6%
-95%
-73%
-32%
-26%
-96%
(FY87-FY82)
18 FY82 :
b The FY82 estimate is significantly higher (denoted by >) than the FY87
estimate at the 0.05 level of significance.
c No statistical comparisons were possible.
19
-------
All of the chemicals had higher estimated concentrations in FY82. However, among the
modeled chemicals, the differences in estimated concentrations between FY82 and FY87 were not
statistically significant for 2,3,7,8-TCDD, 1,2,3,4,6,7,8-HpCDD, and OCDD. These chemicals had
estimated relative changes of less than 23%. Some of the differences between FY82 and FY87 can
be explained by changes in the analytical method. This is evident in the fact that the detection
limits in FY82 were often higher than the estimated average concentration in FY87. For example,
the estimated detection limit for OCDF in FY82 was 19.0 pg/g. The average concentration
measured for the FY87 samples was 2.28 and the detection limit was 1.67 pg/g. In addition to
achieveing lower detection levels, the analysis of the FY87 NHATS composites was enhanced
through the use of additional internal quantitation standards that provided more accurate
measurements of the PCDD and PCDF residue levels. The effect of using fewer internal
standards in the FY82 anaysis may have resulted in over or under estimates of the concentrations
of the penta-, hexa-, and hepta-chlorinated PCDDs and PCDFs.
The profile analysis provided additional insight into the demographic effects on PCDDs and
PCDFs. As previously discussed, the FY87 data demonstrated that there were geographic effects
on the concentrations of 2,3,4,7,8-PeCDF. Although there was a significant difference in the
national average concentrations between FY82 and FY87, the FY82 data confirmed the existence
of geographic effects. In both years, the highest estimated concentrations of 2,3,4,7,8-PeCDF were
found in the northeast. The profile analysis also suggested that there are possible geographic
effects on the concentrations of 2,3,7,8-TCDD and 1,2,3,4,6,7,8-HpCDD. The profiles for FY82 and
FY87 were very similar with the lowest estimated concentrations in the west and south census
regions.
2.2.2 Comparison with the VA/EPA Study
The VA/EPA study was a retrospective study which used surplus specimens from the
NHATS repository to compare PCDD and PCDF concentrations in the adipose tissue of Vietnam
veterans with non-Vietnam veterans and civilians. The study was conducted using approximately
200 adipose tissue specimens collected between 1971 and 1982 from male donors between the ages
of 17 and 46. Each specimen was analyzed for the same PCDDs and PCDFs that were measured
in the FY87 composites.
Although there are differences between the VA/EPA study and the FY87 effort in terms of
study objectives, sampling plans, and size and composition of samples analyzed, it was expected
that the results for the middle age group (15-44) from the studies should allow comparison since
all analyses were conducted with equivalent methodology. The estimated concentrations in the
VA/EPA were two to three times higher than the corresponding FY87 NHATS estimates.
Although no formal statistical tests were performed on these data, the differences are clearly
20
-------
significant. The average values for the compounds across collection year, particularly 2,3,7,8-
TCDD, indicate a decrease in adipose tissue residue levels. These differences may be due to a
decrease in exposure levels through environmental pathways or commercial/consumer products
over time. Another possible explanation may be the result of differences in specimen storage time
between the VA/EPA (up to 16 years) and the FY87 NHATS (up to 2 years) studies.
2.3 RECOMMENDATIONS FOR FUTURE STUDIES
Several recommendations, along with rationale, are presented for consideration for future
studies on PCDDs and PCDFs in human adipose tissue. Although the recommendations are
presented with respect to human adipose tissue, they also apply to considerations for use of blood
(whole, serum, or plasma) as a means to study the body burdens of these compounds in the
general U.S. population.
Recommendation; Increase the number of composite samples and individual specimens
that are analyzed in future NHATS studies.
Rationale; The FY87 NHATS produced estimates of national average concentration levels
of nine PCDDs and PCDFs and demonstrated that the concentrations of these chemicals increase
significantly with the age of the donor. These conclusions were based on the chemical analysis
of 48 composite samples containing an average of 18 specimens each. Although there were
sufficient data to achieve these important objectives, the small number of composites analyzed
presents significant challenges to achieving other NHATS objectives. For example, it is very
difficult to estimate prevalence with composite sample data. Prevalence is established by
estimating the percent of the population with chemical concentrations above specified levels.
Estimating prevalence requires that individual specimens be analyzed to characterize the statistical
distribution of concentrations among individuals in the population.
Another reason for increasing the number of specimens is to detect smaller but important
differences in average concentration levels among geographic regions, race groups, and sexes that
cannot be identified statistically because of sampling and measurement errors. Statistical power
calculations will be needed to establish the optimal allocation of individual versus composite
samples to be analyzed.
Recommendation; Conduct an analysis of PCDDs and PCDFs in human adipose tissue
using a similar study design at intervals of one to five years.
Rationale; One of the objectives of NHATS is to estimate time trends in the levels of toxic
chemicals. Significant trends require data from at least three to five time periods before they can
21
-------
be demonstrated. Considering the promulgation of specific regulations to limit releases of these
compounds from a variety of sources (incineration emissions, pulp and paper effluents,
halogenated commercial products, etc) it is expected that further changes in residue levels may
be observed. Correlation of declining residue levels with the promulgation of the residue levels
over time may provide the Agency with a good measure of the efficacy of environmental
regulation.
Recommendation; Establish and characterize QC samples that are representative of human
adipose tissue levels.
Rationale; In order to ensure that an accurate assessment of data comparability can be
achieved from one analysis event to the next, it is essential that appropriate reference materials
are available for incorporation into the study design. Adipose tissue pools representative of the
three age categories could be prepared in sufficient quantity to allow multiple determinations in
subsequent NHATS analysis efforts. The QC pools could be stored under the same conditions
as the NHATS repository and could be used to address storage stability issues.
22
-------
3.0 OVERALL DATA QUALITY
The overall quality of results in the FY87 NHATS was achieved by performing several
critical activities: planning a well-designed study, validating methods in advance, controlling
implementation, verifying assumptions, and documenting procedures. At least one of these
activities was involved in every step of the study. Figure 3-1 shows the steps of the study and
illustrates how they are related.
The steps are divided into the following five stages:
• Sampling,
• Compositing,
• Chemical analysis,
• Data management and processing, and
• Statistical Analysis.
Each stage involved a design or a plan, a set of controls applied on implementation, and
verification and documentation procedures that were followed. Subsections 3.1 to 3.5 describe,
for each stage, the specific activities that were carried out.
3.1 SAMPLING
A complete discussion of the FY87 NHATS study sampling design is contained in
Section 4.0 of this report and has been fully documented (Panebianco DL, 1986a). The design
followed a multistage process in which (1) the conterminous 48 states were divided into
geographic strata, (2) Metropolitan Statistical Areas (MSAs) were randomly sampled within
each stratum, and (3) cooperators were chosen in each MSA to provide the specimens.
Finally, each cooperator was given quotas, derived from the 1980 U.S. Census, on the number
of specimens to collect from each age group, sex, and race group.
The cooperators were given guidelines on the types of patients and cadavers which can
contribute specimens to the program. The relevant donor information was recorded on
Patient Summary Reports (PSRs). Then, matching bar code labels containing sample IDs
were affixed to each PSR and the corresponding sample vial. Each PSR was then reviewed by
the laboratory responsible for sample collection to determine whether the cooperators
followed the guidelines in selecting appropriate donors.
23
-------
Data Quality
Objectives
QC Plan
Sample Design
Sample
Collection
Composite
Design
Compositing
Chemical
Analysis
/ Outlier
Analysis
|
\
PSR Data
7^
'ting / H
Compositing
Data
Analysis Plan
QC Data
Analysis
Statistical
Analysis
90-48 SEVslanflow 111900
Model
Validation
Results
Figure 3-1. Program steps designed to ensure overall program quality.
24
-------
The sampling design called for collecting 1377 individual specimens with specific quotas
for each combination of geographic region and donor age group, race, and sex. Forty one (41)
of the 47 MSAs specified in the sample design contributed a total of 956 specimens. Specimens
collected within the overall design quota for each MSA were designated as design specimens,
and the remaining specimens were designated as surplus. There were 865 design specimens
in the FY87 NHATS. As discussed in Section 4.0, the marginal percentages of design
specimens from donors in each geographic and demographic category were found to be
consistent with the sampling design.
3.2 COMPOSITING
Composite design criteria were developed to ensure that the data obtained from the
composites could be interpreted for the populations of interest and that the results could be
compared with results from the FY82 NHATS study. Section 5.0 discusses these design
criteria.
The composite design criteria specified that only specimens from the same Census
division and age group were to be composited together. Also, in FY87 special efforts were
made to create more composites containing specimens from donors of the same sex than was
achieved in FY82. For example, 24% of the FY82 composites were pure sex (either all males or
all females) composites compared with 64% of the composites prepared for the FY87 samples.
It was not always possible to form composites containing specimens from the same race
group.
The procedures for preparing the composite samples are presented in Section 5.2.
Quality assurance/quality control practices that were implemented during compositing were
focused on procedures to assure that the study design was accurately followed and reported
and that the individual specimens were composited in a manner that preserved sample and
specimen integrity and avoided contribution from laboratory background.
Battelle provided MRI the study design, which identified each individual specimen that
was to be added to form a single composite. A form was provided for each composite sample
and was used to document the specific compositing activities. This form included the
identification of each individual specimen and demographic information that was used to
confirm the identity of each specimen. All specimens were retrieved from the repository and
grouped according to the composite design. The specimens were examined to determine if
sufficient tissue was present to meet the composite design. The EPA work assignment
managers were notified of potential problems, which were resolved prior to proceeding with
the physical compositing.
25
-------
As the composite samples were prepared, the mass of individual specimens were
recorded on the composite forms. All composite forms were compiled as a report and
submitted to EPA and Battelle. Data from the compositing reports were keyed to a computer
file using double data entry. The data were stored in the FOCUS database system which was
also used to perform range and logic checks.
3.3 DEMONSTRATED LABORATORY QUALITY CONTROL
At the outset of the analysis effort for the FY87 NHATS composite samples for PCDDs
and PCDFs, specific data quality objectives were developed and included in the Quality
Assurance Project Plan. These data quality objectives were identified for calibration criteria
(relative response factors [RRFs]) for each analyte and internal standard, HRGC column
performance, HRMS resolution, and method performance based on the recoveries of internal
quantitation standards (IQS) and compounds spiked into a spiked internal QC sample.
Table 3-1 summarizes the performance achieved versus the specific criteria and data
quality objectives for the analysis of the FY87 NHATS composites.
Table 3-1. Summary of DQOs for PCDD/PCDF Analysis Effort
Criteria
Objective
Actual
RRF calibration
TCDD isomer separation
Mass resolution
Internal quantitation
standards (IQS)
Spiked internal QC
samples
±20%
TCDD/TCDF
±30% all other
<; 25% valley/-
valley
HRMS * 10,000
40%-150%
40%-150%
91% (378 of 416) of all RRF
factors within DQOs
Achieved for all analyses days
Maintained and documented all
analysis days
96% for all IQS standards
(585 of 612); 2.5% (17) of
deviation due to "C-OCDD;
1.5% (9) due to double-spiked
sample
97.5% of all measurements
within criteria (4 of 160
measurements outside DQOs)
26
-------
A QC plan was proposed by Heath (1988) to monitor and characterize the performance
of the analytical method. The plan called for the analysis of a method blank, control sample,
and two spiked samples in each of the five analysis batches. The control and spiked samples
were formed using a composite sample of a large number of surplus adipose tissue specimens.
The method blank was analyzed first to confirm that the sample analytes would not be
compromised by laboratory background or calibration standard carryover, then the control
and spiked samples were analyzed in random order along with the eight to 10 study samples.
The objective of the QC plan was to provide data that could be used to characterize the
method's precision and accuracy and to identify any significant batch effects.
The statistical analysis of the QC data involved modeling the measured concentrations
versus spike level. Estimates of background levels, batch effects, and method recoveries were
established for all of the target chemicals. Also, statistical hypothesis tests were conducted to
determine if there were significant batch effects or if the estimated recoveries were
significantly less than 100%.
Batch effects were accounted for by including their effect in the estimated measurement
precision. Precision, as determined by the coefficient of variation at the control sample
concentration level, was estimated to be 5% to 20% for the dioxin compounds and 13% to 48%
for the furans.
3.4 STATISTICAL ANALYSIS
Prior to conducting the statistical analysis, a formal analysis plan was prepared to define
the statistical objectives and to specify the methods and assumptions that would be used to
achieve these objectives. The primary objectives were to estimate average concentrations in
the U.S. population and in specified marginal populations, and to determine if there are
significant effects associated with geographic and demographic factors. Section 7.0 discusses
these objectives and describes the statistical analysis approach.
The statistical model and analysis approach used for the FY87 NHATS were developed
specifically for the unique problem of analyzing composite sample data. They were developed
and evaluated in a separate study by Orban and Lordo (1989) in which composite data were
simulated from actual specimen data obtained in the FY83 NHATS. Their study demonstrated
the validity of the analysis approach. They also developed and documented special computer
programs to implement the analysis.
Before the statistical analysis was conducted, the data were classified according to
restrictions imposed by the predetermined data quality objectives (DQOs). Only data which
27
-------
met all of the DQOs were included in the analysis. This ensured that the reported results are
not affected by biases attributed to the analytical method. Furthermore, the formal statistical
analysis to estimate the population averages was only performed when there were a
minimum of 30 unrestricted composite measurements, and the chemical was detected in at
least 50% of the samples. Nine of the target compounds met these minimum criteria for
performing the statistical analyses. Summary statistics are provided for the chemicals which
did not meet these criteria.
After fitting the statistical model to the data, the model assumptions were checked
through analysis of residuals (i.e., the difference between predicted and measured
concentrations), probability plots, and correlations. These analyses demonstrated that the
model assumptions were reasonable. Furthermore, as discussed in Section 8.0, eight of the
nine chemicals statistically analyzed had R-squared correlations exceeding 68%. That is, at
least 68% of the variability in the data could be explained by the assumed model.
28
-------
4.0 NHATS SAMPLE DESIGN
4.1 SAMPLING DESIGN
The human adipose tissue specimens analyzed in the FY87 NHATS were collected from
October 1986 through September 1987, following the NHATS sampling design. The NHATS
program uses a statistically based survey design to obtain adipose tissue specimens from
autopsied cadavers and surgical patients. Although the NHATS target population is the
general, noninstitutionalized U.S. population, the sampling population is limited to cadavers
and surgical patients due to the invasive nature of the process required to collect the adipose
specimens from living persons.
Each year approximately 800-1200 adipose tissue specimens are collected using a multi-
stage sampling plan. The 48 conterminous states are first stratified into geographical areas.
From the set of strata, a sample of Metropolitan Statistical Areas (MSAs) is selected with
probabilities proportional to MSA population. One or more cooperators (hospital pathologists
or medical examiners) are chosen from each MSA and asked to supply a specified quota of
tissue specimens. Each year an effort is made to retain the same MSAs and cooperators used
in previous years.
Cooperators are given overall quotas, as well as subquotas for supplying specimens in
categories defined by the donor's age group, race, and sex. The categories are
Age Group: 0-14 Years, 15-44 Years, and 45+ Years;
Race: Caucasian and non-Caucasian; and
Sex: Male and Female.
The subquotas are proportional to the 1980 U.S. census population counts for each sampling
stratum. The donors and tissue specimens are selected in a nonprobabilistic manner by the
cooperators.
Because the survey requires some divergence from strict probabilistic sampling, the
validity of the statistical estimates derived from the data depends on several assumptions.
First, the concentrations of toxic substances in the adipose tissue of cadavers and surgical
patients are assumed to be comparable to those in the general population. Second, it is
assumed that the levels of toxic substances in urban residents are approximately the same as
in rural residents, and therefore the selection of only urban hospitals and medical examiners
(i.e., those located in MSAs) does not introduce any significant bias into the estimates of
average concentration levels. Finally, it is assumed that no systematic bias is introduced by
29
-------
the fact that the cooperators are not randomly selected, and that the donors and specimens
are nonprobabilistically sampled according to pre-spedfied quotas.
4.1.1 Selection of MSAs
Prior to 1985, the sampling strata from which MSAs were randomly selected were the
nine U.S. Census divisions. But in 1985 EPA wanted the ability to obtain estimates of average
concentration levels in each of the ten EPA regions. Therefore, beginning with the FY85
NHATS, sampling strata were redefined to be the seventeen geographic areas of the country
which resulted from the intersection of the nine Census divisions and the ten EPA regions
(Panebianco DL, 1986a). Selecting the sample in this manner made it possible to make
comparisons with previous NHATS results and also obtain estimates for the EPA regions. The
states, Census divisions, and EPA regions that define the seventeen strata are shown in
Table 4-1.
Beginning with the FY85 NHATS, a new sample of MSAs was selected from the set of
seventeen strata using a controlled selection technique, known as the Keyfitz technique,
which maximized the probability of retaining the MSAs used in the FY84 design (Mack et al.
1984). The MSA sample selected in FY85 has served as the base NHATS sample for FY86 and
FY87. The FY87 NHATS sampling design contains forty-seven MSAs. Thirty-eight of the FY87
MSAs were selected in the FY85 sample. The remaining nine FY87 MSAs are replacements.
Some of the FY85 sample MSAs were replaced because satisfactory cooperators could not be
found.
The forty-seven MSAs included in the FY87 NHATS design are listed by stratum in
Table 4-2. Four MSAs (Los Angeles, Chicago, Detroit, and New York) are listed as double
collection sites because their populations are much larger than the populations of the other
MSAs. While given positive probability of selection, no MSAs were selected from strata 13, 15,
and 17 because of the small population size of these strata.
4.1.2 Collection Quotas
Sampling within MSAs is done by quota. A quota of specimens is assigned to each
sample MSA. In addition, demographic subquotas are assigned to each MSA so that the
specimens collected will be representative of the strata with respect to the three demographic
factors: age group, race, and sex. The subquota for each MSA is determined by the
demographic makeup of the stratum to which the MSA belongs and is based on the 1980 U.S.
Census population counts. Each combination of age group and sex is proportionally
represented in the subquota. The race categories are also proportionally represented, but the
30
-------
Table 4-1. Summary of Sampling Strategy
Stratum
Census
division
EPA
region
State
3
4
8
9
10
New England
Middle Atlantic
Middle Atlantic
South Atlantic
South Atlantic
East South Central
East North Central
West North Central
West South Central
West North Central
3
3
5
6
Maine
New Hampshire
Vermont
Massachusetts
Rhode Island
Connecticut
New York
New Jersey
Pennsylvania
Delaware
Maryland
District of
Columbia
Virginia
West Virginia
North Carolina
South Carolina
Georgia
Florida
Kentucky
Tennessee
Alabama
Mississippi
Ohio
Indiana
Illinois
Michigan
Wisconsin
Minnesota
Arkansas
Louisiana
Oklahoma
Texas
Iowa
Missouri
Nebraska
Kansas
31
-------
Table 4-1 (Continued)
Stratum
Census
division
EPA
region
State
11
12
13
14
15
16
17
West North Central
Mountain
8
Mountain
Pacific
Mountain
Pacific
Mountain
9
10
10
North Dakota
South Dakota
Montana
Wyoming
Colorado
Utah
Arizona
Nevada
California
Idaho
Washington
Oregon
New Mexico
32
-------
Table 4-2. Sample MSAs for Fiscal Year 1987
National Human Adipose Tissue Survey
Stratum
Census
division
EPA
region
MSA
1
2
4
5
New England
Middle Atlantic
Middle Atlantic
South Atlantic
South Atlantic
East South Central
East North Central
1
3
4
West North Central
Springfield, MA
Boston, MA
Syracuse, NY(b)
New York, NY
-------
Table 4-2 (Continued)
Stratum
Census
division
EPA
region
MSA
10
11
12
14
16
West South Central
West North Central
West North Central
Mountain
Pacific
Pacific
8
8
10
New Orleans, LA™
Browns ville-
Harlingen-San
Benito, TX^
Houston, TX
San Antonio, TX
Dallas, TX
Omaha, NE
St. Louis, MO
Wichita, KS «
Grand Forks, ND0"
Salt Lake City, UT
Denver, CO
San Francisco, CA
Sacramento, CA
Los Angeles, CA(a)
Portland, OR
Spokane, WA
Olympia, WA*1
Tacoma, WAW
(a) Indicates a double collection site. A double collection site is an MSA whose population
relative to its stratum is so large that its proper representation in the sample requires it to
be selected twice.
^ Indicates replacement MSA. A replacement MSA is an MSA that was not selected in the
FY85 probability sample, but was chosen to replace an FY85 sample MSA for which a
satisfactory cooperator could not be found.
34
-------
subquota does not specify that Caucasians and non-Caucasians are to be proportionally
represented within each combination of age group and sex. The subquotas only specify the
total number of Caucasian and non-Caucasian specimens to be collected from each MSA.
The subquotas for the seventeen sampling strata are presented in Table 4-3. A total
quota of twenty-seven specimens was specified for each MSA, except those that were
designated as double collection sites. In those MSAs the quotas and subquotas were doubled.
Cooperators within an MSA were assigned quotas and subquotas appropriate to that MSA.
The total number of samples specified for the FY87 NHATS was 1377. This is based on
the quota of 27 specimens for each of the 47 MSAs plus 27 additional specimens for each of
the four MSAs designated as double collection MSAs.
4.2 SAMPLE COLLECTION PROCEDURES
NHATS specimens are adipose tissue samples excised by pathologists and medical
examiners during therapeutic or elective surgery or during postmortem examinations. If the
specimen was collected postmortem, the tissue was obtained from an unembalmed cadaver
which had been dead for less than twenty-four hours and had been kept under refrigeration
during that time. The death should have been caused by sudden traumatic injury, such as
cardiac arrest, car accident, or gunshot wound.
The following groups were excluded from specimen collection:
• Institutionalized individuals;
• Persons known to be occupationally exposed to toxic chemicals;
• Persons who died of pesticide poisoning; and
• Persons suffering from cachexia.
These guidelines were stipulated so that the levels of substances detected in the specimens
were a result of environmental exposure.
All NHATS cooperators in the selected MSAs were provided with target quotas for
specimen collections from age, sex and race groups. The cooperators were asked to obtain at
least five grains of tissue from each donor. The cooperators were instructed to avoid
contamination through contact with disinfectants, parafins, plastics, preservatives and
solvents. After collection the specimens were placed in glass jars with Teflon® lids and frozen
35
-------
Table 4-3. Age, Race, and Sex Subquotas for each NHATS Collection
Site within a Stratum«FY87 Design
0-14 Yr
Stratum
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
Census division
New England
Middle Atlantic
Middle Atlantic
South Atlantic
South Atlantic
East South Central
East North Central
West North Central
West South Central
West North Central
West North Central
Mountain
Mountain
Pacific
Mountain
Pacific
Mountain
EPA
region
1
2
3
3
4
4
5
5
6
7
8
8
9
9
10
10
6
No. of non-
Caucasians
2
5
3
6
6
5
4
1
6
2
2
2
4
7
1
2
7
Male
3
3
3
3
3
3
3
3
4
3
3
3
3
3
4
3
4
Female
3
3
3
3
3
3
3
3
3
3
3
3
3
3
4
3
3
15-44 Yr
Male
6
6
6
6
6
6
6
6
6
6
6
7
6
6
6
7
6
Female
6
6
6
7
6
6
6
6
6
6
6
7
6
7
6
7
7
45+ Yr
Male
4
4
4
4
4
4
4
4
4
4
4
3
4
4
3
3
3
Female
5
5
5
4
5
5
5
5
4
5
5
4
5
4
4
4
4
36
-------
at -10° to -20°C. These jars were packed on dry ice for overnight shipment to MRI. Upon
receipt at MRI the specimens were checked to determine that they were intact and frozen.
The specimens were checked versus the cooperators quota, an approximate specimen weight
was determined and the specimens were transferred to storage at -20°C. As part of the check-
in process, the patient summary reports (PSRs) were cross checked for consistency with the
specimen labels and identification data. The PSRs were forwarded to Battelle for processing.
4.3 SAMPLE COLLECTION SUMMARY
In FY87, 956 specimens were collected from the cooperators. Of these, 771 were
collected in accordance with the quotas and subquotas. The rest of the specimens were
collected by cooperators, but went beyond what the quotas and subquotas requested.
In designating which collected specimens to chemically analyze, EPA decided to include
some specimens that were collected outside of the subquotas. This was done in the
anticipation of the use of the linear model, with parameters for the effects of Census region,
age group, sex, and race group, to statistically analyze the results. The maximum number of
specimens from a MSA remained at the original quota of twenty-seven or fifty-four. In effect,
the MSA quotas were maintained, but the MSA subquotas were allowed to vary from what
was specified in the design. This approach, while not ideal, was the best means available to
deal with MSA non-response. Of the 956 collected specimens, 865 were designated for
chemical analysis. These 865 specimens were labeled "Design" specimens.
Table 4-4 is a summary of the collection effort for the FY87 NHATS. Table 4-4 shows
collection information for the nine Census divisions. In FY87, EPA chose not to make
estimates for EPA regions. Instead, EPA maintained similarity to the FY82 geographic
classifications in order to compare FY87 results to FY82 results.
Table 4-5 shows the number of quota specimens, collected specimens, and Design
specimens in each category defined by the four analysis factors in the linear model. Because
the number of samples in the chemical analysis was not large enough to obtain reliable
estimates for all nine Census divisions, the divisions were combined into four Census regions
for both the FY82 and FY87 model analyses.
37
-------
b
re
I
3
en
1
u
CD
•13
0
U
en
H
"^l
rM
2
So
^ "g
III
S a> C
a «rt tu
2 *
'S-aS
X-i QJ ni
nj +•» p
111
11 8,
2 «
MH OJO
o c
»H 'J3 e«
re cu
QJ 4-. g
•^ 9 §
Si S 3 -Q
£ 3 ^ S
M-' ^ CU
. 2 «
t
Tp-
~
"°
re
H
1
«^4 (
^
! QJ ,„
t-* ,-~^ c/i
QJ fZ -^
^r MH
-------
Table 4-5. FY87 NHATS Sample Sizes by Categories
Analysis
factor
Census
Region
Age
Sex
Race
Number of Number of Number of
quota collected design
Category specimens specimens specimens
Northeast
North Central
South
West
Total
0-14 years
15-44 years
45+ years
Total
Male
Female
Total
Caucasian
Non-Caucasian
Total
270
405
432
270
1,377
311
630
436
1,377
668
709
1,377
1,157
220
1,377
195
307
349
105
956
163
353
440
956
499
457
956
776
180
956
175
296
289
105
865
146
318
401
865
436
429
865
707
158
865
39
-------
5.0 COMPOSITE DESIGN
The 865 design specimens in the FY87 NHATS were assigned to composite samples
using specific composite design criteria (Leczynski et al. 1988). The reasons for compositing
samples prior to chemical analysis were: (1) at least 10 grams of tissue were needed to meet
the limit of detection goals for the target compounds, and (2) the budget for chemical analysis
of samples could only support the analysis of 48 samples.
5.1 DESIGN GOALS AND COMPOSITING CRITERIA
The seven goals of the FY87 composite design, listed in order of importance, were
• Create no more than 48 composite samples.
• Maintain similarity to the FY82 composite design.
• Maintain equal tissue mass of individual specimens within the composite samples.
• Specify more pure sex composite samples than in FY82.
Control the MSA effect.
• Provide the best range of race group percentages across the composite samples.
• Maintain a constant number of specimens across all composite samples.
Because of the constraints imposed by the sampling and compositing protocols and the
frequency of collection nonresponse, it was not possible to meet all of the design goals. Each
of the last six goals required a different mix of individual specimens within the composite
samples. Thus, attempts were made to achieve all goals across the design to the extent
possible. The criteria used to design composite samples are discussed below.
(1) Create no more than 48 composite samples.
This criterion was based entirely on the funds available for the chemical analysis.
(2) Maintain similarity to FY82 composite design.
This specification was included to ensure that comparisons between the analysis results
from the two years could be made. The design criterion imposed by this objective is that
each composite sample had to be constructed from individual specimens collected from exactly
one Census division and exactly one age group. Thus, there were 27 distinct categories
defined by the intersection of the nine Census divisions and three age groups.
41
-------
(3) Maintain equal weighing of specimens within the composite samples.
The interpretation of the observed concentrations of the composite samples is far easier
when these concentrations can be interpreted as the arithmetic average of the concentrations
of the individual specimens. Therefore, this design goal specified that each individual
specimen within a composite sample contribute an equal amount of tissue to that composite
sample. This specification allows the lipid adjusted concentration of the composite sample to
be interpreted as approximately the arithmetic average of the lipid adjusted individual
specimen concentrations, with equality whenever all the specimens in the composite sample
have the same percentage of lipid material.
(4) Construct more pure sex composites than in FY82.
Pure male or pure female composite samples were constructed when sufficient numbers
of specimens were available within a particular Census division/age group category. Pure sex
composite samples are samples in which all of the individual specimens were collected from
donors of the same sex. Such composites are needed to achieve more precise estimates of sex
effects in the population. It was anticipated that including more pure sex composite samples
in the FY87 design will lead to smaller standard errors for the sex group estimates (Draper and
Smith 1981, pp. 52-55) than was the case in FY82.
(5) Control the effect of the MSAs contributing specimens to each composite.
Controlling the number of MSAs contributing specimens to composite samples is
intended to reduce the effect of the MSA on the estimated average concentrations. This is
done because MSAs are regarded as being major sources of differences in observed
concentrations across the nation due to their varied exposure scenarios (Panebianco 1986b).
To avoid confounding the MSA effect with any of the geographic or demographic effects, two
design criteria were identified. They were
(5-a) Keep the number of MSAs represented in each composite sample consistent
across the design (2-3 MSAs was the target number), and
(5-b) Maintain approximately the same number of pure male and pure female
composite samples within a group of MSAs.
The first criterion helps to ensure a constant variance of measured concentrations across the
sample whenever the composite sample concentrations are averages over an equal number of
MSAs. The second criterion is intended to prevent confounding a large MSA effect with the
sex effect.
42
-------
(6) Vary the percentage of Caucasian and non-Caucasian specimens in the
composites as much as possible.
For the same reasons it is important to construct pure sex composite samples, it is
important to construct pure race group composite samples. However, this goal is dependent
on the number of non-Caucasian specimens collected in the twenty-seven Census division/age
group categories and the number of composite samples in the design. Since the number of
non-Caucasian specimens collected in the FY87 sample was relatively small, it was decided to
provide the best possible range of race group percentages (i.e., mixes of the Caucasian and
non-Caucasian specimens within the composite samples) across the design rather than focus
on designing pure race group composite samples.
(7) Maintain a constant number of specimens across all composite samples.
This goal, similar to goal 5-a above, could not be fully achieved for the FY87 composite
samples.
5.2 LABORATORY COMPOSITING PROCEDURES
The FY87 NHATS specimens from nine census divisions and three age groups were
divided into 48 composites, as identified in the composite design provided by Battelle-
Columbus Laboratories (Battelle). Battelle provided MRI with the data sheets that identified
the individual specimens and their required weights to be included in each composite. Each
composite consisted of from 3 to 32 specimens. The composite sample data sheets provided
sufficient information (EPA ID number, package number, sample weight, hospital code, etc.)
such that the individual specimens could be cross-checked with the study design. The data
sheets provided by Battelle were used as work sheets to record the actual laboratory
compositing procedures.
Initially, the samples were grouped into composites, and any samples of questionable
weights were noted. Three samples identified below did not appear to have the weight
required by the composite design. These results were relayed to the EPA Work Assignment
Manager (WAM). After consultation with DDE, the EPA WAM forwarded to MRI the
following responses regarding the problem samples on June 24,1988.
43
-------
Composite number Sample number Problem Response
ACD8700023 8706954 Low weight, - 0.1 g Include as is
need 0.5 g
ACD8700032 8701765 No sample Omit
ACD8700201 8703464 Low weight, - 1.3 g Include as is
need 2.0 g
The specimens in a composite were placed on dry ice during the compositing procedure.
An electronic four-place balance was used to weigh the samples. The calibration of the
balance was checked before any weighing was begun and once during the sample weighings
with a Class P set of weights (laboratory grade, tolerance 1/25,000).
To weigh the samples, a clean culture tube was labeled with the composite number and
placed on the balance and the weight tared. A specimen jar was opened, and a portion of the
frozen adipose tissue removed with a clean stainless steel spatula. The adipose tissue was
placed in the culture tube and the weight recorded to three decimal places on the compositing
sheets. Additional adipose tissue was added if necessary. A goal of ±10% of the desired
weight was attempted where possible. The specimen jar was capped and returned to storage.
The weights of the individual specimens were recorded on the data sheets provided by
Battelle.
The weight of the culture, beaker, and adipose tissue was rezeroed, and the next
specimen in the composite was weighed. A new spatula was used for each sample. This
procedure was repeated for each sample in the composite. When the composite was com-
pleted, it was capped and stored in a sample freezer at -10° to -20°C. All data on the actual
compositing procedures were recorded on the data sheets provided by Battelle. All data sheets
were submitted in a separate report to document the compositing activity (Cramer and Stanley
1988).
44
-------
5.3 SUMMARY OF COMPOSITE SAMPLES
The FY87 NHATS Composite Design resulted in the construction of 48 composite
samples, using the 865 design specimens collected from 41 MSAs. Table 5-1 shows the number
of composite samples for the 27 distinct combinations of Census division and age group. The
sex and race group percentages of the composite samples vary across the design depending on
the availability of specimens within specific demographic subpopulations. Table 5-2 shows the
demographic makeup of the FY87 NHATS composite samples.
The 48 composite samples were randomly assigned to five batches. Within batches, the
composite samples were placed in random order for the chemical analysis.
45
-------
Table 5-1. Distribution of FY87 NHATS Composite Samples
by Census Division and Age Group
Age group
Census division
New England
Middle Atlantic
South Atlantic
East South Central
West South Central
East North Central
West North Central
Mountain
Pacific
Total
0-14 Years
1
1
2
1
1
2
1
1
1
11
15-44 Years
1
3
4
1
1
3
2
1
1
17
45+ Years
1
2
4
1
1
5
2
1
3
20
Total
3
6
10
3
3
10
5
3
5
48
46
-------
3
ex
Si
1
8
1
z
00
PL,
•8
a,
2
u
•£
a.
S
1
g
D
s
23
H
Percent
Caucasian
S-S
rj tU
-------
"2
3
J
in
8°3808°8
5J°8°S
rH O
IXOVT^ONOO inM
-------
6.0 CHEMICAL ANALYSIS PROCEDURES AND QUALITY CONTROL DATA
The 48 composite samples were prepared in the analysis laboratory for determination of
low pg/g (ppt) levels of PCDDs and PCDFs using high-resolution gas chromatography/high-
resolution mass spectrometry (HRGC/HRMS). The performance of the analysis effort was
demonstrated through the determinations of PCDDs and PCDFs in 20 quality control samples
(method blanks, controls, and spiked tissues). Chemical analysis performance was also
documented through participation in an interlaboratory effort with two other laboratories
recognized for their expertise in the determination of PCDDs and PCDFs in human tissues. This
section describes the analytical methodology and presents the results (analytical and statistical)
for the quality control samples. A detailed presentation of the analytical results for all PCDDs and
PCDFs in the FY87 NHATS design samples and the quality control samples is given in
Appendix A.
6.1 SAMPLE PREPARATION
The preparation of the composited adipose tissue specimens for determination of PCDDs
and PCDFs required a multistep procedure, which included quantitative extraction and cleanup
through several chromatographic columns. The procedures described below were carried out for
each of the five sample batches for the FY87 NHATS study.
6.1.1 Extraction
After compositing, the adipose tissue samples were stored at -20°C in 50-mL culture tubes
sealed with aluminum foil. The extraction procedure was initiated by allowing the samples to
come to room temperature and then fortifying them with 100 fiL of an internal quantitation
standard (IQS) spiking solution (Table 6-1) containing nine carbon-13 labeled PCDDs and PCDFs.
Ten milliliters of methylene chloride was added and the sample was homogenized for 1 min with
a Tekmar Tissuemizer®. The mixture was allowed to separate, and the methylene chloride was
decanted through a funnel of sodium sulfate into a 100-mL volumetric flask. The homogenization
was repeated two to three times with fresh 10-mL portions of methylene chloride. The culture
tube was rinsed with additional methylene chloride and the remaining contents of the tube
transferred to the funnel. Finally, the funnel was rinsed with additional methylene chloride and
the final volume was brought to 100 mL.
At this point the flask was stoppered and inverted several times to mix the extract. Next,
a 1-mL aliquot was removed with a disposable graduated pipet and placed into a preweighed
(measured to 0.0001 g) 1-dram glass vial. The methylene chloride in the vial was reduced under
49
-------
Table 6-1. Internal Standard Spiking Solution
for chlorinated Species3
Concentration
Analyte (pg/\iL)
Chlorinated Internal Quantitation Standards'1
13C12-2,3,7,8-TCDD 5
13C12-2,3,7,8-TCDF 5
13C]2-l,2,3,7,8-PeCDD 5
13Cirl,2,3,7,8-PeCDF 5
13C12-l,2,3,6,7,8-HxCDD 12.5
13C12-l,2,3,6,7,8-HxCDF 12.5
13C12-l,2,3,4,6,7,8-HpCDD 12.5
13C]2-l,2,3,W,8-HpCDF 12.5
13C]2-OCDD 25
Internal Recovery Standard*"
13C12-1,2,3,4-TCDD 50
13C12-l,2,3,7,8-HxCDD 125
" All internal quantisation and recovery standards were obtained
as solutions from Cambridge Isotope Laboratories (Woburn,
Massachusetts).
b Prepared in isooctane. One hundred microliters spiked. Separate
solutions were used for chlorinated and brominated species.
c Prepared in tridecane. Used for both chloro and bromo analyses.
50
-------
flowing nitrogen until a constant weight of lipid was obtained. The weight of the lipid was
obtained by difference, and the percent lipid for the composite was calculated and recorded.
The remaining portion of the extract (99 mL) was quantitatively transferred, followed by
a 20- to 30-mL rinse, to a 500-mL round-bottomed flask. The extract was concentrated under
vacuum to an oily residue (extractable lipids) using rotary evaporation.
6.1.2 Bulk Lipid Removal
Separation and concentration of the PCDDs and PCDFs from the lipids to achieve a final
volume of 10 \iL is necessary to detect pg/g concentrations. The extractable lipids from some of
the composites was as high as 9 g of oily materials. The bulk lipid was removed following an
acidic silica gel slurry cleanup procedure . This was accomplished by adding 200 mL of hexane
and a Teflon-coated stirring bar to the lipid in the round-bottomed flask Then, while stirring the
extract on a magnetic stir plate, 100 g of 40% w/w sulfuric acid-impregnated silica gel was slowly
added to the extract. The mixture was stirred for 2-hours. During the 2-hour slurry period,
acid/neutral silica gel columns (4 g 40% H2SCVsi]ica gel, 1 g silica gel) were prepared. After the
2-hour period, the slurry mixture was allowed to settle, and the hexane was decanted off the acid
impregnated silica gel through a funnel of sodium sulfate into the acid/neutral silica gel column.
The slurry mixture was rinsed for 15 minutes with two additional aliquots (50 mL) of hexane.
The rinses were added to the silica gel column through the sodium sulfate funnel. The eluate of
the column was collected in a 500-mL Kuderna-Danish evaporation flask An additional 50 mL
of hexane was placed onto the column when the solvent level had reached the level of the
chromatographic packing. The extract was then reduced in volume over a steam bath and the
final volume adjusted to approximately 1 mL using nitrogen blowdown.
6.1.3 Separation of Chemical Interferences
Separation of chemical interferences, such as pesticides, PCBs, and other chlorinated planar
aromatics is essential to avoid false positive measurements. Removal of chemical interference was
achieved using two different chromatographic cleanup systems. The first was prepared as a
layered column containing 1 g sodium sulfate, 4 g neutral alumina, and 1 g sodium sulfate. The
1 mL extract from the acid/neutral silica gel column was transferred to the alumina column,
followed by two 1-mL portions of hexane and 10 mL of 8% (volume/volume, v/v) methylene
chloride in hexane. These eluents were collected and archived. The PCDDs and PCDFs were
eluted from the column with 15 mL of 60% (v/v) methylene chloride in hexane and the eluent
concentrated under a stream of nitrogen to approximately 2 mL.
51
-------
A disposable column of AX-21 on silica gel was prepared and preeluted with 4 mL of
toluene, 2 mL of 75:20:5 methylene chloride/methanol/benzene, and 2 mL of 1:1 cyclohexane/
methylene chloride. The concentrated eluate from the alumina column was added to the
AX-21/silica gel column followed by two 1-mL hexane rinses. The column was eluted sequentially
with two 0.5-mL aliquots of hexane, 10 mL of 1:1 cyclohexane/methylene chloride, and 5 mL of
75:20:5 methylene chloride/methanol/benzene. These eluents were combined and archived. The
columns were turned upside down and the PCDDs and PCDFs eluted with 20 mL of toluene.
The extract was then reduced in volume to approximately 100 \iL, then 10 |iL of recovery standard
in tridecane was added (Table 6-1) and the volume was further reduced to 10 |iL under nitrogen.
The extract was stored in a freezer pending HRGC/HRMS analysis.
6.2 HRGC/HRMS ANALYSIS
Initial calibration of the GC/MS system was conducted by making single 1-pL injections
of the standards listed in Table 6-2. Relative response factors calculated from this calibration effort
are shown in Table 6-3. A CS7 (2.5 to 12.5 pg/|iL) standard was analyzed on a daily basis to
ensure adherence to the initial calibration curve. The traceability and comparability of the
analytical standards has been demonstrated in a previous NHATS effort (USEPA 1986) and
through participation in an interlaboratory comparison study (Bradley, et al. 1990).
HRGS/HRMS analysis of the samples was conducted after initial and routine calibration
criteria were met. Prior to the injection of the first sample, an injection of tridecane was analyzed
to document system cleanliness. If any evidence of system contamination was found, corrective
action was taken by analyzing another tridecane blank or cleaning the injection system. A typical
daily sequence of injections is shown in Table 6-4. A 1-fiL aliquot of the extracts was injected into
the GC/MS system, which was operated under the conditions that previously produced acceptable
results with the daily calibration standard.
Selected ion monitoring (SIM) data were acquired according to the acquisition and MS
operating conditions previously used to determine the relative response factors (Tables 6-5 and
6-6). Instrument performance was monitored by examining and recording the peak areas for the
recovery standard, 13C]21,2,3,4-TCDD. If this area decreased to less than 50% of the calibration
standard, sample analyses were stopped until the problem was identified and corrected.
52
-------
Cft
U
rq N
vd
O
IQ
U
g g g g tg
o
PH
O
o
8
f
in
Cft
U
ir>rq
in
§
•43
O
•43
2
£>
U
I
I
§
§
u
rq
o
en
O
1
.s
o
1
u
§
U
eft
u
co
tn
U
rq
Cft
U
Cft
U
SSS8S
m
ifits
ir>rq
IT>CN
01
1
in
sf
u
iftioinir>c
4 tC
mm
U U
53
-------
in
rH CO in ON O^ rH SO
_ rqONrn^ooco
rHsOOOOONlNSOOOOOO^cqiAoinCNlCNlOOit'
CNlCNlrHrHCNlinCNlCNlOON^;Tj
I
U
T*I
CO
ON OO ON rH ^^
tN in oo in CT^
OO ^» SO CO rH
00 CO
rH CO
in
2
rHOCDrHOCDOOrHrHrHrHrHrHrHCDO
(0
O
O
£
'•a
U
CO
yi
U
CN1
V3
U
C/3
U
O CO rH CO C5 SO
t^ CN *^> ^^
— — CD ^ tx ON O
IN so rq oo in rj< oo
SO SO CO rH CO rH
fNONQCNlCOrHCOCOinoptN
SO in O CO rH
a
rH
CNl
rHOrHOOOOrHrHrH
00 SO OO
CN 00 tN
in rn cq
fc %
in IN
rH rH rH O
ON rH CNl
CO OO CN
o cs ci ^ tx
ON Uj rH sO CO O in tN CO
ON^msotN^cNicNio
CDOrHOCDCDCDrHrH
r
OOsOONOOsOsOCXIONlN^CNlsO
inrHCNrHOinsOtN-'frHCOrH
so
in
U
CNl OO O rH CNJ ON "^ CD
ONinorHCNicNiinso
CNI sO rH in in SO SO CO
CNlrHrHOrHOeDCD
00 IN
rH CO
oo m
CM
CO rH Q M
. . CNI ON O IN
in 00 IN SO rH
rH CO
OrHOrHi—IrHi-HrHrHrHrHrH
I
75
PH
PH O 8
o o -
U U
li?
CO CO^ CNl
fsf CNt rn"
CDD
,8-Pe
PH O
Q G
U U
CM (X
£ K E
£ oi 06
°R. t\ K.
tN SO SO
SO- •*.
PH
O
u
PH G
PH
O R
o y
g3
s^s
^sXT^.'>1,"'v^vrNv-'SV1-H'--«i—T ^J so] so] tNv TJ^ sov
tN. CO. TjJ. CO. CO. CO. ^J. CO. CO. CO.
co C*Q co c*i cq cq co
CNf r-f CNf r-T r-? r-I" C
-------
Table 6-4. Typical Daily Sequence for PCDD/PCDF Analysis
/ 1. Tune and calibrate mass scale versus perfluorokerosene (PFK).
2. Inject column performance/window-defining mixture.
3. Inject concentration calibration solution 2.5 to 12.5 pg/fiL (CS-7) solution.
4. Inject blank (Tridecane).
5. Inject samples 1 through "N."
6. Inject concentration calibration solution 2.5 to 12.5 pg/|iL (CS-7) solution or other
concentration calibration solutions CSl to CSS to bracket observed sample
concentration.
55
-------
Table 6-5. Ions Monitored for HRGC/HRMS of PCDD/PCDF
Descriptor
Al
A2
A3
ID
TCDF
13C12-TCDF
TCDD
13C12-TCDD
HxCDPE
PFK (lock mass)
TCDF
TCDD
PeCDF
13C12-PeCDF
PeCDD
13C12-PeCDD
PFK (lock mass)
HpCDPE
HxCDF
PFK (lock mass)
J3C12-HxCDF
HxCDD
13C12-HxCDD
OCDPE
Mass
303.902
305.899
315.942
317.939
319.897
321.894
331.937
333.934
377.886
380.976
303.902
305.899
319.897
321.894
339.863
341.860
351.900
353.894
355.858
357.855
367.895
369.889
380.976
409.877
373.821
375.818
380.976
383.861
385.858
389.816
391.813
401.856
403.853
445.866
Nominal dwell
time (s)
0.090
0.090
0.090
0.090
0.090
0.090
0.090
0.090
0.090
0.090
0.045
0.045
0.045
0.045
0.045
0.045
0.045
0.045
0.045
0.045
0.045
0.045
0.035
0.035
0.080
0.080
0.080
0.080
0.080
0.080
0.080
0.080
0.080
0.080
56
-------
Table 6-5 (continued)
Descriptor ID
A4 PFK (lock mass)
HxCDD
HpCDF
13C]2-HpCDF
HpCDD
13C12-HpCDD
13C12-HpCDD
NCDPE
A5 PFK (lock mass)
OCDF
I3C]2-OCDF
OCDD
I3C]2-OCDD
DCDPE
Mass
380.976
389.816
391.813
407.782
409.779
417.822
419.819
423.777
425.774
435.817
437.814
429.768
431.765
479.856
380.976
441.743
443.740
453.783
455.780
457.738
459.735
469.779
471.776
513.846
Nominal dwell
time (s)
0.040
0.040
0.040
0.040
0.040
0.040
0.040
0.040
0.040
0.040
0.040
0.040
0.040
0.040
0.06
0.07
0.07
0.07
0.07
0.07
0.07
0.07
0.07
0.06
57
-------
Table 6-6. HRGC/HRMS Operating Conditions for PCDD/PCDF Analysis
Mass spectrometer (Kratos MS50-TC)
Accelerating voltage: 8,000 V
Trap current: 500 pA
Electron energy: 70 eV
Electron multiplier voltage: -1,800 V
Source temperature: 280°C
Resolution: ;> 10,000 (10% valley definition)
Overall SIM cycle time: 1 s
Gas chromatograph (Carlo Erba MFC-500")
Column coating: DB 5
Film thickness: 0.25 jim
Column dimensions: 60 m x 0.25 mm ID
He linear velocity: ~ 25 cm/s
He head pressure: 1.75 kg/cm2 (25 psi)
Injection type: Splitiess, 45 s
Split flow: 30 mL/min
Purge flow: 6 mL/min
Injector temperature: 270°C
Interface temperature: 300°C
Injection size: 1-2 nL
Initial temperature: 200°C
Initial time: 2 min
Temperature program: 200°C to 270°C at 5°C/min
Second hold time: 10 min
Second temperature ramp: 270° to 330°C at 5°C/min
Final hold time: 5 min
58
-------
6.3 OA/OC FOR CHEMICAL ANALYSIS
The QA/QC program for this analysis effort included: demonstration of instrumental
performance, routine analysis of quality control samples (method blank, controls and spiked
tissues), analysis of performance audit samples and participation in an interlaboratory effort for
comparison of results on specific samples. Each of these QA/QC efforts and results are discussed.
6.3.1 Instrument Performance
Instrument performance was characterized primarily by three criteria: (1) mass resolution
(^ 10,000) and calibration; (2) relative response factors (RRF), i.e., adherence to the initial RRFs;
and (3) column performance as indicated by peak separation between 2,3,7,8-TCDD and other
TCDD isomers.
6.3.1.1 Mass Resolution and Calibration
The mass spectrometer was tuned on a daily basis to yield optimum sensitivity and peak
shape using an ion m/z 380.9760) from PFK. The resolution was visually monitored and
maintained at ^ 10,000 (10% valley definition) to provide adequate noise rejection while
maintaining good ion transmission. Static-resolving power checks were performed at the
beginning and at the end of each 12-hour operation period. A visual check (i.e., documentation
was not required) of the static resolution was made by using the peak matching unit before and
after each analysis. Corrective action was implemented whenever the resolving power did not
meet the criteria of ^ 10,000.
Mass calibration of the mass spectrometer for the HRGC/MS analysis of PCDD/PCDF was
conducted on a daily basis. The magnetic field was adjusted to pass m/z 300 at full accelerating
voltage. PFK was admitted to the MS, and an accelerating voltage scan from 8,000 to 4,000 v was
acquired by the data system. This corresponded to an effective mass range of 301 to 593 amu.
Upon completion of a successful calibration step, the five ion descriptors shown in Table 6-5 were
updated to reflect the new mass calibration.
6.3.1.2 Relative Response Factor
As part of the initial and routine instrument performance checks, calibration standards were
analyzed and the responses of the respective analytes were compared to specific internal stan-
dards to establish the RRF values. The initial and routine calibration criteria required that the
precision of the RRF measurements be within ±20% for the tetrachloro congeners and within
±30% for the other compounds.
59
-------
Sensitivity of the HRMS was documented through the responses noted for the first
calibration standard of each analysis day. The method required the analysis of a low level
standard (CS7) to document sufficient instrumental response to support instrumental detection
limits of 1 pg/fiL for TCDD.
Routine checks on the instrument sensitivity, which were documented in the MS log book,
was achieved by monitoring the response for the internal recovery standard (13C12-1,2,3,4-TCDD)
from injection to injection. If the response for this standard dropped by greater than 50% of the
response noted in the previous calibration standard, the analyst verified instrumental performance
by analyzing an additional calibration standard. Additional details on the initial and routine
calibration events are presented in the data reports provided to EPA (Cramer et al. 1989a, 1989b).
6.3.1.3 TCDD Peak Separation
The HRGC column performance was demonstrated at the start of each 12-h analysis period.
This was accomplished by injecting 1 jiL of the column performance/window-defining check
solution and acquiring SIM data for all PCDD and PCDF compounds. The HRGC column
performance was determined based on the ability to resolve 2,3,7,8-TCDD from possible coeluting
TCDDs.
The chromatographic peak separation between 2,3,7,8-TCDD and the peaks representing
any other TCDD isomers was resolved with a valley of s 25%, where
valley % = (*/y)(100)
x = measured height of the valley between the chromatographic peak corresponding to 2,3,7,8-
TCDD and the peak of the nearest TCDD isomer
y = peak height of 2,3,7,8-TCDD
Figure 6-1 is an example of the separation of a TCDD isomer mixture and the calculation of
isomer resolution. The TCDD isomer resolution was documented to range from 18% to 25%
during the analysis effort (Cramer et al. 1989a, 1989b)
6.3.2 QC Samples
Samples included for QC purposes are summarized in Table 6-7. Each of these quality
control samples are described in further detail below. These quality control samples were
included with the analysis of the FY87 samples. The order of preparation and analysis with
respect to the FY87 NHATS composites was specified in the study design.
60
-------
en
§
I
tfl
Q
Q
U
61
-------
Table 6-7. Quality Control Samples
Type Frequency Application
Method blank One per batch Assess laboratory background
contribution
Spiked control adipose Two per batch (two Evaluate method performance
tissue sample different spike levels) (accuracy and precision)
Unspiked control adipose One per batch Evaluate method performance
tissue sample (accuracy and precision)
6.3.2.1 Method Blanks
One method blank was generated with each batch of samples. A method blank was
generated by performing all steps detailed in the analytical procedure using all reagents,
standards, equipment, apparatus, glassware, and solvents that were used for a sample analysis,
but omitting the addition of the adipose tissue. The method blank contained the same amounts
of carbon-13 labeled internal quantitation standards that were added to samples before bulk lipid
cleanup. The five method blanks analyzed with the samples did not contain PCDDs or PCDFs
with the exception of the method blank generated for Batch 1 samples. This method blank
contained a trace of 2,3,7,8-TCDF which was determined to be equivalent to 0.46 pg/g of tissue.
The detailed analysis results for the method blanks are presented in Appendix A.
6.3.2.2 Control Samples
Control samples were prepared from a bulk sample of human adipose tissue. This material
was prepared by blending the tissue with methylene chloride, drying the extract by eluting
through anhydrous sodium sulfate, and removing the methylene chloride using rotoevaporation
at elevated temperatures (80°C). The evaporation process was extended to ensure that all traces
of the extraction solvent had been removed. The resulting oily matrix (lipid) was subdivided into
10-g aliquots which were analyzed with each sample batch. A summary of the QC sample results
is presented in Section 6.5. A detailed presentation of this data by analyte is given in Appendix A.
6.3.2.3 Internal Spiked Control Samples
Spiked lipid samples were prepared using a portion of the homogenized control lipid.
Sufficient spiked lipid matrix was prepared to provide a minimum of two spiked samples, one low
and one high, per sample batch. The native spiking solution concentrations are shown in
62
-------
Table 6-8. Additional OCDD was added to each sample with a 75 pg/iiL spiking solution. Low
and high spike levels in the control samples are shown in Table 6-9.
The spiking solutions were checked for accuracy prior to spiking the adipose composite
with the native isomers. The results of this spike check are shown in Table 6-10. The spike check
results for the separate OCDD spiking solution (needed to reach the higher level for OCDD in
the adipose tissue) are given in Table 6-11. A summary of the results from the analysis of these
spiked materials is presented in Section 6.5. The results are presented for each spiked sample in
Appendix A.
6.3.3 Performance Audit Samples (PAS)
Performance audit samples (PAS) were submitted for analysis before the first sample was
analyzed with batch 1 and batch 5. These samples consisted of unlabeled PCDDs and PCDFs in
a solvent matrix. The samples were prepared by the project quality control coordinator (QCC)
and turned over to project personnel who fortified the samples with IQS and RS solutions and
submitted the prepared sample for HRGC/HRMS analysis. The performance audit samples were
prepared from a mixture of standards used for an interlaboratory effort to establish consensus
values for concentrations (Bradley et al 1989). Results from the analysis were given directly to
the QCC. Acceptability criteria were 70% to 130% accuracy for each of the isomers in the sample.
Table 6-12 provides a summary of the results from the analysis of the two performance
audit samples. As presented the measurements were within the criteria specified for each analyte.
6.3.4 Interlaboratory Comparisons
External QC samples and solutions were submitted to two outside laboratories. The
contacts and laboratories were Dr. Donald Patterson with the Centers for Disease Control in
Atlanta and Dr. John J. Ryan with the Health Protection Branch in Canada. Each laboratory
received one spike check solution sealed in an amber ampule and three blind control lipid
samples. The lipid samples were spiked identically to those used with the analysis of the FY87
NHATS samples and included an unspiked sample, a low level spike sample, and a high level
spike sample. Data from the interlaboratory comparison are presented in Tables 6-13 through
6-16. These data demonstrate that although some differences were apparent in the analytical
standards and hence the results for the analysis of the quality control samples, the data for the
respective compounds from the laboratories are generally within 30% relative percent difference.
63
-------
Table 6-8. PCDD and PCDF Native Spiking Solution"
Concentration
Analyte (pg/jiL)
2,3,7,8-TCDD 5
2,3,7,8-TCDF 5
1,2,3,7,8-PeCDD 5
1,2,3,7,8-PeCDF 5
2,3,4,7,8-PeCDF 5
1,2,3,4,7,8-HxCDD 12.5
1,2,3,6,7,8-HxCDD 12.5
1,2,3,7,8,9-HxCDD 12.5
1,2,3,4,7,8-HxCDF 12.5
1,2,3,6,7,8-HxCDF 12.5
1,2,3,7,8,9-HxCDF 12.5
2,3,4,6,7,8-HxCDF 12.5
1,2,3,4,6,7,8-HpCDD 12.5
1,2,3,4,6,7,8-HpCDF 12.5
1,2,3,4,7,8,9-HpCDF 12.5
OCDD 25b
OCDF 25
Prepared in isooctane. This solution also contained similar
concentrations of the available brominated dioxin and furan
, congeners.
The level of OCDD was adjusted based on the endogenous level of
OCDD in adipose tissue. An additional, separate solution of OCDD
at 75 pg/fiL was used to achieve the higher spiking level needed.
64
-------
Table 6-9. Control Sample Spike Levels
Analyte
2,3,7,8-TCDF
2,3,7,8-TCDD
1,2,3,7,8-PeCDF
2,3,4,7,8-PeCDF
1,2,3,7,8-PeCDD
1,2,3,4,7,8-HxCDF
1,2,3,6,7,8-HxCDF
2,3,4,6,7,8-HxCDF
1,2,3,7,8,9-HxCDF
1,2,3,4,7,8/1,2,3,6,7,8-HxCDD
1,2,3,7,8,9-HxCDD
1,2,3,4,6,7,8-HpCDF
1,2,3,4,7,8,9-HpCDF
1,2,3,4,6,7,8-HpCDD
1,2,3,4,6,7,8,9-OCDF
1,2,3,4,6,7,8,9-OCDD
Low spike level
(Pg/g)
10
10
10
10
10
25
25
25
25
50
25
25
25
25
50
350
High spike level
(P8/g)
50
50
50
50
50
125
125
125
125
250
125
125
125
125
250
700
65
-------
Table 6-10. PCDD and PCDF Spike Check Results
Analyte
2,3,7,8-TCDF
2,3,7,8-TCDD
1,2,3,7,8-PCDF
2,3,4,7,8-PCDF
1,2,3,7,8-PCDD
1,2,3,4,7,8-HxCDF
1,2,3,6,7,8-HxCDF
2,3,4,6,7,8-HxCDF
1,2,3,7,8,9-HxCDF
1,2,3,4,7,8-HxCDD
1,2,3,6,7,8-HxCDD
1,2,3,7,8,9-HxCDD
1,2,3,4,6,7,8-HpCDF
1,2,3,4,7,8,9-HpCDF
1,2,3,4,6,7,8-HpCDD
OCDF
OCDD
Replicate no.
1
2
3
Average recovery (%)
Spike
level
(pgW
50
50
50
50
50
125
125
125
125
125
125
125
125
125
125
250
250
Table 6-11
Spike level
150
150
150
Spike check 1
(Pg/nL)
44.3
51.0
47.3
48.6
46.5
124.0
119.4
104.0
93.0
115.9
118.4
107.3
114.3
114.9
115.4
207.8
226.7
. OCDD
Recovery
89
102
95
97
93
99
96
83
74
93
95
86
91
92
92
83
91
Spike check 2
(pg/nL)
41.1
49.5
56.8
54.7
46.1
121.3
116.2
105.3
107.0
115.1
122.6
114.5
114.4
99.7
118.0
210.6
224.6
Recovery
82
99
114
109
92
97
93
84
86
92
98
92
92
80
94
84
90
Average
recovery
85
101
104
103
93
98
94
84
80
92
96
89
91
86
93
84
90
Spike Check Results
(pg/jiL) Amount found
142.6
154.6
138.1
Recovery %
95
103
92
97
66
-------
cfl
&
"tn
O
I
O
U
2
ts.
0>
%
tn
«
tn
1
re
a
IH
O)
P-,
3
tn
a
H
NO
O
I
!8
PH
d
| 5-
% "
8
o
u
t
CO
O
O
~a 5 i-j
fl * =L
-P r! >>.
•a
2
00
v£)
00 CO tx
O CO
CO
o\
00
OO
OO
O O O O O O O
o o
in in
Q
OO OO tx Cx tx ^ ND
V V V V V V V
" c" TH" r-n" T-H"
rfD Q
c^U U
^H-O O
oj 2T-S 5
H.J2-3 «
"•* x
7 re
P
00 O *J w
K s .8 ^
en B cfi 2
- u eg £
^ w 3 S
i-i ,Q tn IH
67
-------
ro
O
•43
•i
en
OJ
X
U
4
o
en
•c
ro
I-
o
U
S
2
o
,0.
ro
CO
»—i
SO
4)
3
ro
H
ro
3?
T3
re
•S
f-i
O
O)
§co oo CN
^ oo IN oo
o in IN ro
ro ON TJ< oo
00 sO SO sO
rO \o ON tS
O ON O CO
SO O rH
OO OO ON rH
I— 1
ro ro ro ro
ro ro to ro
oo oo oo oo
s § y §
U U PH PH
H H oo oo
oo oo tv IN"
tx Cx ro Tf
ro to tN ro
!iri^)cotsl2'^'
fO v^3 ^D ^h tx CT1* LO ro ON
^o cr* c*^ c*^ ON ^j* o^ *~H co
^OlCCOrHrH^OtNO
CO CN OQ C*Q ^i^ CO C"^ ^^ LO
in
cococococorocoo
OOOOOOOOOOOOOOININ
C1 C? C1 C1 C1 C1 C1 '"^ T-*
LLi [l( P tH P
puPPPPPPp.R
ppppuuuuu
^y^yxxx'SS
^^hrs^'''-as-
iJn ,-L 4-< 4j< oq^ON^oq^oo^oq^
ONOOOOONtN'oo'tN'tN'lN"
OOlN"lN"oOso"tNSOSOsO
t^^}*SOtNT}
-------
Table 6-14. Intel-laboratory Comparison—Control Lipid Results (pg/g)
Analyte
2,3,7,8-TCDF
2,3,7,8-TCDD
1,2,3,7,8-PeCDF
2,3,4,7,8-PeCDF
1,2,3,7,8-PeCDD
1,2,3,4,7,8-
HxCDF
1,2,3,6,7,8-
HxCDF
2,3,4,6,7,8-
HxCDF
1,2,3,7,8,9-
HxCDF
1,2,3,4,7,8-
HxCDD
1,2,3,6,7,8-
HxCDD
1,2,3,7,8,9-
HxCDD
1,2,3,4,6,7,8-
HpCDF
1,2,3,4,7,8,9-
HpCDF
1,2,3,4,6,7,8-
HpCDD
1,2,3,4,6,7,8,9-
OCDF
1,2,3,4,6,7,8,9-
OCDD
Labi
n=l
1.2
8
ND(l)b
16
15
19
9
1.6
ND(1)
11
99
17
36
ND(1)
132
ND(1)
986
Lab 2
n=4
Average
1.3
10.6
ND(O.T)
27.1
26.6
31.7
14.6
3.0
ND(0.3)
20.2
131
18.5
31.5
ND(0.4)
212
NRf
1030
RSD (%)
23.1
9.4
-
7.4
9.4
8.8
13.0
20.2
-
11.9
5.3
16.8
18.1
-
7.5
-
6.9
Lab3
n=5
Average RSD (%)
1.121
9.29
0.48C
25.1
20.8
18.7
10.8
ND(9.84)
ND(0.66)
-
136.8d
21.1'
29.8
1.42
140
4.02*
1,184
-
9.7
-
7.5
4.6
-
21.4
-
-
-
6.8
8.4
6.3
-
6.9
70.6
3.7
a Avenge of two positive quantifiable or trace values (n=2).
b Not detected. Detection limit in parenthesis.
c One trace value (n=l).
d 1,2,3,4,7,8- and 1,2,3,6,7,8-HxCDD reported as total amount in Lab 3 results.
e Average of four positive quantifiable or trace values (n=4).
' Not reported.
8 Average of three positive quantifiable or trace values (n=3).
69
-------
Table 6-15. Inter-laboratory Comparison—Low Level Spiked Lipid (% Recovery)
Analyte
2,3,7,8-TCDF
2,3,7,8-TCDD
1,2,3,7,8-PeCDF
2,3,4,7,8-PeCDF
1,2,3,7,8-PeCDD
1,2,3,4,7,8-HxCDF
1,2,3,6,7,8-HxCDF
2,3,4,6,7,8-HxCDF
1,2,3,7,8,9-HxCDF
1,2,3,4,7,8-HxCDD
1,2,3,6,7,8-HxCDD
1,2,3,7,8,9-HxCDD
1,2,3,4,6,7,8-HpCDF
1,2,3,4,7,8,9-HpCDF
1,2,3,4,6,7,8-HpCDD
1,2,3,4,6,7,8,9-OCDF
1,2,3,4,6,7,8,9-0000
Labi
n=l
70
70
56
50
70
56
56
62
68
60
120
72
100
76
56
56
64
Lab 2
n=4
Average
119
98
91
100
110
120
109
107
95
88
128
93
103
133
260
NRC
NR
RSD (%)
3.8
4.4
6.6
5.7
8.8
13.5
12.7
11.1
5.1
8.5
3.1
3.1
5.6
9.6
7.6
-
-
Lab 3
n=5
Average
95
83
115a
99"
94b
_c
95a
99d
79
e
88a
95
114
115
89s
99a
87a
RSD (%)
7.1
23.3
8.8
17.9
39.1
-
13.6
35.2
15.1
-
32.3
20.0
14.6
4.2
11.5
14.9
13.3
a Average of four positive quantifiable or trace values (n=4).
b Average of three positive quantifiable or trace values (n=3).
c Diphenylether interference observed.
d Average of two positive quantifiable or trace values (n=2).
e 1,23,4,7,8- and 1,2,3,6,7,8-HxCDD isomers were reported as a sum.
f Not reported.
70
-------
Table 6-16. Interlaboratory Comparison—High Level Spiked Lipid (% Recovery)
Analyte
2,3,7,8-TCDF
2,3,7,8-TCDD
1,23,7,8-PeCDF
23,4,7,8-PeCDF
1,23,7,8-PeCDD
1,23,4,7,8-HxCDF
1,2,3,6,7,8-HxCDF
2,3,4,6,7,8-HxCDF
1,23,7,8,9-HxCDF
1,23,4,7,8-HxCDD
1,2,3,6,7,8-HxCDD
1,2,3,7,8,9-HxCDD
1,2,3,4,6,7,8-HpCDF
1,2,3,4,7,8,9-HpCDF
1,2,3,4,6,7,8-HpCD
1,23,4,6,7,8-OCDF
1,2,3,4,7,8,9-OCDD
Labi
N=l
Lab 2
n=4
RSD
Average (%)
74
86
60
62
78
62
72
70
62
58
70
78
100
82
80
81
89
111
91
93
103
120
112
104
105
86
89
93
88
95
120
121
NRd
NR
" Diphenylether interference observed
b Average of two positive quantifiable or trace values
c 1,2,3,4,7,8- and 1,2,3,6,7,8-HxCDD isomers were repoi
2.1
10.0
6.9
6.2
9.7
15.1
13.9
17.2
5.6
9.2
7.7
8.6
6.7
21.3
13.8
-
-
(n=2).
•ted as a sum.
Lab3
n=5
Average
92
92
100
89
88
a
68
78b
74
C
97
91
108
104
90
93
93"
RSD
(%)
4.7
3.4
6.2
9.2
11.5
-
13.2
0.9
13.3
-
12.7
8.8
15.2
10.0
19.3
9.0
29.2
d Not reported.
e Average of four positive quantifiable or trace values (n=4)
71
-------
6.4 SYNOPSIS OF ANALYTICAL RESULTS
The detailed results from the analysis of all samples have been submitted as separate
reports to EPA (Cramer et al. 1989a, 1989b). Appendix A of this report provides detailed
documentation on the results of each FY87 NHATS sample and QC sample. The data in Appen-
dix A identifies: batch number, laboratory identification; NHATS sample identification; number
of specimens in the composite; census region and age group; composition of composite by sex and
race; identification of data quality as positive quantifiable (PQ), trace (TR) or not detected (ND);
measured concentrations and limits of detection; data restrictions; and IQS recoveries.
Successful analyses were achieved for all samples except one, ACD8700425. This sample
appeared to have been fortified with twice the specified amount of IQS. In the remaining
samples, the range of 2,3,7,8-TCDD detected ranged from a nondetect value of 0.0689 pg/g to a
maximum of 15.1 pg/g. Positive quantifiable OCDD ranged from 136 to 1,660 pg/g.
It should also be noted that in many of the analyses, responses to octachlorodiphenyl ethers
(OCDPE) overlapped with the responses of the 1,2,3,4,7,8-HxCDF and the 2,3,4,6,7,8-HxCDF
isomers. In these cases the observed responses were quantitated but were reported as a
nondetected (ND) value.
The recoveries of the internal quantitation standards (IQS) were within the QA data quality
objective of 40% to 150%, with the exceptions that are identified below. As previously noted,
composite ACD8700425 appeared to have been fortified with twice the specified amount of IQS.
Data for this composite are considered suspect. In composites ACD8700023 and ACD8700256, the
recoveries of 13C12-PeCDD (28%) and 13C,2-PeCDF (17%), respectively, were outside the data quality
objectives. Since only one out of nine IQS recoveries per sample were not in control, the samples
were not reanalyzed. The PeCDD and PeCDF data for these two composites were flagged as not
meeting the DQOs. In batches 3 through 5, the recoveries of the carbon-13 labeled HpCDD in
three composites and the labeled octachlorodioxin in 17 composites and/or QC samples were
outside the DQOs.
The analysis results for the spiked control QC samples (Table 6-17) indicated that the
accuracy of the method met the QA objectives of 40% to 150% for all compounds with the
following exceptions: the 1,2,3,4,7,8- and 1,2,3,6,7,8-HxCDD pair in the low level batch 2 spike
(13% recovery); OCDD in the high level batch 3 spike (39%);, 1,2,3,4,6,7,8-HpCDD in the low level
batch 5 spike (32%); and 2,3,4,7,8-PeCDF in the low level batch 4 spike (not recovered). The
PeCDF recovery was affected by a momentary reduction in sensitivity resulting from the apparent
coelution of a high level compound interference. The remaining compounds showed recoveries
ranging from 44% to 137%.
72
-------
£
2
^
£
PH
1
(H ^j
. ^ PH
D ^
u JH
'o D
bu
CQ p t
i
C/>
r-t
I
sO
1
c c
.£ o
3 V4H '-P "^
*C o .*" ff^
Q> 5 ^"^
o 5
U ^
^cJ C
111
a S-&
en TJ
I'g'b
O ** flJ •"—•
&.H, o &
CJ "^ OJ
rt "3 ^H
CQ
C
o
10 w '^ ^^.
rt S S 'oC
I 8 3 IS
< S c ^
0
a3 ^3
11
^
2
111
f
c^osyrj ooino t^^jso co^in
iri^oo ^ TjJ m. m'^1^ ^oC1^
rHOO^ OSrHCO OOOOsO t-xrHSD
CJCJ*^ O^1"1 CJCJ00' rHsO'1*
CO CO in OO ^ 00 Csl rH
ifOrH 'CO'rH 'rHQs 'inOs
Os Os 00 Os *"^ Os tx 00
rQinoq OSSDC^J 2^^ ir1vO.vi?
""Joso" ^tNiri csjrHO incsios
rHrH^H OSrHin ^rHin CSCOsO
tsinin ininin rn^m inmin
csinm inmin in-^m inmin
i— ,OO r- .OO r-^OO riOO
^•'rHin ^rHin ^rHUI "— 'rHin
* D D D
P Q U U
CJ rj o> o»
H H PT ^
oo oo °R. °°v
K tC c\ t\
ro_ co" ^ ^
M CsJ ^ "X
rH N
SO rH O
^ rH OO
|S5 •
o *-!
i o oo <
rH 00
00 00 00 tx
cj o •* oo
CS CO SO rH
m co in rH
in co in rn
,-* o o ^
0 rH in °
_ tu
D D
8 §
-
V [*s,
CO" ^
Vj. co
r^ K
r-H
^ ro
r-5 ^
csq oo
co ^
rH
OS
oq -^
^~^ ^J
rH CO
in ^
tn ^
OS
PH
3,6,7,8-HxCD
c^
rH
in
co
Os
rH
00
£
SO
ol
in
in
a
73
-------
(U
g
u
IX
3
QJ
3
H
4-*
_ §
*2 *o '« S"
^S *"H """^
o 5
u >
73 C
Is ° '~"
"^ '« ifc
43 'E ^
55 TJ
3 QJ IH
O *J 4> - — -
fcri 13 ^ ^^.
•^ "5* 8 — '
to « 2i
pa
-a .°
|i|l
J5 Jj w ""-^
" o
u
^ j
,0 w
g 0
Z^
T3
Q
z
•a
c
TH «* ro os' od so
T— icoos £- ; •*—{ o\ THTHCO cr ^
•^ f? ^ CO" CO" ^>
«N" TH" ^ ^ n
in os in
IX* 't °°.
in *^ os
3 rq oo
in ^ H
o' TH TH
•^ CO
1 TH O
l-H t— 1
oo Os" ro
C5 f^1
IN in in
m m in
P-I
8
OH
oo"
*•
^
1-H~
-------
if
3
1
o
u
[Ss
t-H
SO
$
*j
'0 <4_, V '"~s
o 5
U >
-a c
IB ® '~N
I'^lb
C/2 "Tl
TJ
3 ai ^
o 4S 53 x-^,
1114
»
c
"^ .2
fill
o
(J
J,. ,Q
&> "s
,Q JB
£ °
3 4-*
^
«
z
^^s
CO ^ ^
1
•«*cncq c
co; co^ co- J^ "-
^t *^ °^ TH rH
rH rH TH1
^_^
>
"a>
Tj 05
§ O,
o
^ > 'a,
0 w
•r^ ^Q J> j^
QJ ^3 -J N_ .*• ^3
•|^ b |
S '$ o u
? w g
CA flj *H
^ J " a>
tj O . ^O «*<«
C pT1 n> «3
S S •*•* h
cd s « CL>
* S .H re
°o -g .a
£ S3 n3 T3
"1 i §
3 a O O
c 5 oo oo
u a 3 3
^, rC re re
Z H pa PQ
id £ u
75
-------
6.5 STATISTICAL ANALYSIS OF THE QUALITY CONTROL DATA
The statistical analysis of the FY87 NHATS QC samples for the PCDD and PCDF analytical
results are summarized in Table 6-17. The objectives of the analysis were to
• Estimate the percent recovery of the analytical method,
• Determine if there are significant differences in the analytical performance among
the batches,
• Characterize the precision of the analytical method,
• Establish the relationship between the precision of the analytical method and the
level of the spiked concentration, and
• Identify anomalous results that suggest potential problems in the analytical
measurements.
Of the 68 samples analyzed for PCDDs and PCDFs in the FY87 study, 20 were QC
samples. Each of the five analysis batches contained one method blank, one unspiked control
sample, and two spiked samples. The sampling plan for the allocation of these QC samples has
been described by Heath (1988). Leczynski et al. (1988) determined the assignment of the QC
samples to the analysis batches.
Because it was agreed that population estimates would be calculated using only the data
that met the data quality objectives (DQOs), the same criteria were applied before evaluating the
QC data. The DQO criteria are:
(1) Internal quantitation standard (IQS) recovery must be between 40% and 150%,
(2) Ion ratio must be within 20% of the theoretical ratio, and
(3) There must be no problems with coelution or fragmented peaks.
If an analyte was not detected (ND), the measured concentration of the QC sample was computed
as one half of the detection limit (LOD/2). This same approach was used in the statistical analysis
of the field samples.
A descriptive summary of the QC data is presented in Section 6.5.1. In Section 6.5.2 the
statistical approach to analyzing the QC data is discussed and the results of these analyses are
presented in Section 6.5.3. Conclusions are presented in Section 6.6.4.
76
-------
6.5.1 Summary of the OC
Table 6-17 presents the data for each chemical in the quality control samples. Table 6-17
presents the data by chemical and spike level (pg/g) and provides information on the number of
QC samples for which the DQOs were met, the number of positive quantifiable (PQ) and trace
(TR) measurements, the average measured concentration, the background adjusted recovery
(BAR) for spiked samples, the standard deviation (SD) of the measured concentration, and the
coefficient of variation (CV). The background adjusted recoveries at the low (L) and high (H)
spike levels were computed as
BAR(j) = 100%*[Avg(j) - Avg(O)] / spike level,
where Avg(j) is the average measured concentration at spike level j (j=L or H) and Avg(O) is the
average measured concentration of the unspiked control sample. The coefficient of variation was
computed as
CV(j) = 100%*[SD(j)/Avg(j)]/ for j= 0, L, and H.
All background adjusted recoveries were between 67 and 120% and the coefficients of
variation were generally between 2 and 20% at both the low and high spike levels. In the control
samples analytes at levels below the detection limit of the analytical method generally had the
higher coefficients of variation.
Only one analyte was detected in the method blank The measured concentration of
2,3,7,8-TCDF in the batch 1 method blank was 0.460 pg/g, or 41% of the average measured 2,3,7,8-
TCDF concentration (1.12 pg/g) in the two control samples that met the DQO criteria. Because
this indicated a potential for a bias affecting all batch 1 samples, the measured 2,3,7,8-TCDF
concentrations of the 48 study samples were compared across the five batches. This comparison
did not reveal any statistical evidence of a batch effect on the study samples. Therefore, no
adjustments were made to the measured 2,3,7,8-TCDF concentrations of the study or the QC
samples in batch 1. In particular, all 2,3,7,8-TCDF QC data meeting the original DQOs were
included in the statistical analysis.
Appendix D contains the plots of the measured concentrations against the spike levels for
all study compounds. It is evident from these plots that the relationship between measured
concentration and spiked concentration is nearly linear for all the compounds. Because only one
measurement met the DQO criteria for 1,2,3,4,7,8-HxCDF, a figure is not given for this chemical.
Also presented in these figures are the predicted concentrations with tolerance bounds for
individual measured concentrations. The statistical methods for calculating the predicted values
and tolerance bounds are discussed in the following section.
77
-------
6.5.2 Statistical Approach to Analyzing the QC Data
The QC data were statistically analyzed using linear regression models fitted to the
measured concentrations for each compound. Three models, as described in Table 6-18, were
fitted to the data to determine the best fit and to test for significant batch effects.
Initially the full batch effects (FB) model was fitted for each compound. The FB model was
used to test for two types of batch effects: fixed effects and proportional effects. A fixed batch
effect is the constant amount by which the measured concentrations in the batch differ from the
average for all batches. It is calculated using the intercepts («., i=l,...,5) of the FB model. For
example, the fixed effect of batch 1 is represented by «]-K/ where « is the average intercept.
Proportional batch effects are characterized by differences in the batch recoveries. The
proportional batch effect for batch 1, for example, is the difference between the recovery (slope)
for batch 1 and the average recovery for all batches. Using the notation in Table 6-18, the
proportional effect (also called the recovery effect) for batch 1 is denoted by PrP-
For those analytes providing sufficient data, F-tests were performed to determine the
significance of the fixed and proportional batch effects. In virtually all cases where batch effects
were detected, they were due to variations in the slopes from batch to batch. Therefore, a second
statistical model containing a constant intercept (i.e., a, = <* in the full batch effects model) was
fitted to the data for each analyte. This model is called the batch slopes (BS) model because it can
be used to test for significant differences in the recoveries (slopes) between batches.
The BS model was used to estimate recoveries for each batch, overall average recovery,
and predicted concentrations at each spike level. Statistical F-tests were performed to test for
significant background levels and batch effects. Background levels are indicated when the
estimated intercept is found to be significantly different from zero and batch effects are indicated
when at least one of the estimated batch slopes is found to be significantly different from the
others. Predicted concentrations at each spike level were calculated from the estimated intercept
and average recovery.
For some analytes it was not possible to fit either the FB or BS models because there were
insufficient data after applying the DQO restriction criteria. In these cases a simple linear
regression (SLR) model was used to estimate average recovery and test for significant background
levels. However, using the SLR model, it is not possible to test for significant batch effects.
78
-------
43
re
Q
U
a
&
z
&
>
PH
01
N
X
13
5
o
•d
O)
to
P
tn
"aJ
T3
0
^
.2
"to
so
QJ
&
QJ
rv"
^JH
od
i— i
4
Ol
3
re
H
Remarks
IT
a>
O)
"e 2
C cs
•5 a
fh
#1
i— • AJ
£,
3
o
2
u/
•d MH
0 0
6 &3
.a g en
S* >X
re g fa *
43 S ¥ fi
re <£ -5 o
^ S-d*
O> -iH d -M
* S s p<
+-• 8r D
^ v CJ
C o "o! »H
Ol U -rt W
(«;) are the batch background
pes (PJ) are the batch recoveries. Wh
the average of the estimated batch re
estimated recovery from the SLR mo(
ted intercepts equals the estimated int
y. Oja QJ re
+•> Trl -rt j£ £5
eL,1-" s ^ g
§ - S-ilN
^ "S J 1 % "S
*J S» TJ i O> O
^ 1 2 ir-5 g
or
'•a,—
^r «
^*- IH
, Ol
-4- -i-1
T a>
«rtS
n^l
re
:srP,
O o
m '-'
M" ^<^^
tfi
4^
U
QJ
M-t
M-l
X
W
re
CO
^ «"
rr ^
MH ^— '
UJ
•d
0 T3
6^
W3
g o> __.
O O) O)
*•£"§
SI "8 6
ra S § c^
43 > -»->
re o w ^
•rt W -rt (..
a) is the background level.
are the batch recoveries. When the <
the average of the estimated batch re
estimated recovery from the SLR mo<
quals the estimated intercept from th(
•~""^H» O) 0>
•>-> a. T? x; -M
CL.V — £> ^ fx,
Hi-iS
QJ g-i re 3 o>
•H^ia &<•£
,5 en ^» o> .5
or
en
ca^
+ ^
Ol
« g
II s
re
r:F°-
W^
'fll
U/
"G
O
•*H
2
CA
Ol
{X
175
•s~
•j± en
Cu rv\
» S-
a) is the background level.
s the recovery.
N^^ «^H
4^ y™*\
PnCO-
^
S3 &,
•s o
c '•"^
w en
u"
en ^
ca. «
+ S
Ol
« g
!L§
,:P&H
wSS
S"
•H B?
re -£-
Ol
.s §
J -S
w
0> VI
f-H Q>
& K
g S5
en csj
• *H
-6
.4-t
re
,£>
M-l
O
Ol
D-,
g
re
m
U
a
x
*3
1
«.s
IL> C
> o
Ji V u
ill
_d C CJ
|>sa
^^ ^ P^M Q^ (— ;
S T3 X H.
aj g ^_i ••— N
•5 «^^
*± v ° •£
^ y a ^
^-^ s "S
£ -1-3
^ ^^ ?
re ^o w ^
TJH- _? ol S
m- S 8^^
•S oj ^
_i- C
,° II II
l-H U .I_<
nil rT'r'^
II *sJ U
.rt .— ,W en
n
°i
£
OH
to
X
bO
2
"S
re
4
a,
!0
^
^O
1-J*
o
&
C
8
MH
O
52
01
nO
S
13
3
cr1
01
K
re
a
o>
•S
IH
01
Ol
1
^
•d
I
43 «
(y *C
-O QJ
£J 'C
re u
« o
re O
T3Q
_£ «
^s^
c" -g
•S 1
•+-J C
re c
o ^
A re
^-S
CH™
re ^
•3-S
,2 a.
* g
G re
>-3 (0
J3
79
-------
Although the analysis established that there were statistically significant batch effects for
most of the analytes it was decided that there would be no adjustments made to the measured
concentrations of the study samples. Thus, a model was developed in which the differences in
recoveries from batch to batch were treated as random effects; thus affecting the precision of the
analytical method. The model assumes that the standard deviation of the measured concentration
has two components: (1) a component associated with the within-batch measurement error,
estimated by the mean squared error (MSB) from the BS model; and (2) a random component
associated with the random-batch effects. The standard deviation of the ith measured
concentration (t^) at spiked concentration SCj was computed as:
SD(C,,) =
where MSB is the mean squared error from the BS model, and SD(pj) is the sample standard
deviation of the estimated batch recoveries. According to this model, the standard deviation
increases with the concentration of the sample; however, it is not necessarily proportional to the
concentration. When there was insufficient data to fit the BS model, the square root of the MSB
from the SLR model was used to estimate the standard deviation of the predicted concentration.
Approximate prediction bounds on the individual measured concentrations were calculated
by adding plus or minus three times the estimated standard deviation to the estimated predicted
concentration. The probability that an individual measured concentration will fall within the
prediction interval is approximately 99% according to asymptotic distribution theory.
6.5.3 Results
The results of the statistical analyses are summarized in Tables 6-19 and 6-20. Table 6-19
contains an estimate of the average recovery, its standard error (SB), and the estimated batch
recoveries for each analyte. When there was sufficient data, the individual batch recoveries were
estimated from the BS model, otherwise the SLR model was used to calculate the average
recovery.
For each compound, a hypothesis test was performed to determine if the average recovery
was significantly different (at the 5% significance level) from 100%. The result of this hypothesis
test is denoted by an " * " next to the estimated average recovery. The average recoveries were
determined to be significantly less than 100% for nine compounds, and the lowest average
recovery was estimated to be 66.1% for 1,2,3,6,7,8-HxCDF. There was only one compound,
1,2,3,4,6,7,8-HpCDF, for which the average recovery is statistically greater than 100%.
80
-------
PQ
cu
P;
cu
bo
E
cu
•a
0)
4-1
re
_g
cn
H
£
re
CO
w
v2
cn
cu
cu
O
CU
.1
W
3
cu
cn
.**
cu
$
w
ca
cu
V
II
cn
W
OT ,2
in
CC
CM
o
cu
13
^
8 &
CN|
rq
o6o\
OO CN
sO tv
IN C>
INOO
\o
so rq
00
sO
odo'ocirQ oo IN in rsiomcsc)
CX>O\tsOO Tfl ITlrH OrHOOs
rq co CN
oo
in
tx tvlft rHOOOOCO
so tNO CTvOOON
o\iNOtNC~qtvt7\rOTij in
tx OO so sO O\'& CSl 00 ON in IN SO ^* tx
O O ON IN SO OO O\ OO OO ON rH rHOO
CO
00
O
^ CT*\ f^ s^ ^^ ^*i ^H l^
-Hr-irHTl5tNit-s;rqiri
* *
CO ^
* * * *
r— IsOCSlsD rH
CO
rHtNOOtN SO^COOO OtNCOrHCOrH
CT^ CT^ O^ CTv OO ^O [^ r^* O^ O^ ^z!3 f~^ ^\ Q\ ^H
•3.
PH
b tin P PH P
PgPP PgP
o u u u , o
PH P
P P
U U
H H
°o<
IN
IN IN co
CO rH tSl
*^NOVOI
CO CfT Cff T)J I
rn" rn" rn" rq" <
S ocj^ c>v 06^ od od
,4,7,8-HxCDDc
,6,7,8-HxCDDc
m" cr cr co co
CN
cu
9
&
•8
c
cu
"S
cu
bo
2
cu
re
re
%
cn
cu
u *
#3$
^ « I
O^cl
** bo JJ
1-a-g
O 3 *
S^-2
i
in
cu
re
.8
£
j=
re
en
CU
rt *-
£> be
cn C
c?£
^=5
^ S
£ g g* fi«fi
.12 "PH MM frt "^
(••4 *!"• »—M
tT3 J-< rt»
W3
OJ
firH
re
»
cu
.x
en
•^•3
cu
o
ent
•sis6
llili
£'$>
•%£ < £
« •£
re .jo
•° "g
J-j *-!
C «
SJ cn
'C Si
UH re
^ .a
cn TJ
C C
81
-------
OJ
0)
<3J
'&,
c/i
-6
re
PJ
re
o
•a
re
•c
re
C
cu
•o
QJ
o
u
re
'SB
ex
-------
The individual estimated batch recoveries are shown in the remaining columns of Table 6-19.
Also presented are the results of the hypothesis tests for differences among the batch recoveries.
Notice that there are significant batch effects for virtually all analytes. However, because there
were no apparent patterns to the batch effects, it was decided to treat the batch effects as random.
This means that no "batch" corrections were made to the measured concentrations of the study
samples. Instead, as discussed in the previous section, the batch effects were treated as random
and included in the estimated precision of the analytical method.
Table 6-20 contains the model-derived predicted average concentration and estimated
coefficient of variation (CV) for each analyte and spike level. For each analyte, a hypothesis test
was performed to determine if the predicted concentration in the control sample was greater than
zero at the 5% significance level. For example, the predicted concentration of 2,3,7,8-TCDD in the
control samples was estimated to be 8.90 pg/g and this estimate is significantly greater than zero
at the 5% significance level. This is consistent with the fact that 2,3,7,8-TCDD was detected in all
five control samples. The background concentrations were determined to be significantly greater
than zero for six of the seven PCDDs (not including 1,2,3,4,7,8/1,2,3,6,7,8-HxCDD because the
individual isomers are already represented) and for four of the nine PCDFs (1,2,3,4,7,8-HxCDF is
not included because of insufficient data).
In general, the relative precision of measured concentrations is much better for PCDDs than
PCDFs. At the control level, the CVs of the measured concentrations for the four PCDFs which
had significant background levels were between 13% and 48%. All of the CVs for the PCDFs
were less than 21% in the spiked samples. The CVs at the control level for the six PCDDs with
significant background levels were between 5% and 20%. Although this might be explained by
the higher concentration levels of the PCDDs, the relative precision for measuring PCDDs is also
much better in the spiked samples. With the exception of 1,2,3,4,7,8-HxCDD, the CVs for PCDD
measurements are between 3% and 10% in spiked samples while the CVs for the PCDF
measurements in spiked samples were between 2% and 21%.
In general, the linear model provided a good fit to the measured concentrations for most
analytes. This is evident in the figures presented in Appendix D. There does appear to be some
lack of fit for compounds 1,2,3,6,7,8-HxCDF and 1,2,3,4,7,8-HxCDD. The lack of fit for
1,2,3,6,7,8-HxCDF is most probably due to the rather low measured concentration of 73 pg/g in
batch 3 at the high spike level. The measured concentrations for 1,2,3,4,7,8-HxCDD were reported
as a combined value with the 1,2,3,6,7,8-HxCDD and therefore could not be effectively evaluated
using the linear model.
83
-------
6.5.4 Summary of QC Data
The results from the statistical analysis of the QC data are summarized as follows:
1. Nine of the analytes had estimated average recoveries that were found to be less than
100% at the 5% significance level and one analyte had an estimated average recovery
that was significantly greater than 100%. The estimated average recoveries were
between 66.1% and 124% for all analytes.
2. There are statistically significant differences in the recoveries from batch to batch for
most of the analytes. However, it is recommended that no batch adjustment be made
to the study samples. Instead, the estimated measurement precision will account for
the batch effects.
3. Measurement precision, determined by the estimated coefficients of variation in the
control samples, is generally between 5% and 20% for the PCDDs and between 13%
and 48% for the PCDFs. For the spiked samples, the PCDDs had CVs between 3% and
10%, and the PCDFs had CVs between 2% and 21%. These estimated CVs include
random batch effects.
4. Statistically significant background levels of four PCDFs and six PCDDs were identified
in the control samples.
5. The relationship between measured and spiked concentrations is generally linear over
the range of spiked concentrations.
84
-------
7.0 STATISTICAL METHODOLOGY
There were three objectives for the statistical analysis of FY87 NHATS data:
1. Estimate average concentration levels in the adipose tissue of individuals in the
U.S. population and in various demographic subpopulations,
2. Construct confidence intervals for these averages, and
3. Determine if average concentration levels of chlorinated dioxins and furans in the
U.S. population differ significantly by any of the four demographic factors
(geographic region, age, race, and sex).
The statistical analysis methods used in this report are based on an additive model for the
demographic effects. Previous studies of the effect of using composite samples demonstrated the
validity of the additive model. A technique of iterative weighted generalized least squares was
used to estimate model parameters. The resulting estimates are approximately normal for large
samples. This approximate normality was used in constructing confidence intervals and
hypothesis tests. The remainder of this section provides details of the statistical model and
process as well as references to background work.
7.1 STATISTICAL MODEL
The use of composite samples for determination of the levels of PCDDs and PCDFs created
a need to reevaluate the approach to estimate general population levels of these compounds. The
statistical models previously used to assess NHATS data for OC1 pesticides and PCBs using
individual sample data were not adequate for extrapolating the composite sample data. Section
7.1.1 discusses the background on the development of a new statistical model, the additive model,
which is presented in Section 7.1.2.
7.1.1 Background
Mack and Panebianco (1986) developed and used a "multiplicative" model to analyze the
composite sample data from the NHATS FY82 Broad Scan Analysis Study. In their model the
analyte concentrations in a composite sample are represented as a product of fixed and random
effects associated with geographic and demographic characteristics of individuals who contributed
specimens to the samples. Orban et al. (1987) studied this problem further and recommended the
"additive" model which assumes additive effects of the donors' geographic and demographic
characteristics.
85
-------
The multiplicative and additive models were later evaluated by Orban and Lordo (1989).
They showed that under certain distributional assumptions, both models produce asymptotically
unbiased estimates of average concentration levels in the target populations. However, the
multiplicative model requires that the sampling and measurement errors be distributed according
to lognormal distributions. No specific distributional forms are required to achieve asymptotically
unbiased estimates using the additive model. Orban and Lordo (1989) also compared the two
models using simulated composite sample data which were generated from actual specimen data
obtained in the FY83 NHATS. Their analysis demonstrated, using actual NHATS data, that the
standard errors of the estimates from these models are nearly equal.
Following the study, the additive model was chosen to be used in the FY87 NHATS and
all future NHATS for the following reasons: (1) under very general assumptions, the additive
model produces asymptotically unbiased estimates of average concentration levels in the
population, and (2) the additive model establishes a more tractable relationship between the
distribution of analyte concentrations in individuals and the distribution of measured
concentrations from the composite samples. The second reason is particularly important because
individual specimens are collected but composites are chemically analyzed.
7.1.2 The Additive Model
Table 7-1 lists the categories of the four analysis factors of interest to NHATS. The
additive model assumes that the four analysis factors have fixed additive effects on the average
concentrations in specimens. This assumption creates 48 subpopulations defined by the
4x3x2x2 unique combinations of categories.
Table 7-1. NHATS Analysis Factors and Categories
Analysis factor Category
Census region Northeast
North Central
South
West
Age 0-14 years
15-44 years
45+ years
Sex Male
Female
Race group Caucasian
Noncaucasian
86
-------
In addition to the four analysis factors, there are three ancillary factors that are assumed
to have random effects on NHATS data. They are (1) sampling of MSAs, (2) sampling of
individuals within MSAs, and (3) measurement of analyte concentrations in the composite
samples. The second factor, sampling individuals from MSAs, also includes the effects of selecting
specimens from individual donors. The first two ancillary factors have random effects on the
actual concentrations in individual specimens, and the third has a random effect on the measured
concentrations of composites.
From the assumptions above, the actual concentration in a specimen from the i-th donor
in MSA j, census region k, age group t, sex m, and race group n, is represented by
CRk + \ + sm + *n + MSAJ
where
\i is a constant
CRk is the fixed effect of census region k; k = 1,...,4
A, is the fixed effect of age group f; { = 1,2,3
Sm is the fixed effect of sex m; m = 1,2
Rn is the fixed effect of race group n; n = 1,2
MSAj is the random effect of selecting MSA j; j = 1,2,...
GJJ is the random effect of selecting individual i in MSA j.
Furthermore, to uniquely define the parameters, we let
£=1 fi=l m=l n=l
The random effects MSAjk and eijk are assumed to have independent error distributions
with mean zero and variances o^ and o2, respectively. Also, because of evidence from previous
NHATS and other environmental studies that the variation in specimen concentrations increases
with average concentration levels, it is assumed that o2 = b2|i2, where (is is the average
concentration level is subpopulation s. For notational simplicity we let
87
-------
CRk + At + Sm + Rn
for some particular combination of indices k, {, m, and n.
This defines the model for the actual concentration in a specimen collected in the survey.
However, the specimens are composited prior to chemical analysis. Thus, data are obtained from
the chemical analyses of composite samples. Letting Yh represent the measured concentration of
composite h (h = 1,. . ., C), the natural additive effects of compositing imply that
where Cijs is the actual concentration in specimen i from MSA j and subpopulation s; yh is a
random measurement error; Mh is the number of specimens in composite h; and Ch(i,j,s) is equal
to 1 if specimen (i,j,s) is in composite h, and is equal to zero, otherwise. The error distribution
of yh has mean zero and variance o^.
At this point the notation for representing the model is rather complex. However, the
main points can be illustrated using matrix notation. Let
, Av A2, Sv
be the 8x1 vector of fixed effects and /.A = (/i,,. . . , ju,48)' be a 48x1 vector representing the
48 subpopulation average concentrations. Then
where X is a 48x8 design matrix. Letting Y = (Y]7. . ., Yc)' be the Cxi vector of measured
composite concentrations, Orban and Lordo (1989) show that the expected value of Y is
88
-------
E(Y) = ZXp = Dp ,
where Z is a Cx48 composite design matrix. Thus, according to the additive model, both the
actual concentrations of the individual specimens and the measured concentrations of the
composite samples have expected values that are linear combinations of the additive effects of the
fixed analysis factors.
Orban and Lordo (1989) also show that the variance-covariance matrix of Y is a block
diagonal matrix that depends on 0^,0^, and o^.
7.2 STATISTICAL ANALYSIS
This section describes the specific methods used to achieve the statistical objectives. The
estimation methods are discussed in Section 7.2.1 and the hypothesis testing procedures are
presented in Section 7.2.2.
7.2.1 Estimation
The specific quantities estimated for the FY87 NHATS are the average analyte
concentrations in the adipose tissue of the U.S. population and the averages for each of the
marginal populations defined by the categories listed in Table 7-1. The estimates were calculated
in three steps:
1. The additive model parameters associated with the four analysis factors were
estimated using a method called iterative weighted generalized least squares
(IWGLS).
2. Estimates of average concentration levels in all 48 subpopulations defined by the
analysis factors were calculated from the parameter estimates.
3. National and marginal population estimates were obtained by taking weighted
averages of the appropriate subpopulation estimates. Weights were proportional
to the population counts of the 48 subpopulations from the 1980 U.S. census.
89
-------
According to the model described in Section 7.1.2, the vector of measured composite
sample concentrations, denoted by
Y = (Ylt...tYJr ,
has a multivariate distribution with mean
E(Y) =
and a variance-covariance matrix V. The vector
p = (n, CRr C/^, C/^, Av A
2,
is the vector of fixed effects to be estimated.
To obtain asymptotically unbiased estimates of the fixed effects it is not necessary to make
any assumptions about the form of the distributions of the random effects. If the variance-
covariance matrix V were known, the method of generalized least squares (GLS) produces
estimates of p that are unbiased and have minimum variance among all unbiased estimates.
Furthermore, if the errors are normally distributed, the GLS estimates are equivalent to the
maximum likelihood estimates. The GLS estimate of p is
P = (D'V-lDylD'V~lY
Unfortunately, V depends on three unknown variance components (o^, o^, and o^) as well as the
vector p. Therefore, Orban and Lordo (1989) proposed a method involving iterative weighting.
Thus, the method is called iterative weighted generalized least squares (IWGLS).
90
-------
Starting values for the fixed effect parameters and estimates of the variance components
were calculated using a maximum likelihood procedure. This step was performed using the P3V
program in the BMDP program package. The resulting estimate of V was then used in the GLS
formula to produce a revised estimate of p. Each time the GLS formula was applied, the estimate
of V was updated. This process was continued until certain convergence criteria were met.
Orban and Lordo (1989) discuss this method in more detail and describe special computer
programs for implementing IWGLS. They also provide formulas for calculating the standard
errors of the estimates.
An estimate of the average concentration level in each of the 48 subpopulations was then
calculated by
Weighted averages of the appropriate subpopulation predicted concentrations were then
calculated to estimate marginal averages for the categories of each analysis factor. For example,
the average concentration in the Northeast census region was estimated by the weighted average
of predicted concentrations in all subpopulations in the Northeast region. Marginal estimates
were calculated for four census regions, three age groups, two sexes, and two race groups. The
U.S. population estimate was calculated in a similar manner. An approximate 95% confidence
interval for each estimate was calculated by adding and subtracting two times the standard error
of the estimate.
7.2.2 Hypothesis Testing
Hypothesis tests were performed to determine if average concentration levels differ
significantly by any of the geographic or demographic factors. The specific hypotheses tested
were:
H,: CR, = CR2 = CR3 = CR4 = 0,
H2: A, = A2 = A3 = 0,
H3: Rj = R2 = 0, and
H4: S, = S2 = 0.
91
-------
The hypothesis, H17 for example, states that there are no differences in average concentration
levels among the four census regions. The alternative hypothesis is that there is at least one pair
of regions for which the average concentrations are different. The results from these hypothesis
tests are somewhat related to the confidence intervals for averages in individual subpopulations.
Generally, if the confidence intervals for any pair of subpopulation averages do not intersect, then
the hypothesis of no differences among subpopulations is likely to be rejected. However, the
contrapositive is not always true.
In order to test these hypotheses, it was necessary to make specific distribution
assumptions for the random effects. It was assumed that the errors associated with sampling
MSAs, sampling individuals within MSAs, and measuring concentrations were independent and
normally distributed. The additive effect of compositing specimens suggests that the normality
assumption for sampling error is reasonable because concentrations of individuals are averaged
in the composite sample. Statistical theory states that averages and sums are approximately
normally distributed. Distributional assumptions were tested for all analytes using probability
plots and residual analysis. The model validation results are discussed later in Section 8.5.
The likelihood ratio method was used to test hypotheses Hj through H4. According to
asymptotic theory, the log of the ratio of the likelihood functions (calculated under the full and
null hypothesis models) has approximately a chi squared distribution. The number of degrees of
freedom is equal to the number of independent parameters constrained under the null hypothesis.
Orban and Lordo (1989) wrote computer programs to perform these tests.
92
-------
8.0 RESULTS
This section contains the results of the statistical analysis of the FY87 NHATS for
PCDDs and PCDFs in human adipose tissue. The analysis was performed on data obtained
from 48 composite samples, each containing an average of 18 adipose tissue specimens from
sampled cadavers and surgical patients.
A descriptive summary of the data is provided in Section 8.1 and the results of the
formal statistical analyses are presented in Sections 8.2 and 8.3. Section 8.4 provides estimated
rates of change of selected PCDD and PCDF concentrations. Section 8.5 describes the outlier
detection procedures that identified potential data errors to be corrected prior to conducting
the statistical analysis. Finally, Section 8.6 discusses the steps that were taken to demonstrate
the validity of the statistical methodology applied to the FY87 NHATS data.
8.1 DATA RESTRICTIONS AND DESCRIPTIVE STATISTICS
Prior to conducting the statistical analysis, the data were classified according to the
specified data restrictions and data qualifiers. For each of the target analytes, Table 8-1 shows
the number of composite samples for which the measured concentrations were restricted or
qualified.
Data restrictions indicate whether specific data quality objectives (DQOs) were met
during chemical analysis. In the data listings of Appendix A, the data restrictions are indicated
by C, coelution; F, fragmented peaks; I, ion ratio criterion not met; P, peak separation; and R,
IQS recovery criterion not met. If any of the data restrictions were noted for a particular
sample and analyte, the measured concentration was not included in the data summaries or
statistical analyses. For example, 15 of the 48 composites failed at least one of the DQOs for
2,3,7,8-TCDF. Thus, there are only 33 unrestricted measurements. Preliminary analyses
demonstrated that significant biases can occur if restricted measurements are included in the
population estimates.
Data qualifiers are defined in terms of the analytical method's limit of detection (LOD)
for each analyte. The analyte is reported as not detected (ND) if the measured concentration
is below the LOD, trace (TR) if it is between the LOD and five times the LOD, and positive
quantifiable (PQ) if it is greater than five times the LOD. Measured concentrations are
reported only for detected (i.e., TR and PQ) analytes. Table 8-1 shows the number of PQ, TR,
and ND measurements for each- analyte among the unrestricted composites. For example, of
the 33 unrestricted measurements of 2,3,7,8-TCDF, 32 were positive quantifiable and one was a
trace value.
93
-------
S
"E,
to
(LI
J
PH
C/}
oo
01
-s
a
a
5
(0
D
•d
S
o
_
D
§
U
r
U-.
a
8
•B
%
•a
I
8
o
Is
£
"o
j-.s
(D t
.g
e
I
•>
11
C-t to
C Qj
u o
JH CD
Q; 43
•a
si
.£ .0
c '_£*
a S
D
D
a
a
C-,
>H T3
^ "£
c g
o3 g)
1.1
"C .F^l
_^
01 C
°" Cfl
M a
•B c
_3 w
cS k
>>
c
<-
8O.T«intxOVcNOOcNO t>
ON rH ON ON OO ON rH f""i QN
cONDcOr-iin.-,.txl>.ini-( ON
cOcOTj
CO OO t"x t** tx "*^ ^D VD t^ ^* CO t**»
C"** C^ C<^ ^^ CO CO CO ^* CO CO CN CO^
(rTco^r^covc H I J
|i|| |
cr1 C jS 8"1
jrt O I) « w
"^ § P^ -S * ^
4_> *4J V ,- ?..l^
1 ill "I
's n '•§ S "S 'g
^ ^2 a jg .a ex
5 .y Si 1 | ^
^ fi -s w § a
TJ § 0 ° ^ "S
.S oj ^ S " h
u ri ^Q pj j3
JlS'lcx f|
" -° u "° " -
94
-------
Table 8-2 gives a summary of the unrestricted data for FY87 NHATS PCDDs and PCDFs.
The average, standard deviation, minimum, median, and maximum concentrations were
calculated from the unrestricted measurements. However, in calculating these statistics, the
value of LOD/2 was used in place of the measured concentration whenever the analyte was
not detected. In some cases, the minimum is reported as < LOD, where LOD is the smallest
detection limit reported for samples in which the analyte was not detected. The average LOD
for all samples and the percent detected are also presented. Detailed data summaries are
provided in Appendix E.
The results presented in Table 8-2 and Appendix E are based on simple unweighted
averages of the measured concentrations from the composite samples. As such they only
summarize the data. Statistical conclusions and estimates of population average concentration
levels should only be based on the statistical analysis presented in Sections 8.3 and 8.4.
8.2 POPULATION ESTIMATES
Not all of the analytes provided sufficient data to warrant a meaningful statistical
analysis. Two criteria were used to determine which analytes would be statistically analyzed.
First, the analyte must be detected (TR or PQ) in at least 50% of the unrestricted composites.
This ensures that there will be minimal bias caused by substituting LOD/2 for the measured
concentration whenever the analyte was not detected by the analytical method. Also, because
sufficient data are needed to estimate model parameters associated with the four analysis
factors and three variance analytes, it was decided that a minimum of 30 unrestricted
measurements was needed to achieve sufficient precision for the population estimates. Thus,
of the original 16 analytes (the pair 1,2,3,4,7,8-HxCDD and 1,2,3,6,7,8-HxCDD is counted as one
analyte because they could not be separated in most samples) there were nine that met both
criteria for performing statistical analyses.
For each of the nine analytes analyzed statistically, Table 8-3 lists the estimated average
concentration in the entire U.S. population and in each of the categories defined by the four
analysis factors. Also presented is the relative standard error (percent) for each estimate. The
estimates and standard errors are based on the additive model analysis described in Section
7.2. The estimates are asymptotically unbiased and were adjusted for population percentages
based on the 1980 U.S. Census.
95
-------
to
jy
!"
(0
C/}
^
1
B
o
U
1
tx,,
0
v*-
to
P
T3
I
C
"o
b
(0
B
3
oi
CO
2
(0
H
E
I
*J3
"S 5
o
w
c
E e
c .S
1
c
>
•ir
d
V
oq
00
rH
S
m
"
o
m
B*
P
U
£
00
ex"
^
CS"
^omSg oinS M
i* t*N <^ C\1 rH ^* f'd f\j rH O rO O
iSlrHrH K CSICO S^S
ON r-j m
O tx iri O O VO' r-I t>s* O O rn CO
rn [x ^H rn rn cO
rH OO
*• * * * *
**vOts,O **rH O
rH f^ in
vvvvv vvv v~
r-t 00 fO
inSKSSrH OVOC3 ON
vo T(* co o o c4 m tx o tx r^ ON
•^1 in rH
^T
tx ^xf rH
CO c5 tx tx *^ C\1 00 ON tx O
rHodirioo'c4 c4 iri o ON c4 vo
rH CO rn T-I rn C)
&S8vS"8 &S^SS8
vo „
Stxr^ptocQ o385ofetx
Q B. B. D
(b B, B. B< pPPPP
pPQPQ uP^vJv^
OH oo oox oq^ ox o& tx ox tC oo; bC
oOit>TcCtxioovcxivc?oovvp^rCvps «
of of of m" of of rJ (sf of of of y y
P
O
•s
o 1
ex • o-
^i CQ Q>
= RE
to v^ / n
60 "J B
C x-=
•J3 ^ C
S -a "B
Q U O
« J2 -S
« D.
CL, ^ ^
(9 t/1
W M 5$
"S 8 -2 S
s"fi §
8ti ti o
, S § B
IT! § c
° "u w i>
1 1 D §
' Number of i
' Percent of u
' Not detected
f The minimu
96
-------
"£
Q> ^
1 2
g U U
•BT, I
2 c ">
la I
S D B.
fl? t
oo W
U
u
U
z
w C
™ .°
Analyte
Population perc
•
'
® |g ^6 I
IS
r-5 O ON O Ifj
•>? °°
^, ^^ ^,
1^ in1** i— 1*«* T-H
tx ^H *-
oo pr °
P
Q
UJ
y
>
rt
CM"
§
u
Q>
o-
00
tC
g
Q
00 >-T
s& t«.- cA
tx vO CO
^T m t>J
m^ «t "
ri" r-f rt
"
NO
"*
OO
tc
VO
S
UH
8
a
3
D
C
u
ao
J1 c
oo o
J3 -a
o> J3
pe
conc
tand
population
average
relative
97
-------
The standard errors are used to characterize the statistical uncertainty in the individual
estimates. Uncertainty of an estimate is best expressed by calculating a confidence interval.
Approximate 95% confidence intervals are calculated by adding plus or minus two times the
standard error to the estimate. For example, the national average concentration of 2,3,7,8-
TCDD was estimated to be 5.38 pg/g with a relative standard error of 6%. Thus the
approximate 95% confidence interval for the national average is 5.38 ± 0.65 pg/g (4.73 to 6.03
pg/g calculated as 5.38 ± 2x0.06x5.38).
Estimates of the average concentrations in the population categories defined by the four
analysis factors are presented even if the effects of those factors were not found to be
statistically significant. For example, the regional estimates of 2,3,7,8-TCDD average
concentrations range from 4.54 pg/g in the West to 6.02 pg/g in the North East. However, as
will be discussed in Section 8.3, the differences among regions for 2,3,7,8-TCDD were not
found to be statistically significant.
The age group estimates in Table 8-3 suggest that the concentrations of nearly all
analytes increase with the age of the donor. All of the analytes occur at higher concentrations
in the oldest age group (45+ years). However, conclusions about the age group effects are
based on the statistical tests discussed in the next section.
8.3 HYPOTHESIS TESTING
Statistical hypothesis tests were conducted for each of the target analytes to determine if
there are statistically significant differences in average concentrations between individuals
from different geographic regions, age groups, sex groups, and race groups. The tests were
based on likelihood ratio tests using the additive model analysis as described in Section 7.2.
Table 8-4 lists the attained significance levels for the tests associated with the four
analysis factors. The attained significance level is the smallest significance level that will result
in rejection of the hypothesis that there are no differences between population averages. For
example, the differences among estimated averages of 2,3,7,8-TCDD in the four census regions
could only be considered significant at the 0.15 level of significance. On the other hand, the
differences in age group averages are significant at the 0.002 level. A 5% (0.05) level of
significance is generally the smallest level used to declare statistical significance.
It is clear from Table 8-4 that there are statistically significant differences among the
average concentrations in the three age groups. The differences in concentrations between
each age group for each compound, except TCDF, were found to be significant. For each of
the nine analytes (Table 8-3), the highest average concentration is found in the oldest age
98
-------
«.
o
<4H
en
OH
3
o
IH
U
0
2
a,
re
K
oo
O
g
Ol
£-
rH
O>
O)
i
£
O)
CQ
en en
01 PH
C-> ^-\
s R
S u
X Pi
O>
S -2
D %
»H en
^8
-s
0) O,
*•" tx
tn OO
**-< N^.
tn P"1
OJ U-i
•5 tn
0 H
PL, ^^
£|
g2
0
viS
»— 1
o>
0)
,_)
0)
CJ
re
1
in
3
01
3
&
2
D
3
T3
J3
CJ
£
w
u
a,
3
S
o
8
rrt
c2
u
a,
3
O
IH
U
><
Ol
en
,£)
§^
2
CJ
o>
^
"d
M
0
•a
S
en
3
en
d
o>
u
*
X
1
<
csj I-H m
tx •** o
SO CO SO
O O O
oo es co
rH CO O
OS N SO
O 0 0
» 2- !i-i
—O T*™1 ^^
goo
§00
o ° °
V V
0 t-x ^H
in •«* S
rH tN o
0 O o
u, o R
Q Q w
U U PH
H H oo
oo oo tx
tC tC Tj?
co co co"
rsf Csl tN
co in
Os Os
in K
o o
tx 00
Csl •*
SO SO
O O
V V
rH rH
^*i f~~i
f~*t ^**i
O O
V V
O IX
SO •>*
Os OO
0 0
-PeCDD
,8-HxCDF
oo tx
tx so
co" co"
rJ r4"
rH rH
£
CO
0
•^
co
Os
o
«
rH
g
O
V
1
o
,87
7,8-HxCDD
Is. so
Tf CO"
co cxf
rq TH"
rH
8
o
tx
ON
in
0
V
^^
o
o
0
V
Os
00
cq
0
,9-HxCDD
oo
tx
co
Csf
rH
co o
os rsi
co in
o o
Ttj in
(S OS
Tt< rjn
C3 O
V P
d
o
T3
OJ
en
re
^»
^J
en
£5
o
•43
f3
VH
•n
o
2
3
_S
d
j
u
'"—5 r— 3 i i
d> nj *Tj
> > >
Ol Ol 0)
f"H »— ^ ^^
^ rH 0
C3 O rH
O O O
Ol Ol a>
& A Jl
«-• .M HH
-t-1 H-» 4_(
re re re
a> 01 a>
£ r y
" c d
.5 ,3 _S
'53 'S3 '53
en en en
S Si |
S 5 S
^ fl ^
^J ^3 'O
d d d
HH h— ( ^15
•o v ^_
99
-------
group (45+ years) and, except for TCDF, the lowest is found in the youngest age group (0-14
years).
The only other statistically 'significant finding, at the 0.05 level of significance, was that
there are possible regional differences in the average concentrations of 2,3,4,7,8-PeCDF. The
average concentration in the Western census region was estimated to be 4.49 pg/g compared
to the national average of 9.70 pg/g. The highest concentrations of 2,3,4,7,8-PeCDF were
found in the North East census region (13.7 pg/g). This regional effect is discussed further in
Section 9.0 in consideration of the comparison of the FY87 and FY82 broad scan effort.
Some potential difference in the estimated average concentration between Caucasian and
non-caucasian and males and females are noted for each of the PCDDs and PCDFs presented
in Table 8-3. However, these differences were not statistically significant for any of the
modeled compounds based on the statistical hypotesis tests which are summarized in
Table 8-4.
8.4 ESTIMATED RATES OF CHANGE OF SELECTED FCDD AND FCDF
CONCENTRATIONS
The additive model analysis compared mean concentrations across three age groups,
lower (0-14 years), middle (15-44 years) and upper (45+ years). The analysis showed that
significant differences existed among mean age group concentrations for the nine modeled
analytes. However, the analysis did not quantify rates of change between ages. Therefore, a
second set of analyses using linear regression was performed to address this issue.
For each analyte, the measured concentration in each composite sample was regressed
against the mean age of all individuals whose specimens had been pooled into that composite.
Two linear regressions were performed, the first to estimate the average rate of change in
concentration levels from the lower to the middle age groups and the second to estimate the
average rate from the middle to the upper age group. These rates were taken as the
regression slopes times ten to convert them into rates of pg/g per decade. The plot of the
measured concentration of 2,3,7,8-TCDD versus average age with estimated regression lines is
presented in Figure 8-1.
The average ages taken over all composites from the lower, middle, and upper age
groups were 3.0, 30.8, and 65.0 years, respectively. Therefore, the first rate was the estimated
average rate of change per decade from ages 0-31 years, and the second rate was the
estimated average rate of change per decade from ages 32+ years. These rates were further
standardized by dividing them by the arithmetic average concentration in the first age group
100
-------
o
CO
— o
o
in
— o
o
ro
O
e\j
i I i i
\o
CO CM ~H o
«-H «-H <-H i—I
O/Od' uopiRaueouoo
101
01
W)
ro
QJ
bo
CD
en
3
QJ
>
G
O
U
o
•43
QJ
cc
C
O
-------
within each regression set. That is, the first rate was standardized by dividing by the mean
concentration for the lower age group, and the second rate was standardized by dividing by
the mean concentration for the middle age group. The standardized rates of change were
then reported as rates of change per decade, relative to the mean concentration at the
beginning of the time interval analyzed. Because the average ages tend to cluster around
three points, it is not possible to characterize how the rates change over time. The analysis
only provide estimates of the average rates in the two age intervals.
Table 8-5 displays national average concentrations for selected PCDDs and PCDFs, with
the two rates of changes expressed as pg/g per decade and percentage of initial mean
concentrations per decade, respectively. For example, the national average concentration of
2,3,7,8-TCDD is 5.38 pg/g (as reported in Table 8-3). The rate of change between the lower
and middle age groups was estimated by the regression slope as 0.83 pg/g per decade, with a
standard error of 0.17. The average concentration of TCDD in the lower age group was 2.06
pg/g (not shown), so the rate of 0.83 divided by 2.06 resulted in the standardized rate of 40%
per decade. Similarly, the rate of change between the middle and upper a*ge groups was
1.52 pg/g per decade, with a standard error of 0.20. The average concentration of
2,3,7,8-TCDD in the middle age group was 4.33 pg/g (not shown), so the rate of 1.52 divided
by 4.33 resulted in a standardized rate of 35% per decade.
The rate of change of 2,3,7,8-TCDD of 1.52 pg/g decade from the middle to the upper
age groups is similar to the value of 2 ppt/decade reported by Patterson et al (1985). Their
study investigated adipose tissue samples from individuals of age 35-85, whereas the present
study analyzed composite samples from individuals representing all age groups.
All of the rates of change shown in Table 8-5, except for the first rate for 2,3,7,8-TCDF,
were positive and highly significant (p < 0.0001). For seven of the nine analytes shown, the
rates of change per decade increased after age 31. However, when the rates were converted
to a percent of initial average concentration, eight of the nine analytes showed a decreased in
standardized rates per decade after age 31.
8.5 OUTLIER DETECTION
Prior to conducting the statistical analyses of the FY87 NHATS data, outlier detection
procedures were performed to identify possible data entry errors and errors associated with
the analytical method (Rogers, 1989). Outlier detection was performed on four types of data:
(1) measured concentrations of native analytes, (2) internal quantitation standard recoveries,
(3) LODs, and (4) percent lipid values for composite and QC samples.
102
-------
?uran
T3
d
re
d
"rt QJ
o 3
.is en
°P
1 £
O o
QJ X
»-H SH
QJ 1^
Ux wW
cn ^
^ ^
o C
QJ fy
«>£
c *-<
% *^
nj h+H
X I
i!-s
O en
en d
QJ O
•£ '43
"3 re
Pi h
4«J
T3 d
QJ aj
•£ cj
ro d
S o
•3 u
en
PU
in
06
^H
2>
re
H
2
re
sti
>-<
-L
r
rq
m
QJ
bo
>H
O
c
o
D
E
re
QJ
>H
rH
en
o
OJ
bo
<
>-H
O
C
o
0
g ^
S5 "^
^ £
re ^
c«« ^
^^
pa
aj S2.
-M
cd
Pi
. o
^t
< ^D
"5fc
OH
CJ
S <™
r-J CJ M)
1 s a
o C
•43 o
re U
z
QJ
£
"re
5
<
inoorH oorqoo is.oo
encqc^j rneSrH tsienen
omes sD'*^^ oooooo
rqents. cq-^o cjinrn
cJ ci r-! cj pq TJJ o o o
Q,
rq rq rn is. •* as
insocq osoo envort<
rH c4 Ifi rH rH O O H rH
rH CM en
V"H
OrqrH OrHTf f^rHOS
Tfinso e-q^ttx 'T'm-^
t^en^rH^ trTirr^-^ aT^r?"
rHtsirH rqinoo oenrn
OCSrH OtNO OOC5
en so o en esj en
OOOSrH Tj^C^ CSJOOC5
C5rH\O rHOseN OrHrH
rH rH OO '
rH
oo oo o oo
en is. rH is. oq ts is.
irioin rno-rt; rnosin
rH t^ rH rH CN
rH tS.
§ Q Q HH
o y R o, * R
08 £%& ^843
Dw^ooSoo QaJE
UCHO&tsvosis: UPHOO
enHoOts.vo"oOsO_ ^Hootx
.5 oo isT rj? en" tC •*" y c oo c\ so]
£&
re re
^^
H ^
ll
IH IH
0 O
•+H >*H
'os'u<
QJ OJ
2.S,
QJ QJ
T3 T3
ro re
U >
re re
M— 1 ^*— *
0 0
4-» +•>
d d
QJ QJ
CJ U
S-* l-l
QJ QJ
OH 0,
re re
tn en
re re
T3 -d
QJ QJ
on en
en en
QJ OJ
^ V^
OH OH
X X
H->
re re
pi Pi
ra ^
103
-------
Logic checks were performed to identify obvious inconsistencies in the data. For
example, logic checks would reveal records having recorded concentrations but a data qualifier
of "not detected." The extreme studentized deviate (ESD) test statistic was applied to the
residuals of a simple linear regression model fit to the measured concentrations and recoveries
as another means of detecting outliers. Finally, the secondary outlier procedures were
performed using tests for normality, multivariate techniques, and graphical techniques
(boxplots).
No data problems were detected in the logic checks phase. However, a total of 62 of
1620 data items were identified as potential outliers. Of these, total 45 were found to be
correct readings by the analytical laboratory. The other 17 data items were incorrect readings
which were recalculated. The laboratory also reported additional data changes resulting from
its own review of the outlier analysis. All data corrections were made to the master dataset
before proceeding with the statistical analysis.
8.6 MODEL VALIDATION
Three types of analyses were performed to evaluate the adequacy of the additive model
for use on the FY87 NHATS data. All three analyses were based on comparisons of the
observed (i.e., measured) and predicted concentrations for the composite samples. Predicted
concentrations were calculated using the statistical analysis approaches discussed in Section
7.2. Residuals, which were also used in the model validation analysis, were calculated by
taking the differences between the observed and predicted concentrations.
Model validation analyses included (1) residual analysis, (2) normal probability plots,
and (3) R-squared analysis. The use of the Shapiro-Wilk tests for normality was also
considered. However, in this application, the Shapiro-Wilk test is not appropriate because the
data are correlated and variances increase with increasing concentrations.
The residual plots confirmed the model assumption that the variance of the measured
concentrations will increase with the average concentration. The plots also show that the
distribution of residuals is symmetric. This supports the use of normal models for the
sampling and measurement errors. As discussed in Section 7.2.2 the normality assumption is
important for ensuring the validity of the hypothesis tests. Nearly all of the probability plots
were linear, thus supporting the normality assumption for the errors. Those that were not
linear could be explained by the larger variances at the high concentration levels.
104
-------
Finally, Table 8-6 lists the R-squared correlations between the observed and predicted
composite concentrations calculated for each analyte. R-squared can be interpreted as the
percent of the total variability in the data (observed concentrations) that can be explained by
the model. For example, 81% of the variation in measured composite concentrations of TCDD
can be explained by the fixed effects of the additive model. Overall, these correlations
demonstrate excellent agreement between the data and the model. Eight of the nine analytes
produced R-squared correlations of at least 68%. The lowest value of R-squared (46%) occurs
with 2,3,7,8-TCDF. This can be explained by the relatively small age effect on 2,3,7,8-TCDF
concentrations. Although there were statistically significant age effects for all nine analytes,
the ratio of average concentrations in the lower and upper age groups was only 1.2 (2.45/1.97)
for 2,3,7,8-TCDF. The ratios exceeded 2.5 for all the other analytes. Thus, for each of these
analytes, the additive model accounts for a large percentage of the total variability in the data.
Table 8-6. R-Squared Correlation3 Between Predicted and Observed
Concentrations for FY87 Dioxins and Furans
Chemical R2 (%)
2,3,7,8-TCDF 46
2,3,7,8-TCDD 81
2,3,4,7,8-PeCDF 68
1,2,3,7,8-PeCDD 85
1,2,3,6,7,8-HxCDF 87
1,2,3,4,7,8/1,2,3,6,7,8-HxCDD 88
1,2,3,7,8,9-HxCDD 79
1,2,3,4,6,7,8-HpCDD 88
OCDD 82
a R-squared is the square of the Pearson correlation coefficient. It
represents the percent of variability in the data that is explained
by the predictive model.
105
-------
9.0 COMPARISON OF FY87 DATA WITH FY82 AND VA/EPA DATA BASES
The analysis of the FY87 NHATS specimens as composites provides a reference point
for body burden levels of PCDDs and PCDFs in the general U.S. population. The data
generated from the analysis of the FY87 NHATS specimens can be compared with other data
bases that have been developed from the analyses of samples collected in North America and
Europe. The documentation on the total effort (the compositing design, chemical analysis and
statistical treatment of data) offers a means of comparing the significance of the FY87 NHATS
data set with two other analysis programs conducted using the NHATS specimen repository.
These two studies are the FY82 NHATS broad scan analysis effort and a collaborative study
conducted between the U.S. Department of Veterans Affairs (VA) and EPA's Office of Toxic
Substances (VA/EPA). The comparisons of these studies extends the utility of the data bases
in establishing trends in body burdens of PCDDs and PCDFs and identifies limitations in
comparing the results to other data sets. This section gives an overview of these programs
and presents comparisons of the compositing designs (FY82 and FY87 NHATS), analytical
procedures, results, and statistical methodologies.
The objectives of the FY82 and FY87 NHATS were quite different from those of the
VA/EPA study. The FY82 and FY87 NHATS studies were conducted to develop baseline
estimates of tissue concentrations. The FY82 and FY87 NHATS tissue specimens were
obtained from cadavers and surgical patients. The target population was all
noninstitutionalized U.S. citizens in the conterminous 48 states. On the other hand, the
primary objective of the VA/EPA study was to compare PCDD and PCDF levels in Vietnam
veterans with those found in similar groups of non-Vietnam veterans and civilians. The
VA/EPA study was a retrospective study based on archived NHATS specimens. Only
specimens from male donors born between 1936 and 1954 were included in the VA/EPA study.
Despite the differences in study objectives, it is possible to compare the average
concentrations found in the VA/EPA study and those found in the 15-44 year age group from
the two NHATS surveys. The donors in the VA/EPA study were all between 17 and 46 years
old at the time of their death or surgery. In Section 8.0 it was concluded that the only factor
consistently affecting concentrations of dioxins and furans was the age of the donor.
Additional information on the programs are presented in Section 9.1. Comparisons of the
study designs, chemical analysis procedures, and the significant results are presented in
Sections 9.2, 9.3, and 9.4, respectively.
107
-------
9.1 OVERVIEW OF THE ANALYTICAL PROGRAMS
The FY82 NHATS specimens were analyzed as composites as part of a broad scan
analysis program conducted to expand the utility of the NHATS program beyond the
monitoring of organochlorine pesticides and PCBs. Forty-six composite samples were
analyzed for tetra- through octachloro PCDDs and PCDFs. The FY82 effort was designed to
provide body burden estimates for the general U.S. population based on age, sex, and
geographic region.
The VA/EPA study used approximately 200 individual specimens collected from 1971
through 1982. The specimens were from adult males with birthdates between 1936 to 1954
who potentially might have served in the Vietnam War and who possibly had been exposed
to the herbicide Agent Orange. These specimens were categorized into three groups:
Vietnam veterans, veterans with no military records indicating service in Vietnam, and
civilians. The design of the study was intended to determine whether there was any possible
difference in the levels of 2,3,7,8-TCDD between groups. The analysis program, however,
provided data on all of the 2,3,7,8-substituted chlorinated dioxin and furan congeners. Hence,
this data base has generated a considerable amount of information that can be compared with
other data bases.
9.2 COMPARISON OF STUDY DESIGNS
Similar sampling designs were used for collecting tissue specimens in the FY82 and
FY87 NHATS. Both studies used a multi-staged sampling plan. The conterminous 48 states
were divided into strata; MSAs were selected with probabilities proportional to size; and
cooperators were solicited and assigned quotas for collecting specimens. There was a minor
difference only in the method of stratification. Prior to the FY85 NHATS, MSAs were selected
from strata denned by U.S. Census divisions. Beginning with the FY85 NHATS, sampling
strata were redefined to be the 17 geographic areas that resulted from the intersection of the
nine Census divisions and the ten EPA Regions (Panebianco DL, 1986a). A controlled
selection technique, known as the Keyfitz technique (Mack et al, 1984), was used to maximize
the probability of retaining MSAs used in previous years. As a result, there were 47 MSAs in
the FY87 design compared to 35 in FY82. Otherwise, the sampling designs for the FY82 and
FY87 were essentially the same.
A total of 763 specimens were used to generate the composites for the FY82 study, and
865 specimens were included in the composites for the FY87 study. As shown in Table 9-1,
the distributions of specimens among the various geographic and demographic subpopulations
were similar.
108
-------
Table 9-1. Marginal Comparisons of FY82 and FY87 NHATS Individual Specimens Used
for PCDD and PCDF Analysis
Category
No. of specimens (%)
FY82
FY87
1980 Census
population (%)
Census Region
Northeast 166(22)
North Central 206(27)
South 331(43)
West _60(8)
Total 763
Age Group
0-14 years 178(23)
15-44 years 312(41)
45+ years 273(36)
Total 763
Sex
Male 412(54)
Female 351(46)
Total 763
Race Group
Caucasian 632(83)
Non-Caucasian 131(17)
Total 763
175(20)
296(34)
289(33)
105(12)
865
146(17)
318(37)
401(46)
865
436(50)
429(50)
865
707(82)
158(18)
865
26
22
33
19
23
46
31
49
51
83
17
109
-------
The FY82 and FY87 NHATS had comparable compositing designs. One of the design
criteria for compositing FY87 specimens was to maintain similarity to the FY82 design.
Table 9-2 gives a comparison of the marginal percentages of composites in each of the
categories defined by the four analysis factors. Population percentages from the 1980 Census
are also provided.
Overall, the marginal percentages from the FY82 and FY87 NHATS agree reasonably
well with each other and with the census figures. The only differences are the FY87 NHATS
had more "pure sex" composites (31 versus 11) than FY82, while the FY82 NHATS had more
"pure race" composites (17 versus 8) than FY87.
Because the VA/EPA investigation was a retrospective study using surplus specimens
from the NHATS archives, the method used to select specimens was different from the
NHATS sampling strategy. Of the approximately 8,000 unused NHATS specimens collected
between 1971 and 1982, 528 were collected from males born between 1936 and 1954. However,
there was sufficient background information on only 494 donors. Specimens from 40 donors
who served in Vietnam were selected, along with randomly selected specimens from 80 non-
Vietnam military veterans. Finally, specimens from 80 civilian men were included in the study
by matching the birth year (±2 years) and sample collection year (±2 years) of two civilians
with each Vietnam veteran. Thus, a total of about 200 specimens from male donors between
the ages of 17 and 46 formed the basis for the VA/EPA study. Of the specimens identified,
successful analysis was achieved for 197 individuals.
Another major difference between the two NHATS and the VA/EPA study is that
specimens in the NHATS studies were composited prior to chemical analysis, while specimens
selected for the VA/EPA study were analyzed individually. This difference affects the way in
which the data are statistically analyzed, but, as discussed in Section 9.4.2, it does not affect
the comparison of average concentration levels.
9.3 COMPARISON OF ANALYTICAL PROCEDURES
To compare the data for the three studies, it is necessary to review the analytical
procedures (see Figure 9-1). While the analytical procedures used for the VA/EPA and FY87
efforts were fairly comparable, the figure illustrates that the FY82 approach was considerably
different. The changes in the analytical procedures were incorporated in VA/EPA and FY87
studies as an effort to improve the state-of-the-art of the analytical technology between the
time frames that each study was conducted.
110
-------
Table 9-2. Marginal Comparisons of FY82 AND FY87 NHATS Composite Designs
No. of Composites (%Y
Category
Census Region
North Central
Northeast
South
West
Total
Age Group
0-14 years
15-44 years
45+ years
Total
Sex
Pure male
Mixed
Pure female
Total
Race Group
Pure Caucasian
Mixed
Pure Non-Caucasian
Total
FY82
12 (26)
9(20)
19 (41)
-L(13)
46
12 (26)
17(37)
17_(37)
46
6(55)
35
_5_(45)
46
11 (65)
29
6(35)
46
FY87
15 (31)
9(19)
16 (33)
_8_(17)
48
11 (23)
17 (35)
20 (42)
48
16 (52)
17
15 (48)
48
8 (100)
40
-Q-( 0)
48
1980 Census
population %
26
22
33
19
23
46
31
49
51
83
17
The percent estimates for sex and race groups are calculated as the total number of pure
composites within each study design. For example, 6 of the 11 (55%) pure sex composites
in the FY82 study design were composed of males only.
Ill
-------
9 -13C IQS
1 0g Composites
CO
o
2 s
•* —
Oco
CO J
CM
X
Neutral
Alumina
CO
_o
2-33
Stf
X
<
CO
is
X 0
— 62
| =
UL t LL ; LJ_
r? Q Q Q Q
g QQQ g
gooog
"
>
LJJ
«5
QJ
5
U3
IN
oo
<
Q_
III
CO -2
2 If
w° <55
"~" O)
C31 0
CO
o
2 g
0*"=
CO00
CM
X
CO
s?
6
1SB
Q. —
O CU
-90
CO
O
CO
=i: o
O °°
CD OC
DC
X
LL u. u. LL^
r? Q Q Q Q-
Q ,9 R ,9 D
ea
2?
cs
oo
13
o
T3
03
Q U. Q '"
QQQ Q.
OOO £
HHO 0
666 "
co co ro ^
'"'""" CM
O
Q.
CD
'to
o
u_
— »»
PX-21/Glass
Fiber
6
iSfl
8" 03
•SO
CO
O
CO
DC O
Is
o °°
O oc
DC
n:
CO
2
cc o
T O
=t o
O co
CD CC
cc
Q Q ?5
Q Q g
OOo
O *-*
HO?
13
CO
O
ta
I
o
U
112
-------
Each procedure required fortification with internal quantisation standards (IQS),
extraction, removal of bulk lipid, and separation of interferences from the PCDDs and PCDFs.
The techniques for all three studies were essentially equivalent. Extraction was achieved with
methylene chloride using a Tekmar Tissuemizer to promote thorough extraction of lipids.
Bulk lipid removal for the FY87 and the VA/EPA studies was conducted using identical
techniques, consisting of treatment with sulfuric acid-modified silica gel slurries and further
cleanup via a chromatographic column of the same material. Gel permeation chromatography
was used to remove the bulk lipids in the FY82 composites.
The separation of chemical interferences was achieved using Florisil (FY82 NHATS),
acidic alumina (VA/EPA), or neutral alumina (FY87 NHATS). Neutral alumina was used for
the FY87 samples rather than acidic alumina to improve method recovery and reduce possible
background contributions due to hepta- and octachloro-PCDDs. Previous efforts using Florisil
on the FY82 composites had demonstrated poor recovery of the hexa- through
octachloro-congeners (USEPA 1986a).
For the final cleanup of sample extracts, a carbon-based column was used. However,
as noted in Figure 9-1, three different carbon adsorbents were used between the studies. Two
separate extracts were cleaned for the FY82 NHATS composites. Because recovery of the
higher chlorinated compounds was poor, an aliquot of the extract taken through the GPC
cleanup, but not through Florisil chromatography, was taken through a PX-21/glass fiber
column to determine the hexa- through octachloro-PCDDs and PCDFs.
The AX-21/silica gel column used for the FY87 NHATS composites did not provide the
degree of cleanup demonstrated with the Carbopak C/Celite used with VA/EPA. This was
primarily noted through the detection of octachlorodiphenylethers that interfered with the
determination of 1,2,3,4,7,8-HxCDF and 2,3,4,6,7,8-HxCDF in the FY87 composites. The
HRGC/HRMS conditions varied across studies (Figure 9-1), depending on whether analyses
were conducted using a mass resolution of R = 3,000 or R = 10,000. The higher the R value,
the more specific the analyses and the higher the confidence in compound identification. The
pattern or fingerprint of the major PCDDs and PCDFs observed in the HRGC/HRMS
chromatograms for human adipose tissue was consistent across all these studies.
Two factors have the largest potential effect on data comparability among the three
studies: the type and number of IQS and the consistent use of analytical standards
(Figure 9-2). Only three IQS compounds were available for the FY82 composites. Since the
calculation for PCDDs and PCDFs is based on an isotope dilution principle, the limitation on
the FY82 composites is the assumption that all compounds will recover the same as the IQS.
Recovery data for the additional IQS compounds in the FY87 and VA/EPA studies demonstrate
113
-------
FY82
NHATS
EPA/VA
FY87
NHATS
2,3,7,8-TCDD
2,3,7,8-TCDF
1,2,3,7,8-PeCDD
1,2,3,7,8-PeCDF
2,3,4,7,8-PeCDF
1,2,3,4,7,8-HxCDD
1,2,3,6,7,8-HxCDD
1,2,3,7,8,9-HxCDD
1,2,3,4,7,8-HxCDF
1,2,3,6,7,8-HxCDF
1,2,3,7,8,9-HxCDF
2,3,4,6,7,8-HxCDF
1,2,3,4,6,7,8-HpCDD
1,2,3,4,6,7,8-HpCDF
1,2,3,4,7,8,9-HpCDF
OCDD
OCDF
Internal Quantitation Standards
13C12-2,3,7,8-TCDD
13C12-2,3,7,8-TCDF
13C12-1,2,3,7,8-PeCDD
13C12-1,2,3,7,8-PeCDF
^C^-I^.S.SJ.S-HxCDD
13C12-1,2,3,4,7,8-HxCDF
13C12-1,2,3,4,6,7,8-HpCDD
13C12-1,2,3,4,6,7,8-HpCDF
13C12- OCDD
Internal Recovery Standard
13C12-1,2,3,4-TCDD
13C12-1,^2,3,7,^8,9-HxCDD
Quantitative
Qualitative
Figure 9-2. Comparison of analytical standards from the FY82, VA/EPA, and FY87 studies.
114
-------
that PCDDs and PCDFs recoveries differ depending on the degree of chlorination. For these
reasons, only the data for 2,3,7,8-TCDD, 2,3,7,8-TCDF, and OCDD are directly comparable
between the other two studies. Since the same sets of IQS standards were used between the
studies, the FY87 and VA/EPA studies are comparable.
The second factor, standard traceability, is another serious consideration. Figure 9-2
shows that all standards are directly comparable between the FY87 and the VA/EPA studies.
When considering the standards for the FY82 composites, only the 2,3,7,8-TCDD standard is
directly comparable across all three studies. The standards analyzed with the VA/EPA and
FY87 studies (including the 2,3,7,8-TCDD for FY82) were verified through participation in
interlaboratory comparisons and the analysis of an NBS standard reference material. The
results of these interlaboratory studies support the quantitation of results reported in these
studies and also promote the comparability between human tissue data sets generated by the
other laboratories participating in these studies.
9.4 COMPARISON OF RESULTS
The results from the FY87 NHATS, the FY82 NHATS, and the VA/EPA studies are
compared in this section. Because the .same study design was used for the two NHATS
surveys, it is possible to make a more detailed comparison of those two sets of results. A
statistical comparison of the FY82 and FY87 results is presented in Section 9.4.1. The VA/EPA
results are compared with the FY82 and FY87 NHATS in Section 9.4.2.
9.4.1 Statistical Comparison of FY82 and FY87 NHATS Results
The results from the FY82 and FY87 NHATS were statistically compared to determine if
there were significant changes in average PCDD and PCDF concentration levels over the five-
year period. However, as discussed in Section 9.3, advancements were made in the analytical
method for the FY87 survey. The most significant change was that additional internal
quantitative standards (IQSs) were available for the penta-, hexa-, and heptachloro-
compounds. Therefore, only results for the tetra- and octa- compounds are expected to be
directly comparable. Statistical comparisons were performed on all compounds for which data
were generated in both years. Comparisons were made, not only between the predicted
national averages, but also between the demographic profiles from the two surveys. The
profile analysis, discussed in detail later (9.4.1.3), examines possible changes in the differences
across demographic subpopulations. This type of comparison is valuable even if systematic
differences in concentration levels exist that can be attributed to the analytical methodologies.
115
-------
Statistical comparisons were possible only when sufficient data were available from
both surveys. The criteria for performing a model-based comparison were: (1) the chemical
must be detected in at least 50% of the composite samples each year/ and (2) there must be at
least 30 analyzed composite samples in each year. The additive model, described in
Section 7.0, was used for the model-based comparisons between the FY82 and FY87 NHATS.
Since the FY82 data were originally analyzed using a multiplicative model (EPA 1989), the
FY82 results presented in this comparison are different from those previously published
(USEPA 1990a).
Table 9-3 shows the type of comparison used for each of the target chemicals. Six
analytes met the criteria for a model-based (M) comparison. The other four analytes, analyzed
in both years, were compared in a descriptive manner using weighted averages (WA). This
approach is discussed later. (See 9.4.1.4.)
As discussed in Section 8.0, data restrictions derived from the data quality objectives
were imposed on the FY87 data but were not applied in FY82. Thus, in Table 9-3, the number
of composites listed for FY87 represents the number with unrestricted measurements, while
the number for FY82 is the total number of available composites.
The comparison of FY82 and FY87 results is divided into four parts. Section 9.4.1.1
compares the FY82 and FY87 results in terms of limits of detection (LODs) and the percent of
composite samples for which each analyte was detected. Estimates of the national averages
for the six analytes statistically analyzed using the additive model are compared in
Section 9.4.1.2. The results of the profile analyses are presented in Section 9.4.1.3. Finally, in
Section 9.4.1.4, a descriptive comparison of weighted national averages is presented for the
four analytes that were not statistically modeled.
9.4.1.1 Comparison of LODs and Prevalence Detected. Table 9-4 compares the
percent of composite samples in which the analytes were detected and the average detection
limit (LOD) for each year. In FY82, LODs were only calculated when the concentration was
either not detected (ND) or qualified as a trace (TR) value. Thus, the sample size for
calculating the average LOD in FY82 was often much less than the number of samples
analyzed. For example, TCDD was either not detected or found at a trace level in 13 of the 43
composite samples for the FY82 study. The average LOD reported for the 13 samples was
7.28 pg/g. In FY87, LODs were calculated for all composite samples.
116
-------
Table 9-3. Number of Composite Samples and Types of Statistical
Comparisons Made Between FY82 and FY87 NHATS Results
Number of composite
samples
Analyte
2,3,7,8-TCDF
2,3,7,8-TCDD
1,2,3,7,8-PeCDF
2,3,4,7,8-PeCDF
1,2,3,7,8-PeCDD
HxCDFd
HxCDD"
1,2,3,4,6,7,8-HpCDF
1,2,3,4,7,8,9-HpCDF
1,2,3,4,6,7,8-HpCDD
OCDF
OCDD
FY82a
43
43
-
43
41
45
45
45
-
45
45
45
FY87"
33
36
43
39
35
9-45
39-41
27
46
42
23
32
Type of
comparison0
WA
M
-
M
M
WA
M
WA
-
M
WA
M
a Number of available measurements. Two outliers were removed for 1,2,3,7,8-PeCDD.
b Number of unrestricted composite sample, measurements.
c WA = weighted averages
M = model results and profile analysis
- = no data available for FY82
d Analyte analysis results for specific isomers of HxCDF and HxCDD were combined
(summed) for comparisons between FY82 and FY87.
117
-------
^
00
-d
C
ro
rq
00
fe
H
^1
H"^
J-j
Z
c
01
o>
_g
QJ
CQ
p£
P
U
9
nj
yj
P
P
U
P*
JH
aj
4-1
0)
4-*
QJ
P
4-*
QJ
IH
QJ
PH
I ,
C,
ro
en
Q
O
*— 1
"•4-H
O
eft
r-
*H
O
en
•c
ro
pi j
c2
0
U
•<*
ON
QJ
3
.S
QJ 4-1
bO £
6 S
S VH"
U o
tx
oo
PH
CNl
00
\^
PH
T3
cu > — «v
0 Q
H-d £S
•
-^
" ol
bO >ii?
ro
1-1 P
Is
<. 1— 1
4- -d
^ 4*
C 4J
QJ CJ
U QJ
IH 42
QJ QJ
PH "d
z
,— ,
x?~
aj p^
ro ^~"
*""* P
< J
Z
4- "d
QJ tj
O QJ
QJ QJ
PH "d
a
Z
_QJ
>~i
"ro
C
•*c
* *
^ CO . rx.
K C<1 ' l>1
m T-H •<* in
O ON CO 00
CS CN! ^1 CO
0000
R tx ^ in
2 ON r-i ON
T-H
CO NO CO ON
CO CO Tj< CO
^^ OO
ON r
PH
OO
tx"
CO
^i.
1-H
oo
NO
ON
NO
O
K
^
o
tx
i— i
ON
r-l
T-H
tx
in
PH
P
U
><
PH
«s
bO
ON
i— i
T— I
g
1-1
tx
CO
ON
<*
OO
ON
in
P
U
X
J-i
^ , oq 0 o
i
o T-H tx, tx in
O •<* O NO tx
T— 1 I— 1 *— 1 1— 1 CO
ON ^ o o o
oo ^ O ro O
1-H T— 1
tx NO CNJ co cs
CN) ^ "^ CNl CO
rH
0 00
CT^ ' \O CT^ '
C^ ^H
IX r-H ^ 0
i fn
m ' oo o o
QN Q\ <^JH ^^
in m m m
[*T j fTj J]
P P P
U U U
OH OH OH
oq^ ON^ oq^
tx OO tx
rjn" TJ? T^H" PH U
co" co" co" P P
CN? CNf C
S
in
en 0
P <=>
O 15
1-3 o
*T3 *•"•
-S . N
TO tft _,
r-d QJ W
•d 1 ••§ ,3
N^ at
^x a 2
2 ±i o «
d > o QJ
ro ^ H
CO CO QJ
4-J H^ "u -52
'co en "C QJ
O O *3 o
i-ris
83 §|
"o o ^o "^
S3 S S3 J
rg rQ ,0 ro
3 3 3 "3
Z Z Z £
rt ,Q u T3
c
0
ro
OH
a
o
o
g
<4H
*^->
*^3
a
a
d
xS
•d
c
1
o
QJ
S3
£
Q
P
(J
*T3
d
8
^s
' i '
M-H
o
QJ
O
eft
U
*r^
S
>S
e/5
•^ tx
^ ?2
S p^
VH
'^ «
IS
1^
1 «
tu ^
U rO
0.
.
*d
4J
CJ
0)
4-1
o>
•d
QJ
JJ
^
2
M
0)
a
en
-rH
P'"^
P
U
^^
£
1
lH
O
^— ^
PH
P
x
E.
Ui
^
^o
QJ
MH
O
K^
J—
ro
vy
CJ
"S
5
"is
C
<
P
O
j
QJ
bO
ro
IH
>
ro
o>
•£
QJ
"ro
3
o
13
CJ
-d
4J
en
^"1
CO
•»-H
en
>H
OJ
.S
r^
i ^ j
M
1
eft
^""^
O
ftj
^-»
"o
1
f
H
bO
118
-------
There clearly was a significant improvement in the sensitivity of the analytical method
in FY87. The average detection limits in FY82 were in the range of 9 to 50 times higher than
those in FY87. This, most likely, explains the statistically significant increase in the percent of
samples in which TCDF and TCDD were detected between FY82 and FY87. For example,
2,3,7,8-TCDD was detected in only 74% of the samples in FY82 but was detected in 97% of the
samples in FY87. This difference (23%) is statistically significant; however, the average
detection limit was decreased from 7.28 pg/g to 0.291 pg/g. The average 2,3,7,8-TCDD
concentration in both years was estimated to be around 5.5 pg/g.
9.4.1.2 Comparison of National Average Estimates for Modeled Analytes. Table 9-5
shows the estimated national average concentrations for FY82 and FY87 and the estimated
difference (FY87-FY82) in concentrations for the six analytes statistically analyzed. The
standard error of each estimate and the significance level for testing that the difference is
different from zero also are provided. The test was based on the approximate t-statistic of the
form
NA87 NA82
t =
{•
where NA82 and NA87 are the FY82 and FY87 national average estimates and SE82 and SE87
their standard errors, respectively. Approximate significance levels were calculated using the
standard normal distribution.
Generally, levels less than 0.05 are used to indicate statistical significance. Using this
criterion, the average predicted concentrations of 2,3,4,7,8-PeCDF, 1,2,3,7,8-PeCDD, and
HxCDD are significantly lower in FY87 than in FY82. However, as mentioned earlier,
additional internal quantisation standards were used in the FY87 analytical procedures. Thus,
we cannot conclude that the average concentration in the U.S. population changed between
FY82 and FY87. The differences may be due only to the changes in the analytical method.
Data from the VA/EPA study also demonstrated significantly different concentrations for
2,3,4,7,8-PeCDF and 1,2,3,7,8-PeCDD comparable with the FY87 levels. No significant
differences in the average or predicted levels of 2,3,7,8-TCDD and OCDD were noted between
FY82 and FY87. The same internal quantitation standards were used in both years for these
specific analytes.
119
-------
M-l
O
cn
d
£
H
•C
C
cu
CJ
C
0
U
cu
bo
ro
QJ
>
< a
"3 1
C -22
0 H
'•5 ^
•7 0
2 S,
T3 :=3
(U ^3
•& <*
CJ ^*
'•£ £
xj *3
QJ TO
LH Ci
£ |
IS. l-TH
00 I
>> c
PL, .n
13 rSP
r; P-i
so
rq U
oo Pi
£"§
" S
£ 2
< Q
£a
zy
*-, CH
0
VI
C
o
w
•c
(H
ro
a,
g
o
U
in
ON
O)
i— *
-TO
H
13
o
•43
CO
"J3
ro
4-t
en
OJ
CJ
c
a
Ol
iK
P
cu
OJD
ro
tH
o>
>
ro
"«
f-
#H
O
'^•U
c 1
tJ °
o>
^— *
ro
s
'^J
CO
td
cu
CJ
q
S
•^
.«•
'S3
CQ
00
>H
UH
tx
oo
C.
^
00
oo
>H
H-I
cN
oo
>
fr .
M-(
CU
>
^Oj
)
^)
uT
x2-
iri
^A
P
(0
w
en
bb
>
<
a
s~
en,
bb
^
2
j>,
"ro
^
en I-H r-i m ^ tN
T-I o o ^ c5 o
f i-i m
o o cs o o o
V
X— V
cs NO" ^^ ^^ , — * f-^
NO NO O CO C3 IS,
o en ci C3 in T*
"~" ^ £S CS Ci xS
ON
ON
rt ts ON OO vO ON
cJ in tN ^ T-H tri
' cN SO CO CO ^t
1 1 1 1 1
o" o^ NO" , — , s-^.
m o cs NO ro s-*.
ro 00 -^ i-< Cs NO
o, ci o^ ro ro^ oo'
CN^
oo o
ro Is. ts oq
in ON O NO O r*
I-H 00 T-I (N
l-H IS.
^~ •-— •»
ff\ t**^- J— ^
C* J L^s X^^s x™ •'v X^V <^^X
in in o »-i oo ts,
O ro' O O •<* ON
s— ' — < cs" U
CS" CS" rl H- r-n" O
o
»
'C
ro
D,
a
o
CJ
t-l
&
^
cu
c
a
(0
*^^*
-a
cu
1
a
8
cu
O» JH
So aj
ro S
£ ^
S P
ro . P
•^ u y
- H^
ro CU "-"H
§^^
•J3 HH
« 3 2
^5 S Cs'
"o o a oo
, , O >H
t»o bc.23 PL,
~£b fio o 13
r*i d. ' T; r*
J_L, M-i'-H tn
x-xs^*d «
gg^g
cu cu »H p.,
•^ "2 ^ c
ro ro j2 3}
T3 T3 ^ |
| I S 1
en en pi ^
as A u
120
-------
9.4.1.3 Results of Profile Analysis. Profile analysis is a multivariate statistical
technique that is used to compare profiles of two or more populations. In this context, a
profile is a vector of estimates of subpopulation averages. An example of a profile is the set of
estimated average TCDD concentrations for the four geographic regions (NC, NE, S, and W)
in the FY87 NHATS.
Profile plots were used to make a visual assessment of the differences in the FY82 and
FY87 results. In order to determine what differences may exist and whether these difference
are statistically significant, a formal profile analysis was conducted.
Profile analysis tests a sequence of statistical hypotheses to determine how two or
more multivariate populations may differ. The hypotheses address the following questions in
order:
1. Are the profiles from the two fiscal years parallel?
2. Assuming the profiles are parallel, are they coincident?
3. Assuming the profiles are coincident, do they have equal levels?
Two profiles are said to be parallel if all pairwise differences between the average
concentrations across the levels of a demographic factor are equal. For coincident profiles, all
pairwise differences are equal to zero. Finally, coincident profiles are said to have equal levels
if there are no differences among the average concentrations at different level of a
demographic factor. The different types of profiles are illustrated in Figure 9-3.
Profile tests are performed sequentially. The hypothesis of coincident profiles is only
tested if the hypothesis of parallelism is not rejected. Similarly, the hypothesis of equal levels
is only tested if the hypotheses of parallelism and coincidence are not rejected. The third test
combines the FY82 and FY87 data to test for significant effects of the demographic factors.
This approach described by Johnson and Wichern (1982) was used to conduct the profile
analysis.
To provide the background for the profile analysis, Table 9-6 compares the FY82 and
FY87 significance levels from testing for differences among demographic groups. The additive
model was used to perform the tests, with the assumption of normally distributed errors for
both years. Although these assumptions were found to be reasonable for the FY87 data, the
large measurement errors from the chemical analysis of FY82 composites indicate that the
assumption of normality may not be true for the FY82 data. However, because of these data
in FY82, it is difficult to verify any distributional assumptions.
121
-------
Average
Concentration
Parallel Profiles
NC NE S
Census Region
W
Legend:
D=FY82
A. FY87
Average
Concentration
Coincident Profiles
NC NE s
Census Region
W
Legend:
D=FY82
A=FY87
Average
Concentration
Equal Profiles
— i — i i i
Legend:
D=FY82
A=FY87
NC
W
NE S
Census Region
Figure 9-3. Example of profile plots.
122
-------
co 42
'55 v
CD ^
0 ^
&p
^! C_J
c ^
p^
S £
43 to
350
|8
CO "O
u CD
£
CO CO
O ^H
"§ tn
In **"
>-i 5
ft lH
•a O
C o
to -g
00 ^
S 8
S3 i
i P
2 £
"o >
to -P
ci cu
o oa
CO £-
"C co
CO U
ft, C
£ **
O Ctl
u ig
SO G
O 4-»
H H
C
C
1
oo
*
oo
uu
£
0,
CO
tj
QJ
>*7
^
p>
C
X
CO
cyj
IN
oo
f^
oo
o
3
o
ol
CO
w
oo
ft
tN
00
C
BO
s
CD
^
^3
c
CO
u
IN
OO
OO
ft
£
"re
^
I-H m m in in en o
^* CO Os Os Csl Os CN
en so in en Os en in
o d d J d o d o
rH IN in IN in en
so in oo so en in
00 SO Os OS t^» Os
d d d d ! ! do
tN en IN •«* IN ^ in
en o rq en os csi os
rq so so os in rj< ^
d d d ! d d d d
c P
0 p p ;|X r
Q u u o ir ~. ^
P O) O) H 0& OS 00
U PH ft "— ' tC OO" tC
H oo e» P rti' IN so" _
°°v *\ ^ P m" en^ •*£ P
ts* TJ^ en ^} f>i cq en '~~*
Csl CSI rH MH rH s^
.a
^
i— -j
c
to
f—H
Ol
o
CO
>
^i9
T3
CO
1
OJD
S -H
'5 ^
3 ^
a! in
-q °'
13 £
CJ ^
Oi fO
I.
^H fll
^ 1
^1
O o
t, g
~ s
Co
Co ^
cj co
C 'C
CO CO
c/5 H
B ^
123
-------
In FY82, measured concentrations for four of the six analytes were found to be
significantly different among the four census regions. In FY87, 2,3,4,7,8-PeCDF was the only
analyte with significant regional differences. On the other hand, in FY87, age effects were
much more evident. In FY82, only 2,3,7,8-TCDD and 1,2,3,7,8-PeCDD showed significant age
effects at the 0.05 significance level, and 2,3,4,7,8-PeCDF and OCDD were significant at the
0.054 level. None of the analytes in either year showed significant differences in measured
concentrations among the different sex or race groups.
Profiles of the six modeled analytes in FY82 and FY87 are plotted for each of the four
analysis factors (region, age, race, and sex) in Figures 9-4 through 9-9. The estimated average
concentrations within each analysis factor and fiscal year are connected with straight lines.
The vertical lines define approximate 95% confidence limits of the estimates. The confidence
limits were calculated by adding ±2 times the standard error of the estimated average. The
estimated national averages with 95% confidence limits are also plotted.
The numerical results—estimated average concentrations and their standard errors—of the
profile analyses are presented in Tables 9-7 through 9-10. The estimated difference (FY87-
FY82) and their standard error for each region are also presented in each table. Finally, the
significance levels of sequential tests for parallel profiles, coincident profiles, and equal levels
are provided.
Profile analyses were performed for five of the six modeled analytes. Total HxCDD was
not included, because in FY82 only the total HxCDD was measured, while in FY87 each
individual isomer was measured. To perform the profile analysis on HxCDD, it would have
been necessary to combine the individual isomer data for FY87. This would have required
making certain assumptions that would produce extremely conservative results.
124
-------
CD 00
Cb b*
•O II
c
(D
^^ O
0)
J
U
H
•
oo
X
0)
CO
o
CO
cc
in
in
g
- co "§>
CC
en
-Si
0)
O
_ O
c
o
CO
8
B5
C
O
•J3
.
C "*
P-
« 0,
D |
Q S
y „
^T 3
00 g
J
g
a c
00
2 ^
CO
CO
(6/6d) uoi
125
-------
00 CO
>• >»
DI
0}
G ^6 D
X
G g E3
•€> _
|
f-5
Q
U
00
•s
C3-
-^
D-
X
CD
CO
CD
s
rr
IT)
irt 03
- CO
c
o
"S1
cc
to
13
(O
c
(U
o
_ o
z
c
g
OB
th 95% confidence
concentrations
es
O.
S s
U
OJ
Cu
tissue
oo
r>T ^
^ U3
rj< O
m .S"
csf "o
re
QJ f^
bCoo
sfc
s^
•*- Q
O rq
oo
J2 >
_o u,
^s
oj —
!72 VI
11
Figure 9-5.
o
00
o
fO
o
*t
o
CM
(6/6d)
126
-------
(N r-
00 CO
•O II
C
0)
0^ Q
V
a
Q
u
00
•N
o
o
CO
o
o
CXJ
D-
-df
0 =»&•
- 0.
-5
- in
-5
o
o
X
CD
CO
8
CD
O)
g
co §?
DC
co
"J co
z C
CD
O
O
o
"to
01
C
o>
1
en
O
•.a
03
nj
o a,
0 £
O g
Q a)
U 3
bJDoo
C
CO
o>
"5 U-i
0)
2
a.
(L>
I
E
(6/6d)
127
-------
TJ
C
0)
fN
00
fa-
ll
Q
Q
U
+
— If)
X
CD
CO
o
nj
EC
IT) CD
*" O>
o
'o
0}
tr
CO
CO
C
CD
o
_ o
^
C
o
"
S-l
0
1
C
8
in
^^
••^
C
a
is
c
o>
u «
if
D
> -o
W (5
en
o
o
o
o
CM
o
o
(6/6d)
128
-------
OO
r-
CO
T3
C
V
Ql
0)
J
•x.
Q)
CO
o „ <*
8
CO
DC
a
Q
u
ft
a
00
w
in
«> CD
-co
O
cc
(0
3
CO
C
CD
0
_ o
.0
'cS
2
0
o
0
0
CO
0
0
CM
(6/6d) uoi
129
-------
CM P^
CO CO
>« >"
fc. Cb
•D II
C
0)
O>D
0)
II
— LL
U
O
— in
X
03
CO
in
•a
0}
§
EC
en
C
O
•J3
re
— C
8
8
O
R
U
a>
0
o
CO
0
o
CM
0
o
CO
0
o
(6/6d) uoi
130
-------
w
tu
Q
U
C
CO
g
U
C-,
U
V5
U-l
O
C
_0
S3
i c
c o
era
^ fti
U »
•o 5
1^
£»
tx
>-• 3
£p
M §
OO Q,
"•" 5
||
S.I
6 £
u c
'•— ' .£
"53
15
c
•<^
—
sG
1
"o
s.
s
(U
tt2
f<
o-
_aj
t2
1
13
1
o>
w
J
1
"c
ra
1
"oc
"Sc
a
"aT
^
u
Oj
Jj2
5
'S
£
/erage
ft
T3
ra
_|
1
"3
1
c
*FJ
j
o
Is
1
£
rq
w-
^
^H
PH
tx
UH
00
u.
n
"s
0.
1
Ci.
(S~
tJ2
S
sr
52-
^
w
52-
db
•5
c
'5b
a
t/5
3
g
01
U
I
<.
O i | O O
* *
S S 0 M S
O O O O O
V
o o o o o
SSli" »S2o rHtxvDM OOVCO^
r-.rHO" in<^mO KvgvOg OSCNIQ^ ^SSP3
v-i <-i rH (^
£ S^S3!
III T-1 II 1
^?*cint^ mtxrt^— < tNOOoooj c^c^om co o *c *— i
^ M w ,_ ^
tri vd iri •* omo;^ o^oo ^gSo gggo
^^^
o^ f— i \o . o CN fs
co-^^co Lnc^mm oo\o\oo r-imiriin
i— i CN *— • oo QO r^ vo
"n "3 13 "ra ~5
t-t U Ui t-i Ui
^(ft £1*^ "tfl —yi «trt
uw Uc3 cjw ut3 uw
XJ^j;^ JSX^jj XXXjj XXj;^., XXX^
"§"§"3^ 'o'o'SS "o'o'^S 'S'o'SS "o'o'SS
ZZ«^ ZZt«5 ZZ«5 ZZ«^ ZZcrt^
Q
D
§88^
D Q> Q> CO
U OH CU tC
S K Ji ": g
ts •* mv m tJ
-------
a.
s
o
QJ
60
•<
PQ
1
a
8
H
4)
VI
.s-
<
c
(D
S
ffi
_c
w
PH
Q
U
tx,
C
ra
to
O
p
U
"u
"3
en
"o
c
.2
S
'c
c
0
u
_a>
u
1
C-,
t^
>£
T3
c
ft
CO
tL.
60
C
1"
o
u
.2
"Jo
"3
c
-f
fc v^
5
ST
^_^
00
tL.
•
>
£~?~
52-
(N
?2
u.
1
Cu
3
O
60
OJ
So
<
QJ
^
(0
c
*
o
! ! ! o i
VC
! ! ! o i
* * * *
rH f-H ^ tN. CO
O O O O* O
V V
CO
tx T-I cO ON M CO
O3 en T— < COcOCO i-*CNlfi
r-i T— i ri
T-« \O O CO
ii I ^-* i ii CO ' ri
aro'co' ^^^ ^f^cn' ^^^-,
COLnr^ OOO3O txvOt^ *~ ' O QQ K^ O O
o'o'o i— < r-i r-i o'oo iriirjio iritxco
^^-^
00 t"*» O tv O ON CO
ON CO "^ CC O O CN CO C4 tx CO
rn r-i **ONlx c- 1 ON LO
S-S-S- ^^Ct rJSiS Sco"<£ ^OO'LO'
""^^ ^"^ dC-C-
ON m TH
rn ON in ON *"-» O C4 ITi
CO ^ C3 Ifl I~H p4 CN ^f ITI CO NO *^
T-H r-t r-t r-i 1C tN ON
^^U >^^H >^^w >^^u. X>^
^ 2 >i ^T*|X ^^JX ^4^J>> ^^J^
1~'i+ '"".+ *? i + ^7.+ r7,i
' iri iri ' iri m»i ir\ 10 irs ICN irN (O
® i— t ^ ^t— 1^3 ^r- 1-^ ^r-<^< ^r-irtf
p
Q
LU O ^
n ° Q E"
S u u .
P a; ai oo
U P* 0- r-J
H oi 06 vo" — .
», K K "*, 5
tx Tj< CO CO^ I — '
co*1 ccT c>f c-^ *^
CN! CNJ" r-5* r-" O
_j
>
-22
8
O
TO
OJ
u
c
u
"c
.&
(0
'J3
-—
S
tft
S
I
•a
c
*
132
-------
0.
a
u
Ǥ
^
pa
•
<0
<
1
B-.
'o
j3
3
S
«
ON
O^
1
1
"ro
C
3
tA
1
"o3
I!
"c
RJ
_u
1
in
^
!
(U
u
§
sS
S
1
^-
bJ
2
01
ra
"S
ra
W
ra
3
a-
ta
j_,
c
'u
C
Is
"s
"^
a.
g
1
J
^
"ra
d § ^
! ! o o
* *
C3 r-t C3 CQ ^H
co o o o in
CO O O ^ CX
o o o o o
V
in ts ** c\ co
o o o o o
CO
\O \D O CO
CO ON O IX CO TH f1* ^J ^o
ON CN CN UJ ON
O O ^
1 ,, soin coi-j* coin
^^.^ ^
M in K ^* o *-• fs. ^^ ^-v
sO^O -<*CN txsO COfN O f ^
o c*^ T~* in o CM in tx os in
^ ^G,
M 15 c5
i— < rH i— « O fN I-H in
FH r~l tx EN.
^
5^o~ ??S^ ^-^^-N •— ^i3- -~~
&C- G-SG- S8 t9°^ °^ fn'
^~* ^""^ »— i
o CN o -^ ro ro
roc^l txt*s T^^ OCT\
rl rH OO IT!
C C C C C
VI Ul VI t/1
5
in
o
d
(0
s
ra
CC
I
en
"s
'•g
1
I
1
133
-------
X
CO
at
a
w
^
o
-3
c
(0
E
£
(ft
Q
U
CH
C
W
tfl
D
D
g
-o
if Selecte
u
w
C
_o
2
"c
o
U
' Predicted
r^
CO
>H
•a
c
cd
00
U.
oo
c
"E
S.
o
U
1/1
c
1
CH
'o
_§
"3
OS
o
0^
JU
^
E-"
UH
0
—
-2
1
3
«
•S
en
(U
JJ
C
ra
_y
I
tft
^^
1
C
U
5
5
a
ra
V
re
E
1
3
$
C
0*
C
"o
U
"3
1
CH
U-<
IS
PH
CO
PH
C<1
00
PH
S
"e
D.
8
5
8_
E*
52.
5
t/^
be
w
52-
bb
X
Ot
«
Si
"ra
C
55 CN t-.
o j i o o
* *
ON rn Cj CO
CO O ^ CN Ifi
o" o o o o
r-irH Ifilri QQ* r-*tri COM
P3 rq co T*< fn tx
I-H \0 ^ O
r-t O O^ ON tN CO r** C^ OO T-t
' CO T— < vO Ifi ^O CO CJ
i i t i i r- <
x— x1— «\ x ^~N
oo *~" CQ rs co co co DN x~v s-~^.
& SS
rg oo o
CN rjn tN tx OO ^O
o^ /^^\ «—
1^ II e?I II ££
CO r-<
so M co co mo
iri^o oo OO'ON MO COM
rfCO IX\O OCO COON
T-* r-t \G 00
n>"(3 aj"(3 Qi*(3 Qj"^3 oj"(5
""§ ""§ "" § "w § "" §
S tin «^PH S P-, ,2 BH S P-,
D
D
B. Q y.
n Q D l"
y u u >7
Q 0) at oo
U CH CH tC
t1 00 CO *«O -—.
oo rC tC rj? D
tC TK- en- en- D
en- en- ws csf U
1
i
o
to
c
1
1
VI
~S
~£
•^
3
(A
&
s
1
*
134
-------
Census Region Profiles
Table 9-7 shows that between FY82 and FY87 there are no differences at the 5%
significance level in the profiles of 2,3,7,8-TCDD, 1,2,3,4,6,7,8-HpCDD, and OCDD.
Furthermore, based on the combined FY82 and FY87 data, there are no significant differences
in the concentrations of these analytes among the geographic regions (parallel profiles). The
profile comparisons for 2,3,7,8-TCDD and OCDD are consistent with the individual test results
(see Table 9-6), which indicate no significant geographic effects for these analytes. There was
a significant geographic effect for 1,2,3,4,6,7,8-HpCDD in FY82, but the effect was not
statistically confirmed in FY87. However, the marginal significance (0.088 significance level) of
the combined tests for geographic effects and the fact that the South and West Census
Regions had the lowest estimated average concentrations in both FY82 and FY87 suggest the
possibility of geographic effects for 1,2,3,4,6,7,8-HpCDD.
The profile analysis suggests that the differences in 2,3,4,7,8-PeCDF and 1,2,3,7,8-
PeCDD concentrations between FY82 and FY87 are constant across all geographic regions
(parallel profiles). This could be due to changes in the body burden levels or, as suspected,
differences in the analytical methods. The hypothesis of coincident profiles was rejected at the
5% level for these two analytes. This could be explained by systematic differences in
measured concentrations. As mentioned earlier, because the profiles for 2,3,4,7,8-PeCDF and
1,2,3,7,8-PeCDD were not coincident, the combined test for geographic effects was not
performed. However, in both FY82 and FY87, the highest estimated average concentrations
were found in the Northeast Census Region. The geographic effect was found to be
statistically significant for 2,3,4,7,8-PeCDF concentrations in both FY82 and FY87, but for
1,2,3,7,8-PeCDD, it was only significant in FY82.
Age Group Profiles
The comparison of profiles by age is presented in Table 9-8. The profile analysis rejects
the hypothesis of parallel profiles at the 0.05 level for four of the five analytes, the only
exception being 1,2,3,4,6,7,8-HpCDD. Even in this case, the test was marginally significant at
the 0.10 level. In FY87, the test for age effects (Table 9-6) was significant at the 0.05 level for
each of the modeled analytes. They were also significant or nearly significant at the 0.05 level
for the same analytes in FY82, but the estimated average concentrations of 2,3,7,8-TCDD,
2,3,4,7,8-PeCDF, and 1,2,3,7,8-PeCDD in FY82 were not increasing with age. Also, the
estimated concentrations of OCDD were increasing in FY82, the rate of increase is lower than
the rate observed in FY87. Thus, even though there were significant age effects for these four
analytes in both FY82 and FY87, the hypothesis of parallel profiles was rejected in each case.
135
-------
The hypothesis of parallel profiles for 1,2,3,4,6,7,8-HpCDD was only marginally rejected
at the significance level of 0.10. In both FY82 and FY87, the estimated average concentrations
increased with age group, but the differences among age groups in FY82 were not statistically
significant, possibly because of large measurement errors. Also, the rate of increase of
estimated concentrations of 1,2,3,4,6,7,8-HpCDD was lower in FY82 than in FY87. Assuming
that the profiles of 1,2,3,4,6,7,8-HpCDD are parallel, the hypothesis of coincident profiles is not
rejected, but that of equal concentrations among the three age groups based on the combined
data is rejected at the 0.002 significance level.
Race and Sex Profiles
The profile analysis comparing race groups (Table 9-9) and sexes (Table 9-10) produced
results similar to those for age groups. The findings are: (1) there are no significant
differences in the profiles of 2,3,7,8-TCDD, 1,2,3,4,6,7,8-HpCDD, and OCDD between FY82 and
FY87; (2) the estimated differences in the concentrations of 2,3,4,7,8-PeCDF and 1,2,3,7,8-
PeCDD between FY82 and FY87 are statistically significant and consistent across different age
groups and sexes; and (3) there are no significant differences in the concentrations of any of
these analytes among different age groups and sexes.
9.4.1.4 Weighted Average Comparison of Analytes Not Statistically Modeled. Four
analytes, measured in FY82 and FY87, but which did not meet the criteria for statistical
modeling in both years, were compared using weighted averages. First, the average
concentration of composites in each of the three age groups was computed for each analyte.
For example, the 2,3,7,8-TCDF averages in FY87 were 2.03, 1.34, and 2.50 pg/g in the youngest
to oldest age groups, respectively. Next, these averages were weighted by the population
percentages from the 1980 census. For 2,3,7,8-TCDF, the weighted average concentration in
FY87 was
1.86 (pg/g) = 0.23(2.03 pg/g) + 0.46(1.34 pg/g) + 0.31(2.50 pg/g).
This type of national average estimate is likely to be more accurate than the simple average of
the composite concentrations, because there was strong evidence from the statistical analysis
of FY87 modeled compounds that age has a significant effect on concentrations of PCDDs and
PCDFs in human adipose tissue. Additional calculations were needed to compare HxCDF
concentrations, since each of the HxCDF isomers were individually analyzed in FY87, while
only the total HxCDF was measured in FY82. Thus, the weighted averages of the individually
measured isomers were summed to estimate the total HxCDF concentration in FY87.
136
-------
Table 9-11, which gives the weighted average estimates and standard errors for each of
these compounds, shows large decreases in the average measured concentrations of
2,3,7,8-TCDF and OCDF from FY82 to FY87. The average measured concentration of OCDF
was 56.0 pg/g in FY82 and only 2.28 pg/g in FY87. In addition, there were significant advances
in the HRMS methodology for analyzing the FY87 composites. For example, as presented in
Table 9-4, the average LOD for OCDF was 19.0 pg/g in FY82 and 1.67 pg/g in FY87. The
standard errors of the estimated average concentrations are also considerably larger for the
FY82 data. These facts suggest the differences may be due to changes in the analytical
method.
There also are differences in the weighted averages of HxCDF and 1,2,3,4,6,7,8-HpCDF.
The averages are 47% and 35% higher in FY82 than in FY87 for the two analytes, respectively.
9.4.2 Comparison of FY82, FY87 NHATS and VA/EPA Study Results
The results of the VA/EPA study, as reported by Kang et al. (1990) and Bauer et al.
(1990), are presented in Table 9-12 with data from the 15 to 44 age group from the FY82 and
FY87 NHATS. No statistical comparisons are made, because the NHATS and VA/EPA studies
had different objectives and data collection strategies.
Table 9-12 shows the average concentrations of selected analytes from specimens
collected in three-year periods beginning in 1971 from the VA/EPA study. The combined
averages for all specimens collected between 1971 and 1982 are also presented. Since the
specimens were taken from male donors born between 1936 and 1954, the ages of the donors
were between 17 and 46 years. These averages are compared with the average concentrations
from NHATS composites containing specimens from donors in the 15-44 year age group.
These composites did contain specimens from female donors; however, there has not been any
statistical evidence linking concentrations of these analytes with the sex of the donor.
Therefore, comparisons of the average concentrations from the NHATS and VA/EPA studies is
possible. The standard errors of the average concentrations and the number of specimens or
composite samples used to calculate the averages are also presented in Table 9-12.
There are obvious differences between the NHATS and VA/EPA results. Except for
concentrations of 1,2,3,7,8-PeCDD in FY82, the NHATS averages are considerably lower than
the corresponding averages from the VA/EPA study. For example, the average concentrations
of 2,3,7,8-TCDD from FY82 and FY87 NHATS are 6.87 and 4.33 pg/g, respectively, while the
average concentration in the VA/EPA specimens is 14.1 pg/g.
137
-------
w
en
S2
g
w
IB
T3 <
^5
+-1 7
en "^
__ tx
™ oo
C s_.
re £
% ~r<
f"^ *«J
SH ^
•5 §
re cs
£ oo
C >"
5 ft
° «-
c 8
o o
U 43
a> c«
bOft
s o
S u
<£^
< -a
T3 a>
- «
Jl £
•SPw
a> u
"^ '^-t
5: o
^~H
rH
OS
O)
3
E2
rx
oo
>H
ft
rrT*
M-J
cy}
xrs
3-< ofc
n^ v ZLT
•y 60,3
»-^ cQ
•s ^ a
C
£ rac3
Si,
z
es
oo
>
ft
uT
en
.0 "56
"S o> )fe
•y M3?
P^ TO
•SPSS a
o> > C
£ racj
«
Z
QJ
r^
re
<
OS ^-s
•<* ^~, ^^. 0
OS CS OO OO
O -^ CO Os
S- ii d S.
VO -a OO
oo r% « '**
rQ ™ V
T3 £ "5
«!'§ fi
S -§) &H g
_**< W^J j-- *H
U'a3 S <~
S> O tn
> u a>
8 8-S 2
& rt -s S3
•t? * -d >
o Sn-5 re
tl «° S -d
S S §3
o > B S
o g 3 *bb
M-H _ M^ *ni
o fro ^
>H S »H **
01 £ OJ vw
'§ ^D'§
Z <"Z en
n _o u "n
138
-------
>
01
J3
•M >~>
S S
0 bO
£2
"Of")
01 vJ
i s
r\ S^j
^? ^
0 c
n-S
§1
^c3
"8 >N
4-t O
0 ^
0) en
V! IJ
O) •»H
en T3
WJ SJ
»H H
O £
-+i en
S'fr
fa
m °O
Cu
en 01
"*3P
o <
% •*
£? •*
r£l 1
c m
R * — i
l£
^
li
en"^
*-l IX.
Poo
fc>H
W PU
T3T3
1 S
ra TO
T3 „
JH x^r**
2 ^
^2
'SBU
"Sb oi
d,W5
<;
&3
TO m
Sri
> r*
<£
•"-. E— 1
Cj ^*
Tl hr1
QJ ^Lj
lz
li
•?fe
rq
i— t
ON
^01
3
TO
H
en
H^
^oo
is
I-
<3 00
2Hn
•v,
1?
Cn rH
^
^rq
Is?
en o
^ 0°
•< ON
> ^
(^^
V,
T3 ON
5^
en tx
-< ON
>rH
V,
T3 so
2^?
en rj<
, tx
~
V
"2 «
3^
^S
•< ON
>rH
Ol
•4-t
3?
TO
<
CO 00 OO rH
co rq •<* co
Tt o rq o< o rq
rH rH
tx oo
oo NO in
NO o tx in tx tx
rH rq •<*! rH
rH
oo NO
rH tX ON 00
•^ O IX 00 O tx
rH ON rH ON
rH rH
o rq
NO in co -rjj
rq rn rq tx rn rq
rH OO rH OO
O ON
NO 00 ON O
rH CJ tx tx rH tx
rH in rH in
rq rq
CO ON tx ON
tx! T-H ON CO rq ON
i— i rq rq rq
^S ^3
S^rq rq^rq
p
g e
u t±
H oi
00 tx"
tC CO^
co" rq"
rq" rn"
oo
oo rq
o rq co
IX rH
bO bObO
sO
SK£
rH rH
in
oo •<* rq
in rH 00
rH
in
ON rH tx
in rn in
r— t
0
O 00 ON
ON rH rq
rH
in
£} rq tx
£5 rn rq
P
P
U
$
^o°,
oq^tx^
tCso"
"^f?
co rq"
rq"rn"
rH
CO
oq rH in
ON in ^ NO ON ON
ON rn rq in
tx
tx
•^H CO NO O NO NO
rH rq rH NO NO rH
— H tX
tx 00
CO tx tX O rH tx
tx rH ON tx tx ON
rq rH rq^ T-H
rH
rq
T+I co rq o o rq
•^ CO OO CO CO OO
rq rH T— i
rH
tx
tx NO tx O "* tx
NO rq in co rH in
rq rq I-H
rH
co
rH sO ON CO NO ON
ON co rq o oo rq
rq Tj^rH
rH
3$& |§^
^H
P
P
U
OH
X
oo
tx"
1 §
rq" U
rn" O
S
o
ffi J
2 ^
s 1
"S •*«
•S cj g
"TO O —
CO ^*. *^^
9 ^
OH J C
% •& <"
O 4-1 (H
H *JH QJ
So ^ |
S TJ — ,
0 £P
^ ^2
« Ue
'B u "^
*^ ^H QJ
n w tJ
^2 g^
a *.g
OJ x-x ..
e o o
•3 z c
OJ '^ en
O-, 73 C
en S
1/2 OJ O
? «'-S
i-C QJ TO
H +^ i-
01 *i
csi ^ S
t N j j ^
Mi
^i|^
•*-• s± J^ ?,
•S C TO g
£•0 * ED
H8<§S>
TO 0 C ^
U^ 8 Sj
0) 4^ >,
Lj • V
IH en ^M
QJ O C ^
SH M-i n. i
l^ss
-0 'C c-
-O TO 01 .5
^l^s
2 s^-3
<" +- ^ 0 x-x
^0^ g, M)
•+-• CH en C 5Jh
0 „,- , S a.
e |^ 8a
g ^ c c&
Concentrations f
(age information
Statistics based o
Statistics based o
Arithmetic avera
Standard error.
RJ i u TJ «
P
P
U
o
43
X
ON
00^
tC
co"
Csf
t-H
U
X
+-^
M-H
O
5/3
F-H
QJ
Ol
73
Ol
3
^
• en
T3 TO
Ol Ol
N C
^1
C ^
TO "-
„ T3
SnS
*rT? ^
VJ W
o 71
^
o «°
S ^>
IH 'K
0 o
en CX
P
6 «
'D rq
Ol 00
&,>H
03 IX,
T3 o>
TO 5J
3 X
T3 ^
"H »H
> 0
1^
.a 01
^ "C
o S
>H !X
Ol Ol
•? ^
g TO
I13
2P
*- 60
139
-------
Average concentrations, with standard errors for additional analytes, are provided in
Table 9-13. The VA/EPA results (Bauer et al. 1990) are compared with the average
concentrations for NHATS composite samples containing specimens from donors in the 15-44
year age group. As indicated, many of the compounds were detected in fewer than 50% of
the composites. In those cases, the reported average concentrations may be significantly
affected by the estimated LODs. For compounds detected in more than 50% of the
composites, the average concentrations from the NHATS studies are considerably lower than
those obtained in the VA/EPA study. The only exception is that the measured concentrations
of 2,3,4,7,8-PeCDD are much higher in the FY82 NHATS.
Two explanations for the differences in the levels found in these studies are possible.
First, the apparent decline in PCDD and PCDF concentrations reflects a decline in PCDD and
PCDF residues in the general environment over the same time frame. This is a logical
possibility resulting from regulations promulgated and enforced since 1970 and environmental
awareness that has focused attention on releases of toxic chemicals via industrial effluents and
handling of hazardous wastes.
The second possibility for the differences observed between the VA/EPA and the FY87
NHATS results may be attributed to storage stability. Since PCDDs and PCDFs are very
persistent, stability is affected by the integrity of the tissue rather than the chemicals. Some of
the specimens in the VA/EPA study had been stored since 1971 before being analyzed in 1986
(16 years later). The NHATS composite samples from FY82 and FY87 were analyzed within
two to three years of the specimen collections.
Although further evidence is required, the data in Table 9-12 suggest a correlation
between time in storage and measured concentrations. However, the effect on storage time is
completely confounded with the effect of collection year. Thus, it is not possible to determine
which factor is causing the observed effect on concentration. To address the issue of storage
time versus collection year, further studies will be necessary to either directly study the
storage effect or identify a surrogate measure of tissue stability. Storage stability can be
studied through development of quality control pools that can be stored with the NHATS
archives and pulled for analysis with each analysis program.
140
-------
Table 9-13. Arithmetic Averages (pg/g), Standard Errors, and Sample Sizes for Selected Analytes
Obtained from the VA/EPA, FY82 NHATS (15-44 Age Group), and FY87 NHATS (15-44 Age Group) Studies
Analyte
2,3,7,8-TCDF
1,2,3,7,8-PeCDF
2,3,4,7,8-PeCDF
1,2,3,4,7,8-HxCDF
1,2,3,6,7,8-HxCDF
2,3,4,6,7,8-HxCDF
1,2,3,7,8,9-HxCDF
1,2,3,7,8,9-HxCDD
1,2,3,4,6,7,8-HpCDF
1,2,3,4,7,8,9-HpCDF
OCDF
VA Study3 NHATS"
1971-82 FY82
2.1C
0.177d
197*
0.6
0.062
197
23.0 39.4
1.071 4.62
197 17
21.4
1.005
197
10.9
0.581
197
3.3
0.192
197
0.33
0.021
197
18.0
0.837
197
36.5 21.1
1.822 3.39
197 16
1.4
0.082
197
3.1
0.290
197
NHATSb
FY87
1.34
0.098
9
0.244*'
0.042
16
8.71
0.543
14
7.13
0.817
5
4.63
0.376
15
0.281*
0.001
2
0.370*
0.043
14
10.7
0.298
15
15.9
1.88
10
0.726*
0.048
15
2.89*
2.07
6
a Includes all study specimens; concentrations not detected (ND) were replaced with LOD/2.
b Statistics based on composites in 15-44 yr. age group; concentrations not detected (ND) were replaced with
LOD/2.
c Arithmetic average (pg/g).
d Standard error.
e Number of individual specimens or composites analyzed.
' * Indicates detection in fewer than 50% of the FY87 composites.
141
-------
10.0 BIBLIOGRAPHY
Andrews JS Jr, Garrett WA, Patterson GD Jr, Needham LL, Anderson JE, Roberts DW,
Bagby JR, Paletta FX Jr. 1989. 2378-TCDD levels in adipose tissue of persons with no
known exposure and in exposed persons. Chemosphere 18:499-506.
Andrews JS Jr., Garrett WA Jr., Patterson DGD Jr., Needham LL, Roberts DW, Bagby JR,
Anderson JE, Hoffman RE, Schramm W. 1989. 2,3,7,8-Tetrachlorodibenzo-p-dioxin
levels in adipose tissue of persons with no known exposure and in exposed persons.
Chemosphere 18:499-506.
Bauer KM, Stanley JS, Remmers J, Breen JJ, Schwemberger J, Schultz B, Kang HK. 1991.
Pattern recognition analysis of VA/EPA PCDD and PCDF data. Chemosphere 23:971-980.
Beck H, Drob A, Kleenan WJ, Mathar W, 1990. PCDD and PCDF concentrations in different
organs from infants. Chemosphere 20:903-910.
Beck H, Eckart K, Mathar W, Wittkowski I. 1987. Levels of PCDDs and PCDFs in adipose
tissue of occupationally exposed workers. Chemosphere 18:507-576.
Bradley JC et al., 1990. Interlaboratory Testing Study on 2,3,7,8-Substitutes Polychlorinated
Dibenzo-p-dioxin and Polychlorinated Dibenzofurion Isomers Standard Solution
Chemosphere 20:487-493.
Buckland SJ, Hannah DJ, Taucher JA, Bates M. 1990. Levels of polychlorinated dibenzo-p-
dioxins and dibenzofurans in the breast milk of New Zealand mothers. Short paper
from the 10th International Meeting, Dioxin 90, Bayreuth FRG, Vol. 1 (Ecoinforma
Press, Bayreuth FRG), pp. 225-226.
Centers for Disease Control Veterans Health Studies Group. 1988b. Serum 2,3,7,8-tetrachloro-
dibenzo-p-dioxin levels in U.S. Army Vietnam-era veterans. JAMA 260:1249-1254.
Centers for Disease Control (CDC). 1987. Serum dioxin in Vietnam-era veterans-preliminary
report. MMWR 36:470-475.
Centers for Disease Control. 1988a. Serum levels of 2,3,7,8-TCDD in Air Force health study
participants-preliminary report. MMWR 37:309-311.
Centers for Disease Control. 1988c. Preliminary report: 2,3,7,8-TCDD exposure to humans—
Seveso, Italy. MMWR 37:733-736.
Cramer PH, Stanley JS. 1988. Analysis of Adipose Tissue for Dioxins and Furans: Adipose
Tissue Sample Compositing. Interim Report No. 1. Office of Toxic Substances, U.S.
Environmental Protection Agency. Contract No. 68-02-4252, WA 27.
143
-------
Cramer PH, Ayling RE, Stanley JS. March 1989b. Midwest Research Institute. Determination
of PCDD and PCDF in human adipose tissue. Data Report, Batches 3, 4, and 5. Office
of Toxic Substances, U.S. Environmental Protection Agency. Contract No. 68-02-4252,
WA27.
Cramer PH, Ayling RE, Stanley JS. March 1989a. Midwest Research Institute. Determination
of PCDD and PCDF in human adipose tissue. Data Report, Batches 1 and 2. Office of
Toxic Substances, U.S. Environmental Protection Agency. Contract No. 68-02-4252,
WA27.
Dewailly E, Tremblay-Rousseau H, Carrier G, Grouix S, Gingras S, Boggess K, Stanley J,
Weber JP. 1991. PCDDs, PCDFs and PCBs in human milk of women exposed to a
PCB fire and of women from the general population of the province of Quebec -
Canada. Chemosphere 23:1831-1835.
Dewailly E, Weber JP, Gingras S, Laliberte C. 1991. "Coplanar PCBs in Human Milk in the
Province of Quebec Canada: Are They More Toxic Than Dioxin for Breast Fed
Infants?" Bull. Environ. Contamin. Toxicol. 47(4), 491-8.
Draper NR, Smith H. 1981. Applied regression analysis, Second Edition. New York: John Wiley
and Sons.
Faccheti S, Fornari A, Montagna M. 1980. Distribution of 2,3,7,8-TCDD in the tissue of a
person exposed to the toxic cloud at Seveso. Adv. Mass Spectrom. 8B, 1405-1414.
Fingerhut MA, Sweeney MH, Patterson DG Jr, Piacitelli LA, Morris JA, Marlow DA,
Hornung RW, Cameron LW, ConnaUy LB, Needham LL, Halperin WE. 1989. Levels
of 2,3,7,8-TCDD in the serum of U.S. chemical workers exposed to dioxin contaminated
products: interim results. Chemosphere 19:835-40.
Fiierst P, Meemken HA, Goebel W. 1986. Determination of polychlorimeter dibenzo-dioxins
and dibenzofurans in human milk. Chemosphere 15:1977-1980.
Graham M, Hileman FD, Orth RG, Wendling JM, Wilson JD. 1986. Chlorocarbons in adipose
tissue from a Missouri population. Chemosphere 15:1595-1600.
Gross ML, Lay JO, Lyon PA. 1984. 2,3,7,8-Tetrachlorodibenzo-p-dioxin levels in adipose tissue
of Vietnam veterans. Environ Res 33:261-268.
Heath RG. 1988. Battelle Columbus Division. Review and evaluation of the quality assurance
program for the study of dioxin and furan levels in adipose tissue of U.S. adult males.
Final Report. Washington, DC: Office of Toxic Substances, U.S. Environmental
Protection Agency. Contract No. 68-02-4294.
Heil GA, Fitzgerald BJ. 1987 (November 2). Quality assurance plan for Rutland resource
recovery facility site specific environmental assessment, Rutland, VT. Vermont Air
Pollution Control Division.
144
-------
Huteau B, Gonnord MF, GiUe P, Fraisse D, Bard D. 1990. PCDD and PCDF levels in human
adipose tissue in France. Short paper from the 10th International Meeting, Dioxin 90,
Bayreuth FRG, Vol. 1 (Ecoinforma Press, Bayreuth FRG), pp. 243-246.
Huteau B, Gonnord MF, Gille P, Fraisse D, and the Interim Network for the Development of
Environmental Epidemiology in Vietnam. 1990. New data on PCDDs and PCDFs
levels in South Vietnamese adipose tissues. Short paper from the 10th International
Meeting, Dioxin 90, Bayreuth FRG, Vol. 1 (Ecoinforma Press, Bayreuth FRG),
pp. 239-242.
Johnson RA, Wichern DW. 1982. Applied Multivariate Statistical Analysis. Prentice Hall,
Englewood Cliff, NJ.
Kahn PC, Gochfeld M, Nygren M, Hansson M, Rappe C, Velez H, Ghent-Guenther T,
Wilson WP. 1988. Dioxins and dibenzofurans in blood and adipose tissue of Agent
Orange exposed Vietnam veterans and matched controls. JAMA 259:1661-1667.
Kahn PC, Lewis W, Ghent-Guenther T, Hansson M, Rappe C. 1990. Dibenzodioxin and
dibenzofuran levels in three groups of Vietnam veterans who did not take part in
Agent Orange spraying. Short paper from the 10th International Meeting, Dioxin 90,
Bayreuth FRG, Vol. 1 (Ecoinforma Press, Bayreuth FRG), pp. 247-250.
Kang HK, Watanabe KK, Breen JJ, Remitters JR, Conomos M, Stanley J, Flicker M. 1990.
Dioxins and dibenzofurans in adipose tissue of U.S. Vietnam veterans and controls.
J Am Public Health Assoc 81(3):344-349.
Knick PD, Hosenfield JM, Turman KW. 1986. Midwest Research Institute. National Human
Adipose Tissue Survey operations manual. Final Report. Washington, DC: Office of
Pesticides and Toxic Substances, US Environmental Protection Agency. Contract
Numbers 68-02-3938 and 68-02-4252.
Leczynski B, Collins T, Sasso N, and Clark J. 1988. Battelle Columbus Division. FY87 NHATS
composite design. Final Report. Washington, DC: Office of Pesticides and Toxic
Substances, U.S. Environmental Protection Agency. Contract No. 68-02-4294.
Mack GA, Leczynski B, Chu A, and Mohadjer L. 1984. Battelle Columbus Division and
Westat, Inc. Survey design for the national adipose tissue survey. Draft Final Report.
Washington, DC: Office of Pesticides and Toxic Substances, U.S. Environmental
Protection Agency. Contract No. 68-01-6721.
Masuda Y, Kuroki H, Haraguchi K, Nagayama J. 1986. PCDFs and related compounds in
humans from Yusho and Yucheng incidents. Chemosphere 15:1621-1628.
Masuda Y, Kuroki H, Haraguchi K, Nagayama J. 1985. PCB and PCDF congeners in the
blood and tissue of Yusho and Yucheng patients. Environ. Health Perspectives 59:53-58.
Muto H, Shinada M, Abe T, Takizawa Y. 1991. The tissue distribution of 2,3,7,8-chlorine
substituted dibenzo-p-dioxins in humans who died of cancer. Life Sciences 48:1645-1657.
145
-------
Neal FA, Olson JR, Gasizwicz TA, Geiger LE. 1982. The toxicokinetics of 2,3,7,8-
tetrachlorodibenzo-p-dioxin in mammalian systems. Drug Metab Rev 13(3):355-385.
Needham LL, Allen KM, Patterson DG Jr, Andrews JS Jr. 1990. Human levels of 2,3,7,8-
tetrachlorodibenzo-p-dioxin in Missouri. Short paper from the 10th International
Meeting, Dioxin 90, Bayreuth FRG, Vol. 1 (Ecoinforma Press, Bayreuth FRG),
pp. 259-262.
Needham LL, Patterson DG Jr, Alley CC, Isaacs S, Green VE, Andrews J, Sampson EJ. 1987.
Polychlorinated dibenzo-p-dioxin and dibenzofuran levels in persons with high and
normal levels of 2,3,7,8-Tetrachlorodibenzo-p-dioxin. Chemosphere 16:2027-2031.
Nguyen TNP, Hung BS, Vu DQ, Schecter A. 1989. Dioxin levels in adipose tissues of
hospitalized women living in the south of Vietnam in 1984-85 with a brief review of
their clinical histories. Chemosphere 19:933-936.
Noren K, Lunden A. 1990. Trend studies of poly chlorinated biphenyls, dibenzo-p-dioxins and
dibenzofurans in human milk. Short paper from the 10th International Meeting,
Dioxin 90, Bayreuth FRG, Vol. 1 (Ecoinforma Press, Bayreuth FRG), pp. 263-266.
Nygren M, Rappe C, Lindstrom G. 1986. Identification of 2,3,7,8-substituted polychlorinated
dioxins and dibenzofurans in environmental and human samples. In Rappe C,
Choudhary G, Keith L (eds): Chlorinated dioxins and dibenzofurans in perspective,
Chelsea, MI: Lewis Publishers, pp. 17-34.
Ogaki J, Takayama K, Miyata H, Kashimoto T. 1987. Levels of PCDDs and PCDFs in human
tissues and various foodstuffs in Japan. Chemosphere 16:2047-2056.
Olafsson PG, Bryan AM, Stone W. 1988. Polychlorinated biphenyl and polychlorinated
dibenzofurans in the tissues of patients with Yusho or YuChen. Total toxicity. Bull.
Environ. Contam. Toxicol. 41:63-70
Ono M, Wakimoto T, Tatsukawa R, Masuda Y. 1986. Polychlorinated dibenzo-pdioxins and
dibenzofuranas in human adipose tissues of Japan. Chemosphere 15:1629-1634.
Orban J, Lordo R, Menton R, and Unger A. 1989. Battelle Columbus Division. Statistical
analysis plan for chlorinated/brominated dioxins and furans from FY87 NHATS.
Task 2-34 Interim Report. Washington, DC: Office of Toxic Substances, U.S.
Environmental Protection Agency. Contract No. 68-02-4294.
Panebianco DL. 1986a. Battelle Columbus Division. Sample design for the fiscal year 1987
NHATS. Draft Final Report. Washington, DC: Office of Pesticides and Toxic
Substances, U.S. Environmental Protection Agency. Contract No. 68-02-4243.
Panebianco DL. 1986b. Battelle Columbus Division. A review of hospital nonresponse and its
effect on standard errors of sample estimates in NHATS. Draft Final Report. Washington,
DC: Office of Pesticides and Toxic Substances, U.S. Environmental Protection Agency.
Contract No. 68-02-4243.
146
-------
Patterson DG Jr, Needham LL, Pirkle JL, Roberts DW, Bagby JR, Garrett WA, Andrews JS Jr,
Falk H, Bernert JT, Sampson EJ, Houk VN. 1988. Correlation between serum and
adipose tissue levels of 2,3,7,8-tetrachlorodibenzo-p-dioxin in 50 persons from Missouri.
Arch Environ Toxicology 17(2): 139-143.
Patterson DG Jr, Holler JS, Smith SJ, Liddle JA, Sampson EJ, Needham LL. 1986c. Human
adipose data for 2,3,7,8-TCDD in certain U.S. samples. Chemosphere 15:2055-2060.
Patterson DG Jr, Holler JS, Lapeza DR Jr, Alexander LR, Groce DF, O'Connor RC, Smith SJ,
Liddle JA, Needham LL. 1986b. High-resolution gas chromatography/high-resolution
mass spectrometric analysis of human adipose tissue for 2,3,7,8-tetrachlorodibenzo-p-
dioxin. Anal Chem 58:705-713.
Patterson DG Jr, Hoffman RE, Needham LL, Roberts DW, Bagby JR, Pirkle JL, Falk H,
Sampson EJ, Houk VN. 1986a. Levels of 2,3,7,8-tetrachlorodibenzo-p-dioxin in adipose
tissue of exposed and control persons in Missouri-an interim report. JAMA 256:2683-
2686.
Patterson DG, Fingerhut MA, Roberts DW, Needham LL, Haring-Sweeney M, Marlow DA,
Andrews JS, Halperin WE. Levels of polychlorinated dibenzo-p-dioxins and
dibenzofurans in workers exposed to 2,3,7,8-tetrachlorodibenzo-p-dioxin. Am J Ind Med
16:135-146.
Phiet HP, Anh TK, Vu DQ, Schecter A. 1989. Preliminary observations on. the clinical
histories, polychlorinated dibenzodioxin and dibenzofuran tissue levels and
2,3,7,8-TCDD toxic equivalents of potentially dioxin exposed patients living in the south
of Vietnam. Chemosphere 19:937-940.
Phuong NTN, Hing BS, Vu DQ, Schecter A. 1989. Dioxin levels in adipose tissue of women
living in the south of Vietnam in 1984-85 with a brief review of their clinical histories.
Chemosphere 19:933-36.
Pirkle JL, Wolfe WH, Patterson DG Jr, Needham LL, Michalek JE, Miner JC, Peterson MR.
1989. Estimates of the half-life of 2,3,7,8-TCDD in ranch hand veterans. / Toxicol
Environ Health 27, 165-171.
Rappe C, Nygren M, Lindstrom G, Hansson M. 1986. Dioxins and dibenzofurans in biological
samples of European origin. Chemosphere 15:1635-1639.
Robinson PE, Mack GA, Remmers J, Levy R, and Mohadjer L. 1990 Environmental Research
53, 175-192.
Rogers J. 1989. FY87 NHATS chlorinated dioxins and furans outlier analysis, Final Report to
EPA's Office of Toxic Substances, Contract 68-02^1293, September 29.
Ryan JJ. 1986. Variation of dioxins and furans in human tissues. Chemosphere 15:1585-1593.
Ryan JJ, Lizotte R, Lau P-Y, et al. 1985. Chlorinated dibenzo-p-dioxins and chlorinated
dibenzofurans in Canadian human adipose tissue. Chemosphere 14:697-706.
147
-------
Ryan JJ, Schecter A, Lizotte R, Sun W-f, Miller L. 1985. Tissue distribution of dioxins and
furans in humans from the general population. Chemosphere 14:929-932.
Ryan JJ, Schecter A, Sun W-F, Lizotte R. 1986. Distribution of chlorinated dibenzo-p-dioxins
and chlorinated dibenzofurans in human tissues from the general population.
Chapter 1 in Chlorinated dioxins and dibenzofurans in perspective, Rappe C, Choudhary G,
Keith L (eds), Chelsea MI: Lewis Publishers.
Ryan JJ, Schecter A, Masuda Y, Kikuchi M. 1987. Comparison of PCDDs and PCDFs in the
tissue of Yusho patients wtih those from the general population in Japan and China.
Chemosphere 16:2017-1025.
Schecter A, Ryan JJ, Gitiitz G. 1986. Chlorinated dioxin and dibenzofuran levels in human
adipose tissues from exposed and control populations. In Rappe C, Choudhary G,
Keith LH (eds): Chlorinated dioxins and dibenzofurans in perspective, Chelsea, MI: Lewis
Publishers, pp. 51-65.
Schecter A, Ryan JJ. 1988. Polychlorinated dibenzo-p-dioxin and dibenzofuran levels in
human adipose tissues from workers 32 years after occupational exposure to
2,3,7,8-TCDD. Chemosphere 17:915-920.
Schecter A, long HY, Monson SJ, Gross ML, Constable J. 1989. Adipose tissue levels of
2,3,7,8-TCDD in Vietnamese adults living in Vietnam 1984-87. Chemosphere 18:1057-
1062.
Schecter A, Constable JD, Bangert JV, Tong H, Arghestani S, Monson S, Gross M. 1989.
Elevated body burdens of 2,3,7,8-tetrachlorodibenzodioxin in adipose tissue of United
States Vietnam veterans. Chemosphere 18:431-438.
Schecter A, Ryan JJ. 1988. PCDD and PCDF levels in human adipose tissues from workers
32 years after occupational exposure to 2,3,7,8-TCDD in 1953 in Germany. Chemosphere
17:915-920.
Schecter A, Fiirst P, Fiirst F, Groekel W. 1990. Levels of dioxins, dibenzofurans, and other
chlorinated xenobiotics in human milk from the Soviet Union. Chemosphere 20:927-934.
Schecter A, Ryan JJ, Gasiewicz TA. 1990. Decrease in human tissue levels of dioxins and
dibenzofurans over nine years after exposure in one male patient. Short paper from
the 10th International Meeting, Dioxin 90, Bayreuth FRG, Vol. 1 (Ecoinforma Press,
Bayreuth FRG), pp. 275-278.
Schecter A, Vu D, Tong HY, Monson SJ, Gross ML, Constable JD. 1989. Levels of
2,3,7,8-TCDD and 2,3,7,8-TCDF in human adipose tissue from hospitalized persons in
the north and south of Vietnam 1984-85. Chemosphere 19:1001-1004.
148
-------
Schecter A, Mes J, Davies D. 1989. Polychlorinated biphenyl (PCB), DDT, DDE and
hexachlorobenzene (HCB) and PCDD/F isomer levels in various organs in autopsy
tissue from North American patients. Chemosphere 18:811-818.
Schecter A, Constable J, long H, Arghestani S, Gross M. 1987. The use of tissue
measurements of 2,3,7,8-TCDD to characterize elevated dioxin body burden of dioxin
from Agent Orange in U.S. Vietnam veterans up to 20 years after exposure.
Chemosphere 18:531-537.
Shecter AJ, Ryan JJ, Constable JD, et al. 1990. Partitioning of 2,3,7,8-chlorinated
dibenzo-p-dioxins and dibenzofurans between adipose tissue and plasma lipid of 20
Massachusetts Vietnam veterans. Chemosphere 20:951-958.
Schecter A, Ryan JJ, Lizotte R, Sun W-F, Miller L, Giuitz G, Bogdasarian M. 1985. Chlorinated
dibenzodioxins and dibenzofurans in human adipose tissue from exposed and control
New York State patients. Chemosphere 14:933-937.
Schecter AJ, Ryan JJ, Gross M, Weerasinghe NCA, Constable JD. 1986. Chlorinated dioxins
and dibenzofurans in human tissues from Vietnam, 1983-84. Chapter 3 in Chlorinated
dioxins and dibenzofurans in perspective, Rappe C, Choudhary G, Keith L (eds),
Chelsea MI: Lewis Publishers.
Schecter A, Ryan JJ, Gitiitz G. 1986. Chlorinated dioxin and dibenzofuran levels in human
adipose tissues from exposed and control populations. Chapter 4 in Chlorinated dioxins
and dibenzofurans in perspective, Rappe C, Choudhary G, Keith L (eds), Chelsea MI:
Lewis Publishers.
Schecter A, Pa'pke O, Ball M. 1990. Chlorinated dioxin and dibenzofuran levels in human
blood samples from various countries, including Vietnam, the Soviet Union, the United
States and Germany. Short paper from the 10th International Meeting, Dioxin 90,
Bayreuth FRG, Vol. 1 (Ecoinforma Press, Bayreuth FRG), pp. 271-274.
Schecter A, Furst P, Fiirst C, Groebel W. 1990. Human milk dioxin and dibenzofuran levels
and levels of other chlorinated chemicals form various countries, including Vietnam
and Cambodia, as compared to the Soviet Union. Short paper from the 10th
International Meeting, Dioxin 90, Bayreuth FRG, Vol. 1 (Ecoinforma Press, Bayreuth
FRG), pp. 267-270.
Schecter AJ, Ryan JJ, Constable JD. 1986. Chlorinated dibenzo-p-dioxin and dibenzofuran
levels in human adipose tissue and milk samples from the north and south of Vietnam.
Chemosphere 15:1613-1620.
Stanley JS, Cramer PH, Ayling RE, Thornburg KR, Remmers JC, Breen JJ, Schwemberger J.
1990. Determination of the prevalence of poly chlorinated diphenyl ethers (PCDPEs) in
human adipose tissue samples. Chemosphere 20:981-985.
149
-------
Stanley JS, Boggess K, Going JE, Mack GA, Remmers J, Breen J, Kutz FW, Carra ], Robinson P.
1986. Broad scan analysis of human adipose tissue from the EPA FY82 NHATS
Repository, pp. 161-179, Chapter 14 in Environmental Epidemiology, Kopfler FC, Craun
GF (eds), Chelsea, MI: Lewis Publishers.
Stanley JS, Bauer KM, Turman K, Boggess K, Cramer P. 1989. Determination of body burdens
for polychlorinated dibenzo-p-dioxins (PCDDs) and polychlorinated dibenzofurans
(PCDFs) in California residents. NTIS Publication No. PB90-148289.
Stanley JS, Bauer KM. 1989b. Chlorinated dibenzo-p-dioxin and dibenzofuran residue levels
in food. NTIS Publication No. PB90-146523.
Stanley JS, Boggess K, Onstot J, Sack T, Remmers J, Breen J, Kutz F, Robinson P, and Mack G.
1986. PCDDs and PCDFs in human adipose tissues from the EPA FY82 NHATS
Repository. Chemosphere 15:1605-1612.
Stanley JS, Ayling RE, Cramer PH, Thornburg KR, Remmers JC, Breen JJ, Schwemberger J.
1990. Polychlorinated dibenzo-p-dioxins and dibenzofurans in human adipose tissue
samples from the continental United States collected from 1972 through 1987.
Chemosphere 20:895-901.
State of Connecticut. 1986. An act establishing a program for monitoring dioxin emissions
from resource recovery facilities. Public Act 86-332.
Tong HY, Monson SJ, Gross ML, Powell LBM. 1989. Elevated levels of 2,3,7,8-TCDD in tissue
of agricultural sprayer of herbicides. Chemosphere 18:469-476.
USEPA. 1986a. Broad scan analysis of the FY82 National Human Adipose Tissue Survey
specimens. Volume IV. Polychlorinated dibenzo-p-dioxins (PCDD) and
polychlorinated dibenzofurans (PCDF). EPA Publication No. EPA-560/5-86-038.
USEPA. 1986b. Analysis for polychlorinated dibenzo-p-dioxins and dibenzofurans in human
adipose tissue: method evaluation study. EPA Publication No. EPA-560/5-85-022.
USEPA. 1987a. USEPA. Interim procedures for estimating risks associated with exposures to
mixtures of chlorinated dibenzo-p-dioxins and dibenzofurans (CDDs and CDFs).
Washington DC. EPA/625/3-87/012.
USEPA. 1987. 40 CFR Parts 707 and 766. Polyhalogenated dibenzo-p-dioxin/dibenzofurans:
testing and reporting requirements; final rule. Federal Register 52:21412-21452.
USEPA. 1989. USEPA. Interim procedures for estimating risks associated with exposures to
mixtures of chlorinated dibenzo-p-dioxins and dibenzofurans (CDDs and CDFs) and
1989 update. Washington DC. EPA/625/3-89/016.
USEPA. 1990a. NHATS broad scan analysis: population estimates from fiscal year 1982
specimens. EPA Publication No. EPA 560/5-90-001.
150
-------
USEPA. 1990b. Pattern recognition analysis of VA/EPA PCDD and PCDF data. EPA
Publication No. EPA-560/5-90-006.
USEPA. 1991. Water programs; guidelines establishing test procedures for the analysis of
pollutants. 40 CFR Part 136. Federal Register 56(26):5090-5122.
VA/EPA. 1990. Dioxins and dibenzofurans in adipose tissue of U.S. Vietnam Veterans and
Controls. EPA Publication No. 560/5-89-002.
Van den Berg M, Van der Wielen Frans WM, Olie K, Van Boxtel CJ. 1986. The presence of
PCDDs and PCDFs in human breast milk from the Netherlands. Chemosphere 15:693-
706.
Weerasinghe NCA, Schecter AJ, Pan JC, Lapp RL, Giblin DE, Meehan JL, Hardell L, Gross ML.
1986. Levels of 2,3,7,8-tetrachlorodibenzo-p-dioxin (2,3,7,8-TCDD) in adipose tissue of
U.S. Vietnam veterans seeking medical assistance. Chemosphere 15:1787-1794.
151
-------
APPENDIX A
FY87 NHATS COMPOSITE DATA LISTED BY ANALYTE FOR EACH COMPOSITE
A-l
-------
a
UJ I/I
j y
*
3 D
I/I U.
uj o
M 9
VI
O I/I
a. z
O X
uo
M
1/1 a
«s
^£
P* Z
>- at
u. O
U.I
Ik ^
ou
a ae
zo
M U.
I/I
^
_J
ti.
a
u
i-
00
J^"
.
CO (D
-r-
U. 1 O
Q CM
O ~ n
1- U
i n o
GO ^ *^
r» u »
- ac
c}I§
CN TQ U
(S -f
C I/I Ol
3 L.
OCA
O +>
U n
c
m
§
«i
c
01
•M
c
M
L.
01
a o
« u
01
0
c —
o +*
M a
c
o
fll -f
*• o
flj .„
O L.
VI
01
a
___ ^
0 Ol
n HI
82
-—,
0 O)
Ox.
^•"l
ft
•^
1.
01
(0 M-
^ '^
1C *-
Q (0
*^
C 0)
«^
**
8S
0) 2
a.
+rf
C 01
0 «
01
Q.
01 S"
O) O
"*«
3§
in •-
01 01
UK
C
* §
i"
a
^^
a
n
1/1
X
O (.
*• 0)
(Q J3
i. B
o 5
r
u
(B
CD
r-OOaJUJf-^tor-oi^'CitD
(OCM(0(0(O(Ococ--(0(0'~oiin
r-lOiflCOr-cor-r^f-r-oor-r-'
^
r- (DO
•^ O ~
O CM
t^oi^oocooin^oicMiflcofo
^O^^^PJ^^^^PJOO,
O) 0) CO Ifl ^ CO (O O)
* O co in »• IB ID CM c-
do ddo o odoo
o o o
ooKoaaetaxaaaa
or-ooiaiocootuoocncM
otincMMincocooioO'-oi
^ ^
oooooomooo^mo
inOOOOO'-QOOVCMQ
CM ino(oooov(oo
" *" "
W W ^ ^ ^ CO ^™ CO CO W CO W W
U (J CJ U U U U
W O) O ^ ^ O CO N CO O 00 CM ^
'-'-CM'-'-CM'-COCMCM'-COeM
r~incooocMcn — TfO(D(OOfO
^ W N ^" ^ (O (0 04 Ot (**• U1 f*** 00
CS^OCO^O^*"CO^CS^CS
8888888888888
QOQOQQQQQQQQQ
^mr^oooo^cNCo^r^ooot
minifluimtDtotofocototDU)
C4C4O4MC4C4CNCNOIWMWOI
** CO P* CS O> ^
c? ^r o1) co in o
oo r* r- r* p* r^
'-»-«-
CM r~ ID
oo co t »-
^
O O O "O
o in
»- CM
o a a o a a:
in o ID in » in
to co CM (D i r-
o o o CM o co
8 S ° 5 8 2
"
CO CO CO ^ *" CS
^ ^ ^ ^ z" 2
r- in w r^. oo (0
»-««'-'-*-
P* O> (0 (D ^ *
00 00 O) ^ *™ (**
O ^ O ^ O c^
888888
g § § S § S
<<<<<<
o CM «t in us t-
r^ P* r^ r^ r^ r^
CM (M CM CM CM CM
CM CM CM CM CM CM
O
CM
ID
4f
in
-
^
o
a
ID
O
Ol
O
O
CO
UJ
CM
CO
in
co
8
S
u
Ol
f^
CM
ID
CO
in
n
IO
CM
^
d
|H
ai
ID
CM
O
at
a
CO
01
O
o
CO
*
(D
CM
Ol
CM
8
CO
a
u
o
a
CM
(D
CO
CM
qi
in
^
*
T-
CM
CO
o
a
CM
co
O
o
CM
«/,
Ol
^
in
CO
8
co
a
°.
^
co
CM
ID
CO
co
CM
ID
in
in
o
u.
|H
f-
co
0
in
CM
OS.
r-
0)
o
o
o
CM
z-
<»
CM
^.
CO
CO
a
^
CM
co
CM
(0
CO
A-3
-------
_^
< 1-
i§
_j a
a. u
CO < 00
at
iu 3 r-
1- U.
IH co (o
f- f*«
CM •
L. 1 O
z a CM
o - < ii C
u. z re -H
IH TJ -M 0»
U. OS C (/) 0)
O O 3 L
-i o c re
(•IT n o ^
w ^ u u ^
z o F-
IH O +•
t— ae u re
l/l O -f
-j +•
^
re
£
(U
+J
c
M
X
L.
W >
a o
w O
(U
c.2
0 **
IH re
a
§
re -H
•H O
a I
VI
u
et
^^
00,
c32
a ro
o \
-1 O)
L.
0)
fll OJ
u at
c
f£
U O
Q) O
O
IH
0)
re
10
X
L
O C.
re n
o 3
re
r
u
re
oa
N
CO
(O
CO
in
CM
in
CM
CM
d
o
CO
CO
en
o
o
CO
2
in
on
O
CO
8
00
a
u
CO
oo
(O
CO
CO
CO
(O
in
en
»-
p.
^
d
a
o
8
o
8
T~
u
z
"
CO
oa
CO
CM
8
00
00
CM
(O
CO
OOcoincMoo*-in
inmo«-coa)coen
<0tt,<0tf,(D(0r-l0
fo o r-
O CO «»
•*• «- o
H4 I*H hH
"?-SSSoS
CM'-t-CO'.-'-i-CM
en (0 ^~ (O co
co in «- — O
^ T* f^ ^ T^
d d d d d
o m o
»• CM *-
ataooaaatoc
tr-omocMOfo
oooooooooooOoo
O'-OOOCMOOO
Or*c3oo'-ino
in o co * *
COCMCMCO»-COCOCM
UJ O O IU
W*ZZ»tOZZ
r~*O(Dinr^«too
^ ^" CM ^" ^ CM ^~
S2£*5co8S
COCM'-CJCM'-*-'-
88888888
aaaaaaao
uuuuuuuu
oocooocncncnonci)
CMCMCMCMCMCMCMCM
cocococo^^v^
inininooocoo*ioco(oor-
oncMCMVcoooocointoocnoo
CD co co ^* (o r*" co in co CD co in in
CM co t- in
CM o o on
~ ~ - b
IH IH IH IH
« in
in coMtoincncor^cn cocn
O^"O~CM^»-CO^»-CMCMCM
r^ ^^(ocncoco^
a> v cncncMcn^nooen
CO CO •^•»-»-O'--^CM
o o oo'ooooo
in in in o
CM CM CM »-
eraoeasoaaooaaao
r^ooocoocoinoocMencMO
coinr^'*o^iT"Oininco^o
r^or^ooocDCOOoooocovO
OOtOOOCMOOOOOO
§OtOOO(OOOOOOO
r^wo* «CM oo
^ *" ^ ^*
CM'-MCMCOCMCM'-COCMCO'-CO
w o ^j y u y^
tOZZZXifl^XOOZZ^Z
onooocoo'frcoinot-oof-en
woncM^cnco^incMOcMr^co
^^co'VCMtticMcor^inincM^
CO^OOCMCO^CMCOO^CO*
8888888888888
88gS88888gg8S
en o ^ w ^ co oo w O ^ co ^ in
enoOOOOOO^'"^'^'~
CMCOCOCOCOCOfOCOCOfOCOCOCO
CO(O(OCO(D(DCOCO(OCOCOCOCO
^r^nv^uiminininininin
A-4
-------
K
< 1-
1- Z
< o
Q U
UI
0*
5 2
en 3
Q£
UI 3
t- U.
en Q
21
otn
o z
K4
en x
1- O
< W
* Q
r- ui
oo H
>• 4
u, z
M
ti Of
O O
t- oc
en o
M U.
~*
U.
a
o
i
oo
.
r-
.
n 03
• f^
CM •
U. I O
8M
«- II
t- O
i n o
co — -
. +J
t- II (S
"He
(N M O
13 **
n C
<« *•
C-S8,
3 L.
O C 10
Q.O H
u m
•p»
c
ra
§
»•"•
«i
c
Q)
c
fr"l
X
Jll
w
l/l >
a o
at
o
c —
o **
n fl
a
c
0
(fl **
+• o
a I
in
01
a:
-— *,
0 O)
82
_
Q cn
O '**^
w
0)
4^ .^
ni ^
a n
•j
»
C 01
01 •H
0 •-
tt) 2
a.
44
s«
01
a.
8,1
< c.
u
U) C
3 0
VI •-
C 0)
01 01
u «
i c
f" 3
0
0
01
1
t/l
X
L.
o s.
+* 01
n 5
C. E
o 2
(Q
•J
j=
u
A
00
T
ID
in
cn
o
oo
M
^
b
a
^
f^
«
o
8
M
to
1-
(0
n
n
8
oo
Q
U
(0
CJ
(O
^
in
i—
r~
^
en
M
cn
oo
n
O
o
a.
o
b
00
o
8
n
en
in
r-
O
i
r-
s
u
f-
a i- a. o < at
t- z a z ui
I/I UI Z UI O M >
UI O O t- " X Q
at z w z t- < 3
O t- ui < Z ui
< 03 z a ce
i- -i 3 x
< u ui < z < en
a o ae o ui or
CM u u. M a. M
ce ^
O O n ii ii ii ii
u.0
_i o u- M Q. a
a
UI
en
3
en
UI
O
O
u
A
-------
t-
^
a
Ul • at
u. O
ii T
oo
s§
1-
co
-1
a
o
o
1-
1
00
^
-
n co
cs •
Q i O
O M
O ••- u
H U
1 CO 0
^
r- II <8
- a
CM (Q O
"1"
C CO O)
3 L.
o c ra
f^K
U
a o
i O
01
o
c —
o **
11 (8
a
c
0
(fl -H
+- 0
(8 •"
a L
VI
0)
or
om
rt H*
82
a o>
0 v.
-1 0)
a
^"
c.
.^
a <*•
(8 —
O (H
bJ *V
O
C OJ
o|
0) 2
a
4^
C 01
on
0)
a.
a
u 3
O)O
< «-
(9
M C
3 O
VI -~
C O)
C
o o
uu
a
o
0)
„
X
L.
O (.
Si
si
o f.
(8
£
O
a
r-
co
O)
^
(O
8
d
a
o
§
^
0
in
(N
M
111
z
w
1^
CM
8
00
a
u
<
in
CM
CO
O«coooooincnr~O(OOc)incnNoo(Moenoon<-
c»jooc>fM»-(oeoo»(oco^tnin*enr^*-»oo)int-»
LncDCOf"oo(DCDiOr*>o>r*>t>*>[^t0(DCQr>-(Dt^(Qtfl[N*
*™
oo n (o n in
<» r- in 01 o
0-0 0 -
DC M M M MM
— cs — n ooooco 01— »• incni*-c»f-t—
— CM — <- — (Hoor^neninenOcno — c^ncnoovn
^ ^K ^
en «- MioiD en — invin toen en*-
Sn d csoomMncnincnao owr-^oi
w — '-{^'•'•'"wwwoi cscotoncM
oo o dodoooooo o o o o o
in in o in in
ooaaaoroaoaaaaaooraoaoaa
r-ocnenocaenuoocnNinocomvinioooNr-
cnS3S!coS«cooi8^*|5SSr25£iS)SS
^
oooooinooO'tinoooocMooooooo
OOOOO — OgO^CNOOlOO'-OcoOOOO
inocaooo^ioooin ^o— o
"""" ~ "
CM *~ ^ ^ CO ^ C*) C) CM CO CM O 09 C*) CO *" *"" CM CO CO CM! CM
uo o uj ujuj u o o o w u
jg^mi/i^t/i^pi^^jF^^^f^^^^^jg^pt/*^
28*28$2S3g2SIS-SS-™«?522*
innooNen-^oucoonr-enuiov^incjiv^
§^O^OCS*~C4C*)^
§000000000
ooooooooo
coooeococooooococooaoooooooooooocococooaaooo
aoaaaQaaaaaaQQQQaaooQQ
(JOOOOOOOOOOOUOOOOOOOOO
<<<<<<<<<<<<<<<<<<<<<<
(OCO(OCDCOCOCD(DCO(OCD(OCD(0(OCO(OCOCO(OCOCO
A-6
-------
~
HZ"
aS
U \J
3~ §
to
Z i
at
U ^3 C^
1- U.
IH co (0
CO If) 00
CO W CO C*5
N
^^
a
0
r- oo
*~ cn in
in co *
*"
en »
oo in CM in
cn *- »- CM
O O O o
So a o
O « r~ o
o CM in in
O 00 CO 00
o o <- o
§o r« o
in o
»- CO CM CM
U U
co r- «• o
^* ^ w
00 ^ O ^
o oo ^ to
N CO N *-
8888
o Q a a
^ (0 00 O)
00 00 00 00
C* CM (D en »•
ooooooo o o o
o o o o
ooaaaoactaaaaQca
inocMOcot^ooocoocoinoo
oooooooor-(or«oooiO(OQco
OOCMOOOOOVOOOCMOO
oo'-moootoootooo
OOV^f Ot~*O* *CM
CO^COCOCMCM'-CMeMCOWCM'-CO
O O LU LU O O
Z2(/9ZZCOZZZ7(/)CO7(/I
(Oinr^^rooaioooioovromo
iiiiiiiiiiiiii
§00000000000000000000000000
aaaaaaaoaaoao
uuouuuuuuuuuuo
Soo en o
o o ^
CMCMCMCMCMCMCOCOCOCOCOCOCOCO
co^^^^f^^^^^tninmin
oo
in
r-
co
O)
^
0
d
o
CM
in
00
o
o
CM
Z
r*
O4
i
8
00
a
^
T-
CO
(0
in
o
en
(••
„.
*
(0
q.
J:
d
a
CO
00
00
o
8
CO
z-
oo
CM
in
I
00
a
CO
CO
(0
in
cn
in
r-
CM
cn
,-
CO
CO
CM
d
o
CM
«
0
8
^
to
r-
r-
CM
CO
8
00
o
U
«f
CO
(0
in
in
(O
(O
,_
CO
d
H4
N
^
t^
O
at
o
8
o
o
CO
2*
cn
CO
I
oo
a
in
o
(0
in
A-7
-------
. ,
< K
< O
O U
Ul
-I
Q.
Z CO
3 z
CO <
at
UJ 3
1- U.
^
co a
a <
SCO
z
(/) X
I- O
Is
a
t- UJ
co 1-
£!
n
u. a
o o
a x
z o
n
t- a
co a
HH U,
a
a
u
a
~
r-
-
n (o
. js»
CM •
§1 O
CM
O "- ii
1-0
i co o
00 »- —
r» n M
at
CO B
- L C
CM IB O
ii C
ID -H
•B -< z
CO UJ
> o
_J CO O!
3
n -I O
K- < -1 h-
(S>Z< CO
•iSo zo
H w < t
< -1 Z 0
h- < UJ CO U.
CO 0 X 0
11 O t- CO
Z Z ui UJ
M ui Z Z >- O
x M co z
o u H 3
u Q o a at
co Z ui Z UJ
am- t- z
0 « «
§z i- z n a
o z at < i-
M M U y jiJ ^ Q
1- t- CO 9 H UJ Z
U < U UJ M CO
n a at a. a >-
DC i- a. u < a
i- z a z ui
CO UJ Z UJ O <-> >
UJ U O H >- X O
K Z M z 1- < U
O H UJ < Z uj
< u D i a a
H _l O X
< « UJ < Z < 10
a o ct o uj o
CM o u. M a. M
a \
O O n n n u n
U. O
_i o u. 11 a. at
a
UJ
CO
a
CO
UI
Q
O
u
A-8
-------
1-
^
a
UJ U.
UJ Q
h- Z
*•« 4
00
O (/I
a. z
Z IH
O X
u o
l-l
i/) a
i-
^S
CO l-c
>• a
u. O
O U
a at
20
i-
M
-1
u.
a
u
UJ
a.
00
CO
u. CM in
a -in
o •*• •
u i »-
a. CM
i *- u
oo U
-co o
co" it ^J
- a
CM 'O
•"- S
^ (Q O
•DM
ii C
•O -H 0)
C I/I 01
3 t-
6cm
o **
O (8
£
c
m
3
_
(Q
C
(U
4*
C
1-4
X
0)
O 0
a
o
c —
0 -H
w flS
Of
c
0
IS -H
v U
a .-
a L
VI
0)
>-v
U 0)
C v.
82
"— •
a o)
3m
2
i.
01
ns <*.
Q m
3
+*
§2
gr
L. £
01 2
a.
+i
c u
01 ^*
0.
0)
a.
a
0) 3
no
< L.
(9
u> c
3 O
c 5)
U 01
u a
B C
gj
(/I
a
01
*
X
L.
O t-
ns J3
i. g
o 3
u
(TS
CO
o
0
CD
CM
V^
0
o
(O
S
o
§
o
8
»•
o
in
CM
CM
z
fs.
^f
M
8
CO
a
^
^
in
CM
(0
O *~ M W
in o r- r~
r- * r- eo
CM CM CM
X^ "\ X^
at. a a a
o o o
00
CM
^
o
CM O> CO t~
o ? »- 2
O O 0 O
2 i § §
r~ o en en
* in CM CM
o o o o
O O O ^J
in
CM •»- »- »-
S S w ui
o> o * *
in co oo CM
N W ^ ^
§co *~
8 8
00 CO 00 00
8888
< < < <
U) r** oo o)
in in in in
CM CM CM CM
(0 (O 10 (O
focor-^CT-enenMCMcneno
P5CMr-vcM(OP-iOoooo(OeM(O
oooooococncMQor**cnoooocooo
CMCMCMCMCM CMCMCMCM CMCM
•x^ v^ -x^ *s^ ^^ -^ ""s^. "x^ X^ «x^ "x^
aaaaoaoaao oo
ooooo oooo oo
in
ID
oo
o
co^ooco co co t^ in ts co
"ScomSSSwnSSSS
ooooo»-oooooo-o
gggggggogocfgo
OcocniooocncMinoioin^1
uici(oa>OQ'*cn(OoocM(o«
omooo^inoooocMO
§»-OOO*CMOOtDO^'O
^lOOOOtCDOOUl "TO
CO^'COCOCMCOCMOrOCOCO^"*"
O lu liJUJ O OOO
Cs"fOCM°2ocM-S(N-2
ai<-«O(0(OOnr-en(D(o«
fcocACflr^mf^oofloooo)^^
f^1 O ^ O
OOOO
0000000000000
(JUOUOUOUUUOOO
<<<<<<<<<<<<<
S^cMCo^r^coenocM^infO
cototDODiOtocor^r^r^r*^
CMCMCMCMCMCMrMCMCMCMCMCMCM
(O(OIOIO(0(O(0(OIO(DIOCO(0
(0
q.
CO
CM
•x^
a
0
in
o
a
in
00
00
oo
CM
*
(O
q.
r-
CM
8
CO
a
<
r-
r-
CM
CO
N
co
w
(0
CM
>^
Q
O
^
(0
o
i
(0
o
0)
o
o
m
UJ
CM
in
ao
8
co
a
<
en
r-
CM
co
CO
r~
(O
CO
CM
«x^
0
o
f*
•0
o
i
oo
CO
0)
0
o
CO
•*
CO
CM
en
CM
8
ao
a
^
o
00
CM
CO
CO
1-
_
00
CM
X^
a
0
^
CO
(0
o
§
CM
00
O
O
CM
I/)
en
^
in
8
00
8
<
v-
co
CM
(O
0
r-
00
CM
X^
Q
O
en
in
O
Q
t~
T-
0)
O
o
o
CM
z1
CM
^.
CO
8
CO
a
u
<
CM
oo
CM
(O
A-9
-------
^
< K
< O u.
O U O
— u
UJ UJ
-1 O.
0. I
Z (/) 00
< z
(/) < r-
at
UJ 3 CO
1- u.
M U. CM U)
01 Q o -in
o z o •- •
O. < UJ I *-
§0. CM
i/) i »- u
Z CO U
M -tO 0
< n C
u. Z £
ft) 3
a
^1
C ft)
ft) ^"
0 «
01
a.
a
01 3
J'g
*4 *•
(3
V) C
3 0
I/) —
C Ol
01 01
u a
c
'*" 3
f-
^
00
to
8
00
8
<
S
CM
CD
CO
r-
00
CM
°
O
^.
en
b
at
t*
in
00
*•
r-
in
CM
X
«
o
CM
8
S
u
oo
00
CM
(0
r* to in r** o oo co ^ oo oo to in co CM ui ^ CM ^ CM
O in ^ o ^ r™ CM CM oo in ^ CM co CM co en r** o CO
oocoooencncncnenr-cnoocncncDoooocnoooo
to en co
o> '-in
CM >- b
CM CM CMCMCMCNCMCM CMCM CMCM CM
Q QMQQQQQQM o o <-> o o o
o o oooooo oo oo o
*• ao co in
CM «- CO CO >- 00
* in to r- co oo
*- o o o o o
^-co^ cnco^eninen csr^^ ^cMcno
oa »• in W'-ocM'-en r-en* o n- oo co
CMCMCO
-------
^^
Zfe
< o
QU
Ul
Q.
5 "*
CO <
at
Ul D
I- U.
n
00 Q
O Z
o. 9
§co
Z
M
oo x
H 0
< M
Z O
Z
a
r> u
Ui Z
u. a
o o
O X
z o
" a
CO O
11 U.
u.
Q
U
Ul
a.
00
^
-
CO
u. CM in
a -in
o *- •
Ul 1 »-
a. CM
i i- u
00 O
-co o
•H
CO II A
- tt
CM TJ
- i. C
•- A O
•DM
ii C
A +•
•D +* 01
C CO Ol
3 L
OCA
O *•
U A
44
•—
c
A
3
«
41
C
»H
c.
0)
CO >
a o
H U
at
a.
o
c —
o *•
11 A
C
0
A -M
•H U
A •-
O L.
M
0)
a
got
2 x^
82
>^
^^
O Ol
o -».
n
>_^
C.
0)
A *
*< —
A —
a A
v
C 0)
01 +>
u •«-
t. £
at 2
a.
+J
C 01
u "A
t-X
01
a.
S3
o
i.
O
3 0
in —
C 01
01 01
ua
B C
ffy
Q
01
0
A
X
s.
o c.
+> 01
A ^
Q 3
43 Z
A
J
£
U
A
m
•*- o
CM Ol
en o
*-
CM CM
^^ NS^
o a
3 3
in CM
(D 10
6 6
§ §
*- o
r-^ 6
« 00
0 0
8 8
CM CO
CO CO
t- in
S S
CO *
80
0
r- r»
8 U
^ ^
(0 r^
n ro
ID 10
^^ ^
in in
P^
v^
a
0
CO
S3 S
2 «/. §
< M Ul
z i/i a.
co -i o
-i M < in
o > * ^
c^j e
i/l Z < I/I
M < U Z O
H M < •»
< -IX U
h- < Ul I/I U.
I/10X 0
n O t- CO
Z X Ul UJ
nutzx>- a
X M 03 Z
a u t- 3
ui a o a a
CO Z Ul Z Ul
D w t- )-Z
i-« i— . ni \/ nj ft r^
K H I/I < h- Ul Z
U < Ul Ul 11 CO
MCE a a. a >
at t- a. u < a
1- Z Q X Ul
co ui Z UJ O M >
ui o o H M x q
a z« z t- 2 Q
O t- ui < X Ul
v
Q Q II II II II II
u. o
_i o u. 11 a. at
Q
Ul
3
CO
Ul
Q
O
U
A-ll
-------
J-
0
Ul (/>
s!z
Z at
I/) U.
oi Q
f- Z
l-f <
(/)
O M
1 5
Sxx
0
01 Q
1-
< a
^5
[s. 2
00 N4
> a
u. O
U_ T*
u* ^
oo
sg
M U.
D>
3
u.
Q
0
U
a.
00
.
L. CM in
a -in
o *- •
Ul 1 T-
a. M
00 O
-M O
v ii ra
- at
n TJ
• L C
w ra o
ii C
TJ +• 01
C I/I O)
3 (.
o c ra
o Z
u ra
•*•
c
ra
c
01
c
(-
0)
a o
M U
at
0
O v
M ra
Of
§
ra •f
•H 0
O t.
I/I
01
0 O)
82
a ai
o ^
a
C
01
(Q <*-
(0 i—
Q (w
a
+*
c v
u ^
?i
01 y
+4
C 0)
ura
01
Q.
«. 3
< 2
Ifl —
C 0>
SOI
at
c
i C
O O
So
(/I
o
01
ra
X
L.
O L
•H 01
ra 3
c- E
O 3
fl
£
U
CO
OO'-««noor~'-
Oinror-r-cocMf-^
cnr^^r^cooooocooo
M r* »t
CM (D CM
•» (M in
^
KM M M
(O (O n (0 en co M
cnm^rocor^CN'-cn
CN ^
ID * (0 — *
(O cp o> in r- w
O O O O O O
o in in
aoaaaaSoa
art k^ rt kd A_ i^ ft A_
U, p^ (L ^^ (L M & &
or«oo>o>oooan(0
8SSSSSSS3
^-
ooooooinoo
tnooooo^-op
M Ul O '-tO
iiiiiilii
oooocooooooooooooo
QQQQQQQQQ
S^ W CO
(0 (D (0
Ck4NCc5(O'-r*
in >- CM oo «t TO a) to en
^r-fflwvr-onencx r- r»»-
coencMcMCocococMn in- in t
ooooooooo o oo
o o
aooatoaoaaatoatao
OoenNinofflin^inioooesr^
S8^*i2ooSt25wSSSS
^
otinoooocMOoooooo
StNOOCDO — OooOOOO
^ us o o in *o»- o
(MCOWCOCOWW'-'-CMCOCOCXW
UIUI O ^^^ ^ ^
85SSSSS£?!£S£2S
"'"'^'SooScSS^'SSinco
liiilliliiiiil
oocnooooooooooooooaooooocooo
Qooooaaoaoooaa
^r^oowOc^^finujr^OTO^oj
(0(O(0(ot^p*p*t**ri"r^ts»GOocjco
O4C4dWWWO4NWWNWdW
(0(0(DH)tOCDCD(D(DtO(0(0(O(0
•^ M M M CN N CM (N CS CN 00 CO CO CO
A-12
-------
^^
< K
t- Z
< O
_
w
5 *"
I/) 4
0!
U 2
K U.
I/I Q
25
8$
M
(/I JC
52
*Q
si
1-4
u. at
O 0
i- a
en o
HI U.
J
fc
U
o
UJ
a
00
r-
.
CO
u. CM in
8-"?
UJ 1 »-
a CM
i *•- u
*2.2
V II 19
w-? c
"52
ii C
(9 •H
"O +> 0)
CM ft
O C 19
fft w
.2
Is
V
c
19
3
^
m
E
Q)
•w
C
1-4
X
t.
0)
(/) ^
a o
DC
0
c •-
o •H
1-4 (9
a
c
o
19 V
•H U
(0 —
a L
VI
01
a
^-*,
00.
A pjt
cSa
•^s
-— .
0 O)
o ^
Q
s_ '
01
19 H-
0|
C 0)
m ^
Si
a.
s*
iy ^
01
a.
0) §"
O) O
< a
vi c
3 O
C O)
01 0)
u a:
c
§
2.°
i/i
a
H
01
(9
in
X
o c.
(9 j5
(. E
19
U
19
m
en
CM
en
O
^i
r-
a CO
oo r^ co
CM *-
CM C~ CM
CM t~ (0
CO •r- CM
odd
a o o
rt rt fl_
u, u. u>
« c^ O
CM in in
oo oo oo
O «- O
d ^ d
in o
CO CM CM
IU
i/> •» z
r> * O
«- o r-
GO ^" (D
ON*"
888
00 00 CO
a a o
u u u
00 00 00
CM CM CM
(0 10 10
CO 05 CO
coint^oco4
co (D'-cn»- in«cM
ixcMT-iocor-coooen^cooocMio
CM '-»-'- •»• '-
cn^cooo oocoeAincn^ co
01 CM — co inoar-inomoo r-
*™COCMV^
a
o
^
co
CM
d
§
^
CM
«
O
8
^
w
r;
r-
CM
CO
8
oo
Q
3
CO
CO
in
CM
ID
oo
«
(O
^
^
(D
CM
d
a
o
8
0
0
CO
2*
en
CO
^
^J-
8
oo
O
3
in
CO
CO
in
A-13
-------
,_,
< H
I— Z
< O u.
Q U Q
- U
HI 111
-1 O.
Z CO 00
3 Z "
co 5 t**
a
ui 3 n
H u.
M u. CN in
co a a -in
Q. < UI I T-
§Qh CM
CO 1 v* II
Z oo U
•i -no
co x r- *• —
I- O - -H
< 11 W H «
I a -at
Z co ^
a - c. c
r- in M ra o
00 H '0 M
X ii C
U. Z 104'
11 -a *• oi
u. at c co a
o o 3 s.
-J O C (Q
OX i O H
n 0 -H
i- a u m
CO O -H
M U. —
~* C
3
»5
«
c
o
£
X
L
01
CO >
0 0
H U
01
at
0
o Z
W CQ
o
M •?
•H 0
a L
M
01
at
(-_ ^
0 0)
O 0)
u a
,— V
a a
O -v
-i m
a
L.
0)
ID
CO O
o *
o o
o o
oo oo
a a
u u
(0 r-
W CO
(O (O
in in
^
^^
Q
O
CO
CO t—
** z
CO 111
S" O
_i co at
Z co a.
CO _J 0
-i w < in
< CO Z <-
u > <
M _l O
t- < -1 t-
CO Z < CO
•i 5 L) ZO
1- M < *
< -1 Z U
t- < UJ CO U.
CO U Z O
11 u K- m
Z Z UJ in
M UJ Z Z > (9
5n oa 2
i- 5
u o g a a:
co Z u Z ui
3 «l- I-Z
n 11 UJ X UJ O. O
1- t- (/I < t- UJ 2
U < UJ UJ M CO
M a a a. a >-
a i- a. u < at
HZ a z ui
CO UJ Z UJ O 1 >
UJ U O H M X O
a 2 « z H < o
O H UJ < Z UJ
< t)3 z at at
H -i a x
< il 111 < Z < CO
a a ae o uj a
CS O U. w Q. n
at ^
O O n if ii ii ii
U. O
_i o u. 11 a. at
a
UJ
CO
CO
UJ
o
o
u
A-14
-------
a
UJ I/)
J 2
CL 3
Z Q£
5 13
I/I U.
u a
£3
(A
O I/I
O*>
^
u o
i/> a
51,1
H
^
oo M
>- a
u. O
n T
00
u a
ZO
M U.
t-
*•!
-1
g
U
U
a.
i
oo
^
K
CO
§CM in
-in
IU 1 *-
a CM
oo u
-co o
t*» ^ '^
•H
co H n
- a
N- 1 c
^ A O
(9 +>
C V> 8)
3 c.
OCA
P'2
O 4*
u ra
c
n>
g
^
(Q
C
d)
c
1-1
X
L
0)
CT O
fi O
01
at
c°
o *•
1-1 at
c
0
n •M
V O
a L
i/i
Q)
QC
00,
SO)
Q.
*""
X-*,
a w
o \
Q^
^r
01
A 'm(Or-'-t
cntocMcocooor-oocncnt^cnr^cncoocn
^ ^
CM »• «
cn co cn
*• d »•
ox a
co »- in »- (0 * oo
vi^eoiocMcoin^cocnocoinooooiou)
oocoO'-cocnin*inenco'-ocor-o«
«- CM'- '-'-'-CM'-CM
to f oo r** M to
n «CM CM oo (D cn oo oo CM ID
in mm in t r~ CM CD ro in CM en r-
o oo d N o »-'-«- '-cMod
o in o o
»- CM »• »-
ot^ocnonooomiooooncMinoaiin
§*- en to oo CM (0
r- r- r- oo oo r«
ooooooinooo^tnoooocM
go "-
^
^ ^ ^" *" ^^ *~
CMCM*"^^CO^COCOCMfOCMCOCOCOCOT*
IUOU O IU IUIU (J (JO
Z22MI/)ZMZCOZZV)XZl^ZZ
S!2S**CM2cOCMS?COCM^CMCM^
t-U)WeocMen-»ocD«OOcot-en(0(0
4f M M v v (0 0 A O) Is* If) f^ 00 00 00 O) *~
888§§§oo88888S§o§
oaaaaaaaooooooooo
^WtN"OOOTO^CNCO^t1*COO) O CM ^ ifl
u)iAU)tnif)(0(0(0(0(0(D(0(Ot^'tN
CM
8
00
o
00
CM
CO
CO
CO
oo
cn
8
*•
CO
«-
o
at
CM
^
00
o
Q
CM
cn
cn
^
in
CO
8
00
8
^
00
CM
(0
0?
oo
v
en
P.
O
cn
»•
O
*
r-
,_
cn
o
^
CM
O
z
CM
^
CO
i
00
0
CM
CO
CM
(O
0?
A-15
-------
(— v
11
UJ ""
-i
z c/>
^ z
at
UJ D
1- U.
co a
oz
a*
82
IH
(/) X
t- 0
< W
2
a
1** U
u. a
o o
(3 T
Z 0
i- a
in a
j
a
Q
U
HI
a
00
.
**.
CO
a CM in
a -in
O ^
Ul 1 »•
a. CM
i *- it
oo O
-co o
•H
CO II IB
- a
CM •a
- (. C
^ (Q O
„?"
IB +•
TJ -H U
c in 01
3 L.
0 C IB
f-21"
U
00
HH O
(U
Q
o £
M ID
or
c
o
IB •*•
•H O
(B •-
0 t-
U)
01
C£
00,
3a
^^
.— ^
Q 0)
0
L.
0)
(0 M-
a m
3
4^
£ *
o —
L JZ
01 X
a.
4^
s *
u "m
L Z
0)
a.
a
01 0
u
U) C
3 O
V) —
sl
ft] ^
i c
ol
41 O
a
0)
A
§•
m*-*'-cM CM Tt in in
inior-oiDcMioco <-(ocoo
~66*-o*-~~ ~ ^ ~ ~
O O 0
oaaoroaaaaoiocroKo:
«r-OiflOCMOCor~Ooocoocom
cMininp^oooococoinr^^Of^
oooooooooooooor-(Or*ooOCD(D
O'-OOOCMOOOOOVOOOCM
or-ooo'-inooo^oooio
m o co * w o f- *t o *t v
v* ^ ^
COCMCMCO'-COCONCM'-eMCMCONCM
U CJ U U UJ tj U
cn»z2a
-------
~
HjE
Q 8
-V
UJ
_J
00
00 <
u.g
H u.
(fl Q
o z
a3
82
M
00 X
1- O
^ *•*
I a
a
r- ui
£i
u. a:
0 0
w
H a
00 O
M U_
J
i
u
UJ
a.
i
oo
CO
Q CM in
a -in
O'- •
Ul 1 »-
a. M
CO 0 "
-co o
^ *— •—
*
co ii m
- at
CM TJ
- L. C
'-wo
1,2"
A v
•g v 0)
C 00 0)
OCR)
g J
u a
44
,^
44
C
A
3
*••
(0
^
Q)
c
X
L.
0)
t/) >
a
o
c ~
O +*
M (Q
a
Q
A •£
+* O
A ••*
a c.
M
01
a
6^1
U Q.
•^f
a oi
^
~~~,
^
01
n <*.
A *-
a A
^J
^i
C 0)
OI-H
11
Q.
S *
UA
01
&
ft
&3
0
U) —
C D)
h
jj3
s.
g
0)
1
X
o c.
(Q S
O 3
X3 2
(0
-J
£
U
m
o
CO
^
CO
in
an
^
an
O
O
a
i—
S
o
8
M
(/)
P*
(D
CO
CO
ACD870O
(0
CO
(O
in
o
CO
"
m
f-
^
an
in
CO
a
a.
o
6
oo
O
§
CO
I/)
in
t>-
0
ACD87OO
^
^
CO
ID
in
CM
§
1/1
H4
I/I h-
n Z
I/) UJ
>• O
-IV) Of
< n UJ
z co a.
00-1 O
_j n < m
< oo z «-
i J 0
1- < -1 1-
l/l Z < OO
n< U Z O
^ J Z (J
(_ < uj (/, u_
00 O X O
IM UK CO
Z Z u ui
X ^i 00 2C
Q O H* ^
ui a Q o at
vi z ui 2 ui
3>-il- HZ
iiz^sl"
PPtoSSuli
(J < UJ UJ M I/I
M a K a. K >
ot fc a. ^ o < a
HZ a z ui
I/I Ul Z UJ O M >
Sz'Sz'H'SS
O H UJ < Z UJ
< UD z a at
H -10 X
< If OJ < Z < 00
a o a o u o
CM (_> u. « a. w
a -v.
O Q ll II II u ll
u. O
-i o u. M Q. at
o
Ul
(/I
3
00
Ul
o
o
u
A-17
-------
IU I
_i;
5S
I/I U.
UJ I
*-;
in
i x
10
- at
u. a
u. X
O U
a at
2 o
a
u
Z
U. CO
a -
O CN CN
T7-
CO CN
- v- H
t- o
co M n
- a
:ii
II C
ra +^
?-H 0)
0)
^§J
o «
It
n
01
u
v> >
a o
o -H
11 (0
a:
•H O
CO —
a L
4->
VI
01
a
u 01
32
a o>
a n
8
c u
0) ^H
iir
01 2
a.
c ai
01 -
u n
I*
< L
O
(A C
3 O
I/I •-
C 0)
Q) 01
U 0!
a
at
I
n
x
L
O (.
n A
li
Sco (0 ^ co (0 O oo CN to 01 (D co o O) r*- o (D 01 o o) '** *™
r-totDcocor-o>oooo(Dcor-coin*rcocooor-mr^r-
CO 00
* ci
UOKCJMUUUUCJU UUOUOi-iCJUUUOU
(o CN en in in CN
ojiniouicor-cocoiDtocoiQaioicflincnNiDlDUltDCN
^coco^(O(0(0^*t™(OCNincn(0^^"^oooooo(Oin
r- * ,
co CN V
O co
O b
in o
o OCNOOOOOO ooooo<-inooooo
i-a--tt''-'-''-«-'-*-ar'-'-'-'-'-o£CN'- — ••-'-••-
OQ-Oi-iOOOOOOQ.OOOOO>-iOOOOOO
or*-Oo>cnooo0>(DOOcncNinO(oin
oo
Q
f~
ao
O
o
o
r*
CO
a
o
o
r*
oo
o
i
r>
co
8
iO
r>
oo
OP5r-o)ioco
OQQO
OOOO
r-
oo
r»
CO
r^
00
i~
CO
r-
00
a
§
r~
co
a
s
o
t^
oo
Q
r~
co
a
u
« in CN *
r~ co en in
CN r- CN CO
I I I I
Sco co co
a a a
o o u u
MMCNfNCNCNCNCNWCNCHMPICN
CN CN CN CN CN CN CN
ID f- en
!«• r- r»
CN CN CN
ID ID ID
O *-
oo co
CN CN
ID ID
§
CN
00
CN
CO
CO CO CO CO
A-18
-------
~
il
tj \,j
u
J
a
52
i/i <
at
t- u.
V) Q
O Z
aS
o co
oz
(/) X
$2
5°
a
si
M
u. a
o o
i"
t- a
,
01
«
I/) >
o o
M O
01
a
o
c •-
O +•
M n
0£
C
0
n v
•H O
a tl
V)
01
IT
uo,
5nt
a
a a
_
^
0)
fl <+•
^ •*•
^"
a n
B
+>
C 01
!i
Q.
4-*
s u
U (0
L3E
01
a.
« §*
0) O
o
M C
3 O
VI —
C O)
01 01
oa
c
fll ^
ic
ol
Q
Q)
ra
i/)
X
L
0 L
A Ij
^ g
0 3
£
U
03
^a.co^cMr.o^m^^o
CNin^OnCOCMCOCOCOOO
f*r^r"(DlD(D(fl(OlD(Or*-(O
^.
r-
^
CM
•X^
a o ooooo o t-"
3
(0 r- CM r-
t- (oioco^-cMinor^ooco
»- CM CM CO N CO »-
CO 00 (0
O «- in o
in »- » *
o co b b
in
__o_oooooo«
Q.ZOO.OOOOOQ.OI->
coo^r^oinofMOcor^o
0)OoooOooaOOoOOoor-(0
OOO'-OOOCMOOOOO
o mo oo » t o t-
^ ^ ^
COi-PJCMCMCO'-COrOfMCM'-
uj u o o uj u
^^m^^^^ty>jg^g»tyt^
f
*
innr-*^o(Oinr-^coo)O
^" ^" "^ N ^ ^ O4 ^ ^ W
O)oo*-Or*co^ococoino)
CO«(*> r~ - in -
«- oo r- o> ^
r* (O (D Is* r*
^
oo
b
M (J
* * CO
co ^> p. 9 (0
^
oo in
CM CO
CM in in
b b b
o
»- o
fV ^V ft ^ ft
^ U \J ^ V
IH a. a. o a
oo ro o co in
r- »- o » '-
r- oo o (D ID
^
<» O O O CM
» o * «
^
CM CM CO CM CM
O ^^
Z Z 3 00 00
oo (O O t co
CM »- cn co *-
CO » CM ID CM
§6 CM CO »-
8888
a a a a a
«- CM <» (0 00
ooooo
CO CO CO CO CO
(a in
8
oo
a
o
in
CO
(O
in
A-19
-------
^
< K
< 0
Q U
Ul
— 1
Z to
to <
at
UJ 3
t- u.
n
to o
21
to
z
n
to x
H O
I Q
Q
r> UJ
CO 1-
11
_J
O ^
11
i- at
to o
M U.
U.
O
o
*.
1
CO
(yf
V
.
u. n
a -
U M CM
X -W
1 1 i-
00 CM
•»• II
r- t)
• n o
^ i— •»-
* ^
P) il (0
- DC
"fe§
(0 -M
c to o>
o c (0
Q.O 1-
o +•
u
o o
>-i O
01
OK
0
c -
O -H
11 (0
c
o
(0 V
•H O
(0 •-
a L.
VI
Q)
a
00,
32
Q 0)
o -^
n
s_,
3)
.^
(0 <4-
(0 —
a to
§
Is
o •-
L £
01 2
(L
50)
^_
u to
a> z
a.
*l
u
VI C
3 O
u> •-
C O)
Of Q)
01 4J
o i
(U U
f\
Ifl
o
IM
Sample
X
i.
0 C.
(Q 5
(_ E
O 2
Q 2
m
j
_£-
O
(0
00
*- r>
(0 W
(0 (O
U 0
M O
CM n
o o
o o
" O
r- O
^ 00
O O
8 8
C4 CO
to i/>
^ in
ID r~
n o
n *
§ §
Is- r-
§00
o
u u
4 ^
(D r-
CO CO
ID ID
in m
ts
g
to
M
2 5
to uj
-J 10 Of
< 11 UJ
i 2 °-
en -i o
-J n < in
<
M — 1 O
t- < -1 t-
tfl Z < to
«5o zo
1- M < *•
< -1 Z U
1- < UJ tO U.
to U Z O
1 U H CO
Z Z UJ uj
M UJ Z Z >- C9
511 00 Z
H <
Ul Q O Q Of
to Z ui 2 ui
3« K <»-Z
11 M UJ X UJ Q. O
K- 1- tO 3 t- Ul Z
O < Ul Ul H tO
M Q£ Q£ O. DC >
a. i- a. u< a
H- Z Q Z Ul
to ui Z iu O H >
Ul (J O 1- n X O
06 Z w Z f- < O
8H U < Z Ul
3 z a oc
K- -JO X
< u ui < Z < to
O O OC O Ul O
CM u u. n a. n
ac ^
O Q it
-------
1-
Q
Ul VI
li
(X) U.
Ul Q
h- 2
M ^
(/)
O (/)
§x
o
M
I/I Q
^m
< o
Z £
r» Z
00 M
bu O
ii T
06
(9 OC
ZO
M Ik
I/)
1
U.
s
X
00
p*.
Q -
U CM CM
X -CM
T »»
1 1 ^
00 CM
tx O
-CO O
•H
co ii n
at
*" <9 O
,,?M
tQ 4^
?* 92.
o>
3 t.
o c n
Q O b-
Uk W ~
O +•"
u n
c
ra
a
ra
c
t.
V
C
X
(.
u
0 0
at
0
c •-
0 v
w 4
n
It
n
cn
X
c.
0 (.
(Q 2
^ g
0 3
r
u
n
a
o
o
(0
CM
a>
CO
00
CM
b
O
a.
o
8
•^
o
in
CM
CM
111
z
CM
fs.
^
CM
8
00
a
^
in
<*4
(O
o
CO
r*
at
CM
0
CM
in
00
o
b
a
a.
r-
cn
O
o
CM
f\
rp
en
in
CM
^
8
00
Q
in
in
CM
to
Oen*Ooo*«cMcneMO'-onixr»
(O^cotoooocMincflvOcooenixo
(OlOoooor-OTODoocooor-ooini-co
O (0 cn co
r~ o (n in
*- CM i- i-
CM
^
O 1-4 H H l-l
3
CM (o cflintxcococnincMcooo t-
CM'- oo»-(Dtxinoo»oocnrx«cM
o in oo on t ix ««ao o oo
(O CO (0 CM CO « CO CO CM COCO
b b CM b b b bob bb
o o o o
ooQoaaacotxoooacra
o.a.za.i-1-ii-ia.Ma.a.a.i-io.a
OcncnooocncoooencMinotoin
S5fSSS?enSS8^12|2Scoi2
^*
oooomoootinoooocM
OOOO'-OOO'»CMOQIDO'-
inoiDOoo^mooin •»
" "
'-'-'-CO'-COCOCMWOIfOCOCOCO'-
u c_j w u u cj o cj
jgmt/i^mym^p^t^^jgc/tjp^
W'-^N^COWM^'OW^-WW*-
COOOCNO)'-fl'O(D(DOCOt^OJ(DtD
c4^MT*(0(OO)O).r-u)rs'Ooooooo>^
O CO ^™ O ^ ^" CO *™ CS ^ W iQ ^ O ^
888588888800000
QQQQQQQDQQQQQQQ
r^GOAO^cHCo^r^oootoc^^ui
U)iniri(DU)(0(0(0(OtO(0^r>>r4-tVi
C4MCV(SCMMC4C4C4C4C4C4C4CNM
(0(0(OtOU)(0(0(0(0(OCDU}(0(D(0
»-
-------
^
< H
1- Z
^ O
Q U
Ul
j
Q.
Ul 3
I- u.
H4
CO Q
O Z
|*
CO X
1- O
1 3
Q
1** M1
00 (-
u. ce.
o o
01
20
IH
CO O
J
u.
a
u
*•
1
00
.
*
U. CO
Q -
O CM CM
X -CM
I *- •
00 CM
r- O
-co o
(0 T- —
CO ll IB
- IX
CM TJ
-U C
T?M
IB -H
c co 0)
3 «.
o c m
f£*~
U IB
C
IB
3
"«
c
L.
01
c
>*
0)
(/) >
a o
OL
o
0 -H
i-* ni
a
c
o
n -H
v u
s.
a
o
^
on
on
^
on in
* O
o co
a a
a. z
co o
co o
en o
o o
° I
CO •*•
UJ
in co
§00
CO
CO CM
8 8
00 00
8 8
< <
CO *•
CO 00
CM CM
(O IO
CO CO
oo
^
t^
on
,J
«•
(D
CM
00
0
O
a
*
CM
00
O
O
CO
CO
t>
00
CO
8
8
<
(O
00
CM
(0
CO
«
o
(0
en
on
O
g
r-
in
CO
*•
in
CM
*
*
o
CM
8
00
<
00
00
CM
IO
CO
CM
^
(O
o
IO
CM
^
d
a
a.
o
in
00
o
8
CM
U
2
o
10
8
8
<
on
oo
CM
(0
CO
r-
CO
(O
tn
V*
(0
d
o
a
in
oo
o
o
CO
^
2
ID
00
8
00
a
<
en
CM
(O
CO
0 r.
CM CO
(0 (O
w
(0
**
M
(O 00
„• ^
,_
r>
CO
d
o
u. a
t- M
0 N
O eo
O CM
o *•
•*• CO
a to
in r»
" 0
8 8
§ 3
° -" Q M hH M
O
~"
(O^cn (D »- in en CM in into
r^uicM *c^«in^(binoocMen
«- 10 t 10 in CM en
CO^inCMCMIOOO 00 t^O
in o in CM m o v CM ' o r~
O'-OOO'-O" O CMO
o in o o
f CM *- •-
aaaaaooaaaaaao
1-iaQ.ZQ.O.Q.Q.i-ii-iQ.i-cl-a.
cor-ooocoocoinoocMcnrMO
cor-(Or~oooiO(OOcocoo3
-------
^_
< H u.
1- Z Q
< O U
Q U X
UJ 1
-J oo
a.
Z 1/1 r-
5S V
at
iu 3 u. n
H u. a -
M CJCM CM
i/ia x -CM
OZ I «- •
0- < ii*-
§00 CM
I/I -»• II
Z Is- L>
M -mo
in x (0 »- —
I- O - V
< M nun
I a -at
Z w u
O • L> C
r» lu »- nj o
oo t- •Q M
U. Z M *•
11 -g -M oi
u. a c 1/1 n
o o 5 c.
-i g c «
ax 9.0 H
H a u n
«/i o v
M U. —
C
ra
^"
AS
£
Q)
c
M
X
c.
0)
" §
IX
o
0 -M
1-1 n
at
c
0
« -H
v u
nj —
a t-
M
01
a
^^
00,
82
O Ol
>—
L
0)
(Q **-
o IQ
§
+j
C 0)
8Z
I. £
01 X
a
+J
C 01
8«
0)
a
gj 3
O) O
< L
M C
3 O
c ro
41 0)
uae
c
0) ^
i c
u o
0) U
a
M
O
Q)
1
X
L
O (.
(t J3
c. E
?3
2
£
U
fl
0}
•^ r-
S*»
(0
Sin
00
»r ao
v ai
r> ao
10 Is-
b b
CM
•x^
o
O O O
a a -i
M
«~ O l/> H
r-- O - O
-i 1/1 a
< l-l IU
4 ^
O O l/)_l O
_j i-i < in
8 8 S>5
»• - « _| 0
t- < -1 t-
l/l Z < I/)
•H 3 o zo
H n ^ ^
CM n < u z u
H- < IU - (9
5 HI CO Z
1- <
iu o Q a at
I/I Z iu Z iu
3 w K HZ
giz§5SH
r- in tfMtuxujao
.••- »- HHWI
a: H a. o < at
HZ Q Z UJ
(O Is- IO LJ Z UJ O >-" >
« o uiooHi-ixa
co « a z « z i- < o
O O OHiu uj
^ ^ i/i
co n 3
(0 10
t- t- i/i
IU
a
8
in m
A-23
-------
£
4
o
IU CO
-J Z
i
i/i a.
UJ Q
S3
(/)
o vi
a z
o x
u o
n
1/1 a
<£
5
!••• Z
00 11
> a
u- O
n T
00
eg a
zo
M U.
5
-1
u.
8
X
1
00
p."
-
*
O (N CM
X-N
1 1 *»
00 M
- ^ II
Is* o
-co O
(Q «<• .»»
- 4^
* II (0
N (Q O
TJ 11
n C
(0 -H
1"?
0 C 19
IS
u to
•M
ra
o
_
a o
Q
O 4^
M <0
cc
c
o
(0 •£
+•» U
10 -
a L
VI
d)
a
U O)
O Ol
^-
Q U)
Ja
01
,^
(0 <4-
10 —
a m
3
8,u
-
c.r
01 3
Q.
^
C 01
01 i—
U (0
L Z
oio
< L
in c
3 O
VI •-
C 0)
01 01
U ft
I c
u o
s°
a
a
M
01
n
CO
X
L
O L
fl .Q
gl
(0
£
U
n
a
o
o
ID
CO
CM
n
Ol
6
in
CM
ae
n
O
§
O
in
CM
IU
CM
r-
*
i
00
a
- CO CM CM
00 CM 01 - •» 0 ID
•• *• jo (D 01 oi r»
i i 8 I § i i
a a o a a a a
o o u o o o u
S*~ W CO ^
(O CO (D (O
CM CM CM CM M CM CM
(0 CO ID CO ID (0 ID
01
O)
(O
Ol
CM
00
r>
o
a
o
0
^
»
«
«•
CO
z
00
^
(D
in
8
00
o
O
f*
(D
W
(0
w
CM
ID
CO
CM
§
^
s
o
§
01
J^
in
CM
ID
CM
VI
f*i
n
Q
f-
i
00
00
(D
(N
(O
Oi
o
n
t-
CM
Q
O
r*
-
i
CM
«
O
8
CO
s
^
CM
n
00
i
CO
a
u
O)
(Q
M
(D
W
'-enr>r-'-cocoint-in
Oenr-ocooiooi'-'-
coin^oocooor-mr^r-
^
r-.
d
CM CM CM CM CM
ooaaoiioouu
00000
n
ui ^ o ^ ,;
M CM »• ^
c^ CM in
CO » ID 00 CM
O O O O "• -
o
ooaoo5S222
ZZZZZnOOOO
inoiDintintooocsr--
|5co3|5?5eno!SoI
OOOCMOoOOOOO
§(DO'-OooOOOO
in * o *• o
COCOCO'-«-eMCOCOCMCM
ZCOZZZ3Z3COZ
^incor^oococMCDOi^
'-CMCM'-'-'-ro'-'-eM
r-.oi(DiD«t*ineM»»
eooooi»-'-r-oooiinco
§o o o o o o o ^J o
o o o o o o o ^s o
oocooocooooooooocooo
8QQQQOOQQQ
oooouooau
^5 W ^ 1A (D ^" O) O ^ W
^•r^r^r^r^p^r^cooooo
WW«MWCM
-------
^
£z
<0
5i
IU
1
51
H U.
IH
I/) Q
O Z
jj
82
I/I X
H O
cnOT"CM^(poooi
cocncncnocncnOOOOOOO
(O(0(0(0(DCOCOCO(0(0(1)(0(D(0
coco^^^^^^f^^^tninin
T-
^
r-
U
CO
CO
^
O
o
o
in
00
o
o
CO
o
N
fv
CO
8
00
<
o
CO
(0
in
0
r-
CO
U
00
00
O
o
CM
in
CO
o
o
CM
z
1^
O
Q
8
00
8
<
^
CO
co
in
(0
,_
r-
U
O
CO
w
o
0
en
00
00
o
8
CO
z
CO
CM
in
^
8
S
u
<
CO
CO
(O
in
oo
co
in
CM
a
o
(0
w
CM
5
O
8
^>
r^
r-
CM
CO
8
00
Q
<
f
CO
co
in
(0
CO
CD
U
CM
CM
^*
O
o
0
8
^
o
0
CO
z
en
CO
^f
8
co
a
<
in
CO
(O
in
A-25
-------
, — ,
< 1-
1- Z
< o
a u
UJ
J
a
f VI
-l
ra -H
c oo 6)
3 L.
o c
a o
i-i O
0)
at
0
c •-
o •*
at
c
o
10 +>
•H O
^ •••
a (.
VI
0)
IX
00,
82
Q 0)
o ^
-1 O)
a
^
01
n> *•
10 ^
O ID
a
+^
C 01
St:
"r
at as
Q.
4^
S5
O (9
(. Z
01
Q.
0,1
< t.
V) C
3 O
VI —
C 0,
i c
o 3
5 o
oL
i^
a
Sample
X
t.
O L
** S
ra j3
t. E
Q 3
J3 Z
10
i
£
U
*>
A
CD
_
(O
(O
o
o
CN
0
o
r-
0
8
^
(/)
£
S
CO
8
r-
oo
Q
U
ID
CO
(O
in
r~
^
(0
u
4_
CO
CM
0
o
o
0
oo
0
8
n
i/i
in
o
CDS 700
^
^.
»-
CO
(O
in
CM
Q
o
1/1
CO t-
>-i Z
00 UJ
>• o
_i uo at
< l-l UJ
2! ^
CO -I O
o> 1 "
w _l 0
1- < -I 1-
00 Z < CO
«5o zo
1— i-i < V
< _IZ U
1- < 111 00 U.
co u x o
IH (J h- CO
Z Z UJ w
IH U Z Z >- (9
51-1 CO Z
1- <
u o oo at
i/l Z UJ Z UJ
3 lit- t- Z
§<<<-!
ooz SL
H4 1-4 111 ^ UJ & O
t- t-
a: t- Q. o < a
t- z a z in
00 01 Z IU O i-l >
UJ 001- •- X 0
a z w z i- < o
O t- UJ < Z UJ
< OD Z flC OS
i- -i 3 x
< n ill < Z < I/I
Q O 0£ O HI C3f
CMOU. M a«
at ^
o a n ii i
U. O
_i o u. <-• Q. at
a
UJ
I/I
3
CO
u
a
8
A-26
-------
o
UJ I/I
O (/>
S5
8S
v> o
< Q
o-S
ss
IH U.
I/J
O
CJ
X
u. ro
a -
CJ CM CM
X -CM
en CM
- V- II
oo CJ
-co o
CO II (Q
- at
™lc
-as
ii C
I/)
c «
.o t-
O (0
£
01
X
01
a o
ce
o
o **
•M O
a I
+<
M
0)
ce
u ai
82
a oi
3S
L
ra 4-
o 15
§5
a.
*/
§2
£*
a.
at
< i.
So.
0) 01
U Oi
I;
O C
a
03
lOr-ioiDajcor-cncocococnr^couitcococof-inr-t--
8
CMCMCMCMCMCMCMCMCMCMCMeMeMeMCMCMeMCMCMCMCM
•HoroaaaaaaaoaoaoooooaaaQ
ooooooooooooooooooooo
_i_i_i_i_i_j_i_i_i_i_i_i_i_i_i_i_i_i_i_i_i
en
(0
b
O) O) f1^ 01 (O^CO '"CO (D ^ CO
bbbbbbbbbbbbbbbb--'-»-cMO»-
in
£§§§§§§§§§§§§§§§§§§§§§§
Or-OcnenOooenioOOcneMinouJin^-uiioeofMc-
opooooinooo«inoooocMOcooooo
to o o o o o ^m o o o ^r w o o (0 o ^ o oo o o o o
CM CM »•»•'-
uiui
»- »- CM CO CO CM CM
U CJ
» Z S c/) Z
ID CM ID en
»- CO *- *-
CM
§
r»
oo
in
(M
ID
in
CM
1
in
in
CM
ID
88888
CM
03
r- r-
2
(M CM
ID (0
I-
00
o
CM
ID
f-
00
a
f»
s
uo
<<
CM
ID
CM
03
8
r-
§
S
CM
CO
8
r-
s
u
CM
(0
8
r»
oo
CM
ID
888
a
o
CM
(0
CM
(0
CM
ID
p-
00
o
u
CM
ID
r*
00
a
CM
(O
8
r-
oo
o
CM
(0
r-
oo
a
u
ID
CM
ID
«t in cs «r *r
> oo en in eo
CM *- CM eo *
88888
r- r- r- r- r»
a a a a o
u u u u u
CM
ID
CM
(0
O -
oo to
CM CM
ID ID
CM
00
CM
ID
CM CM CM CM CM CM CM
CO CO CO CO
A-27
-------
< I-
< o
O (J
HI
li
10 <
at
UJ 3
I- U.
HI
I/I O
o z
en x
t-o
< HI
* =
fs. U
00 t-
£2
HI
u. a
o o
25
I/I O
er
X
V
to >
a o
~S
0
C -
O *•
w 18
o
18 -H
S.S
O L
+rf
V)
01
K
O 0)
C v.
O CTI
o a
O 0)
S2
<8 *.
C 01
u ^
y
a.
S2
U (8
0)
OI
(/> C
3 O
I/I •-
c a»
0) 0)
O K
-*i
U 0
01 U
%
X
O L
01
N
QOOQi-i 00000000000000000
oooo ooooooooooooooooo
-J -I -I _l _I_I_I_I_I_I_J_I_J_I_I_I_J_I_I_I_1
N
m CD
o> o
b <-
M » in en to (0 i^ en *- n o> m en co r- —
tt CD r- 00 PJ(MMO)'-00»-«O)OOOOCM'-»-COOOOOM
O ("J »- O OOOOOO'-OOO'-O'OOOOCMO
in
iiii2eilgiiiiii;iii§§i§§
cnooocoooooocoOoof>(0(^eoo(OCDOoocooo«O
OOO'-OOOMOOOOOtOOOCNOOOOOO
dooi^-ooO'-inoooi-oobcDoooooo
O mo oo »* or~»o« *e>< oo
*- *- CM »-
•>- N »- *-
M (M *• »- «-
88888888888888
r~r»t~t^r-r-r-f-i>t^r-r~i»-r^
aaaoaaoaaaQaaa
888888888
oo
3
oo
10
oo oo
10 to
oo
co
O) O)
CD CD
a>
co
a
co
at at
(O (0
IO
O O
(O CD
r»
oo
o
o
co
t^ t^
00 CO
o a
CO
o
(0
r»
00
a
o
(0
t^-
00
a
t^
CO
a
r»
oo
a
r»
oo
a
r»
co
O
CDCO
CO
CO CO
A- 28
-------
^
HZ Q
§8 |
U I
-1 00
5z *"•
wi 3 ^
a
iu 3 u. n
t- u- a -
w U CM CM
1/1 a x -CM
oz z~ •
Q. < ii*-
§O> M
(/I - V* II
z a u
M -co o
I/I X f- «- —
HO - *>
< M CO II (0
5 a - at
CM 'O
t- iu "-mo
oo H ^ w
>< M C
u. Z « +•
u. S e JJ o)
o o 3 c.
W Z ^L O ^
2° 5 v
HOC u n
I/I O -H
M ^ -^
C
(Q
Qf
^
m
c
01
c
X
£.
fl)
V) >
a o
OL
o
c •-
O -H
I-H (Q
a:
c
o
<0 -P
(Q —
Q L.
U)
a
00)
O W
u a
>— n
a 01
o ^
-1 O)
a
t.
0)
(Q *^
•V •••
O "ffl
^
C 01
01 -H
o -
(. £
01 2
a.
4^
C Q)
Q) ^
0 «
01
a.
en <5
4^ L*
O
VI C
3 O
5 a!
c
i c
U 0
2^
W
O
HI
V
1
1/1
X
L
O i.
(0 fi
l- E
Q 2
(Q
j
£
U
00
*—
s
CM
a
o
in
oo
6
§
^
r*.
*
0
8
CM
tfl
^
(0
CO
CO
8
r*
00
U
^
(0
CO
(0
in
f«.
10
CM
Q
0
CM
O
^
Q
O
d
00
o
8
CO
I/I
in
*
8
r-
00
a
y
^
f*
T-
n
(0
in
CM
x^
a
o
(/)
^H
^ g
I/) IU
_i in at
^f ^^ ^j
Z 1/1 a.
I/I -1 O
-i n < in
o > * *~
M _l O
H < -1 H
< -1 Z U
H < IU I/I U.
MUZ 0
M U H CO
2 Z IU IU
M IU Z Z >• O
5n CO 2
H <
iu Q o o at
1/1 Z iu z iu
3 w H HZ
O < < w
Z Z H Z w tt
O O Z OS < H
11 M iu x ui a o
H H I/I < H IU Z
U < IU IU 11 I/I
•i a at a oc >
ana. u < at
HZ a z iu
I/I u Z iu O M >
IU U O H n X O
a z n z H < u
O H U < Z IU
< o => z
-------
t-
o
UJ I/)
si 3
51
or
u. o
o5
(9 O£
Z 0
*•* U.
V)
1-4
-1
a a
o a
OO
X X
XX
1 1
co co
r»~|x~
(0 (0
CO CO
- -
CM CM CM
- -CM
•x. 1 f
00 CM
- *- n
r- O
-co o
CO II (Q
- at
CM TJ
- c- c
^ (Q O
H?M
«*•
TJ *• M
C I/I 0)
0 C Ifl
ft O 1—
g-21-
o -M
o ra
**
£
c
At
a
^_
(Q
C
C.
c
X
L
10 >
0 0
"1
O
0 -H
w CB
o:
c
o
ra -H
•H 0
a L.
w
u
a
. *-*
U 0)
8D)
a
am
S^i)
r»
^^
L.
0)
.^
10 4-
O (Q
3
^
S2
u •-
L £
4) X
C U
u ra
<-*.
CD
a
fi. 3
5)5
< L
u
vi c
3 O
C 5)
OK
Cd -H
3 c
•^ 3
|8
a
0)
0-
1
X
L
O C.
v 0)
(0 j]
L E
Si
a Z
(0
£
U
ra
a
o
oo
(O
co
(0
V-
cn
(0
o
etc
o
§
o
in
CM
M
UJ
Z
CM
r-
»
CM
§
co
3
in
CM
to
*
*
UJ
_l
1
I/I
01
O
O
(14
£
O)
in
CM
^
8
CO
a
in
in
CM
(0
O ^ ^ (O U) co *~ LO 01 oo (o (0 U) ^ r** r^ w co
co w co o) Is" oo o r* (0 ^ w ^ r^ co ^™ co to O
to (0 oo r** (o co oo oo (0 r*** r^ r*- co (D P*" oo r*™ co
O) (O
*• r^
*~ ^
M
IXCMV ID oen ix cocoinr-
^-Oin^incMOcococMvcocM~-oicn(oin
1- r- ^ *- ^ r- »-
CM'- p- in n*- in t co
fo oicom'-oi»'-co(D coor-CMCO
CM * tcocncnocovcnr- to r- 01 •- 10
'-o oo'-O'-O'-oo o b -o -•«-'•
0 O
»- i^
ooaoaoooooooaoooao
oo>enococn(ooocncMino(Din«in(o
S5fgSS5SS§8?:SI5S3!S5So§
oooomoootinoooocMOcoo
OOOO"-OOO*CMOO(OO*-OooO
inoioooovioooin t o >-
*™ T" ^ ^" ^ V* T~
^«~*— (O^COCOC^CONCOCOCOCO^^CNCO
O O UJ UJ UJ L> O O O UJ
O^^ooc^coocow^fr^uicor-cotof^
M*-^-C4'-COWCM»-CON*-«CN'-'-'-CO
«co«oi-*otO(oo«^cn<0(D'»*in
cM^^cptooienfxm^cocococn*— ^-r^co
O C*) ^ O *^ ^ CO ^ CM ^ CM O ^ O ^ O CM ^*
QOOOOOOOOOOOOOOOOO
oooooooooooooooooo
COCOCOCOCOCOCOGOCOCOCOOOCOCOCOCOCOCO
aoaoaaaaaoaooaaoaa
oooooooooooooooooo
CMCMCMCMCMCMCMCMCMCMCMCMCMCMCMCMCMCM
CM
CO
(0
CO
"
0
CO
a
co
CO
en
O
O
CO
3
(0
CM
en
CM
8
§
o
0
CO
CM
(0
CO
co
(^
^
r-
0
^.
*
V-
a
CM
t
CO
o
o
CM
*
en
«
in
CO
8
co
8
CO
CM
(O
CO
CO
CM
r-
r-
P.
CO
co
b
a
p-
^
01
o
§
^
CM
Z
^
CM
^
CO
^
8
r*
CO
8
CM
CO
CM
(0
CO
A-30
-------
^^
< H
K- Z
< 0
Qi\
\j
UJ
-1
X (/)
5 z
i/i <
a
U 3
1- U.
too
o z
§2
H O
$ °
a
f*. m
525
U. Z
u. £
o o
«jj5
t- a
M O
M U.
J
a a
o a
(JO
z z
i i
oo oo
i- r-
(D (O
CO CO
M CM CM
- -CM
*- »• •
00 CM
- *- II
r- t)
-co o
*<
co u n
- at
" ? c
« -H
TJ ** Ot
CMS)
3 L.
OCA)
l^1"
u ra
c
(Q
i
1
01
c
l-l
X
c.
01
o o
a:
o
o •£
H «
at
« v
+• O
ra —
a t-
i/i
ac
*— v
u ai
ft n\
8 a
*-\
O O)
-J O)
a
L.
01
ID -OOO
in«r.»0(Dinr.*oo»ooo
in
in
(O
CM
8
oo
a
<
cn
O
CO
(0
in
'-
(O
r-
^
[~.
cn
(0
^
a
o
in
00
o
o
CO
M
0
CM
CO
8
00
a
<
o
CO
(0
in
*"
^.
!*•
»
^
r-
T~
^
00
6
o
CM
in
00
o
o
CM
*
r.
CM
o
in
o
8
CO
a
u
<
^
r*
CO
(0
in
CO
in
CO
CO
*~
ID
in
v-
o
cn
oo
oo
O
8
CO
i
00
CM
in
i
CO
a
u
<
CO
T-
co
10
in
oo
00
(O
to
r-
CM
CM
(O
^
o
^
CM
*
O
8
-------
^
£!E §§
< o u u
UJ ^ II
_l co oo
Z (/) P« r-
co <
0 0
~!
0
0 V
11
O >,
-1 0)
a
t.
01
4^ -~
(0-
a ra
§
C 0)
01 -M
u •-
L £
0) X
a.
4^
S -
u n
c. z
01
a.
0) O
3 §
VI —
C O)
U 01
HI ^
ic
fQ
y
a
0)
I
ra
X
L.
0 (.
•H 01
m ja
^ g
0 3
Q Z
n
£
U
n
CD
^
00
(0
in
01
(0
(0
(O
^
CT
a
^_
f*.
If
o
8
CM
CO
£
(0
CO
§
co
8
(0
CO
(0
in
U)
t-
r-
in
CM
CO
^
^>
0
a.
o
6
00
0
8
CO
CO
in
O
i
oo
^f
^
t—
CO
(0
in
CN
o
a
to
n
I/) h-
i Z
5
M -1 0
H < -I h-
CO Z < I/)
M O
X H 03 Z
a u i- <
uj Q g a at
to Z u Z uj
DMH K-Z
O O Z S < H
11 n uj x ui a. o
I- H co < f- ui 2
O < UJ UJ (1 l/l
M at at a. a >
st H a o < a
HZ a z u
CO iu Z iu O n >
U UOH M X O
a z n z i- < o
8H UJ < Z IU
3 z ae a
H -JO X
< n UJ < Z < I/I
a o at oiu o
CM u u. M a. M
a \
O Q II II M II II
U. O
_i o u. t-i a. at
a
UI
CO
2
CO
UI
a
a
u
A-32
-------
a
a
u
4C V
2 I
Q oo
UJ I/I (•»"
CL< 10
5 S5 Q m"
i/) u. a -
O CM CM
H- Z I *- •
M 3 1 1 *"
c/> en CM
2^ - ^ n
Z oo CJ
Z 11 -co o
° 2 co" u «
MO -a
K CM 'O
Ig ;|S
00 M B *• 01
x a CMS)
u. O 3 L.
••J 0 C (Q
o **
So °2
K •£
(/) C
11 «
-1 3
_
(Q
£
4^
C
01
a o
a
o
c —
o +-
1-1 ra
a
c
o
(9 •£
•M O
a C
(A
a
^^.^
00.
o 01
u a.
O O)
3 w
a
*—
L
01
ra H-
(S >-
0|
^
C 01
U^
*•
In £
01 3
+J
C 01
u n
J. Z
0)
013
0) O
< L.
C9
M C
3 O
01 0)
u oc.
c^
1 §
U O
So
a
o
«,
X
t.
0 «-
•P 0)
"f2Scol2?£§
ooooooinooo«inoooocMOooo
UIOOOOO'-QOOfCMOOUJO'-OooO
CM inocoooo«coooin * o •-•
V- *• ^ 1- W" ^ ^
CMCM^**^CO^COCOCMCOCMCOCOCOCO^^'CMCO
lUOO O ul UJUJ O CJCJCJ UJ
Z2Z(/)(S>2ZZZ3Z
CMeno^^ocorMcoooocM^c-incor-ooiocM
'-»-CM»-'-CM»-COCMCM'-COCM*-eMCM»-'-»-CO
^cMCMTvvw(ococno)r**(nrN>oooooocn^*^r*>oo
§0000 ooooooooOoooooo
O O O O O O O ^^ O ^ ^^ o o O O O o Q ^
SOOQOCOOOOOOOOOOOOOGOOOOOCOOOOOOOCOOOOO
OOQQQQOOOOOOOOOOOOO
UOOOUOOOUOOOUOUOOOCJO
^in^oocno^cMCO^^cocnOcM^inco^en
ininininmcocDtocococDCDco^f^f^r^^F^r**
CMCMCMCMCMCMCMCMCMCMCMCMCMCMCMCMCMCMCMCM
COCDCOCO(OCO(0(OCO(0(0(0(0(0(OCDCD(OCOCO
CM
00
-------
^
< 1-
1- Z
< o
_,
UJ
_l
X I/I
2i
at
IU 3
h- U.
1-1
Q
?!
(L <
jp
5 vi
u z
M
(/I X
<2
i S
t** UJ
521;
U. Z
M
00=
u
fit T
z5
H a
in o
j
Q
Q
X
I
1
oo
[V,"
(0
a m
o -
OWN
5*C * W
?7-
O) (M
00 U
• m o
m H a
at
^ (Q 0
u !~
fl V
u +^ 0)
C co 8)
o c «
O v
u «
V
*•
c
IB
d
n>
c
^
di
•*•*
c
IH
X
L.
tt\ %
2§
o/
o
O -H
M at
c
0
ID •«->
•M O
(Q •^*
a L.
M
U O)
0 Ol
u ex
^.^
Q O)
3S
c.
(U
(Q t*>
-4-* —
Q (0
g
SO)
^
£r
tl 3
a
is
tu ^«
om
01
a.
a
01 3
OIO
< t.
u
3 O
«| —
C O)
01 0)
ua
c
OJ ••-'
M 2
I1
Q
jfl
ns
(O
X
i.
0 L.
Si
JJ Z
ra
JZ
u
n
oa
«r«mr-.-" Q. >-" Q. >-"
0
J
co in r* in o i/)
r- O'-O)O>c<<*>CiNO«O)'-«oo»-Moa'-eoc<5Oi
•*• M^ ^ ^^.^^ ,-,- ^ ^^^ ^
oo »- r^ in in r-
co (0 in in in in f t M«^-n 01 co M
Ooan^-(ooo (OOMcnooNcnooto ^-oo- r-
o o in o o
ooo'ooooiooooo'oo'O'ot/iaiao'c/iaa
cootr-omoiNOor-OconocoinoocNoicNO
WOCOOOOOOOOOOOOOt^COr-OOOCDlOOOOOOOOTO
^~ *• T* ^ ^ ^
OOO'-OOOCMOOOOOtOOOMOOOOOO
OOOr^OOO*-lflOOO«OOO(OOOOOOO
O ino oo»* or-*o* *•« oo
^ ^ ^ ^ ^ ^
C*)^(V)C^CNCv)^B"OC004Cs4v"CSCs4COCNtN^~*COC4fr)^CO
uj uj o uujujoo cj o y
^^m'K^^jpm^^gtjn^^^y^/ifjflyi/t^^f/i^
*
wor»*otf)tnr»-5rcoo>Ooo(OO^c5iflOr^oor^(OCO(0(D(D(D(0(0(D(0(0(0(0(DCO(D(OCO
• *»««ininininmiOLnin
A-34
-------
^^
< H
< 0
a u
!j
Z I/I
< ?
i/i <
a
UJ3
^ li
r^ u*
i/> a
o z
a*
O I/I
uz
(/} X
1- O
IQ
a
00 H
U. Z
u. a:
o o
n
vt o
_i
§
u
00
pj
B
g n
O CM" CM
X -CM
X «- •
1 1 • U
-i i/> a
^ *•* in
i 52 "•
°. °. ^23 S
§O < I/IZ «-
o o>S
»- «- w _l O
1- < -J H
I/I Z < I/I
>i (9
X i-i 03 Z
1/1 i/i a u i- 3
IU O Q O £t
i/l Z ui Z UJ
D 1-1 1- HZ
0 « «
§z H z M a
o z a < H
P- U) i-l l-l UJ X UJ Q. Q
»- «- HHI/>
ana. u < a
HZ a z iu
ID P- (/) UJ Z UJ O 1-1 >
co o iu y o H M x o
m t a 2 1-1 z H < o
O O OHUI
o a a o a o uj a
u u cs o u. w a M
O Q II II II II II
u. O
_i u u. M a. a:
a
(OP- IU
f «- W
r> ro 3
10 (0
»- «- i/i
IU
a
o
u
in in
A-35
-------
S
Q
UJ I/I
j Z
& <
ig
M U.
UJ Q
I/I
Ii
So
tn a
i-
< a
Z K
r- Z
00 M
>• at
u. o
U. I
o u
O or
Z O
^ Urn
H-
in
J
L_
a
o
a.
X
i
00
r-
ID
^
CO
.
CM
•
H
1
3
0
§
°
a
u
i
oo
r>
ffl"
•
^
CO
CM CM
-O
»—
CM
O "
co o
4*
ii a
at
•o
L. C
?O
w
HI
(0 4-«
** SL
M a
i.
OH
n
+3
c
m
a
^
2
^
0)
c
(/)
a
H4
g
2
a
g
Q
U
a
o
2
44
Of
u
Q}
a
4*
c
u
0)
a.
0)
01
<
tfl
I/I
tJ
c
f"
>
L
O
•H
n
L.
o
Jd
n
X
V
0
1
+•
nj
a
c
0
0
C
VI
0)
a
*-s
O)
-^
O)
X..
01
a
t-
.2
»*.
(Q
^
0)
^
^fc
V
IS
2
Q.
3
8
L
C
O
0)
0)
i
u
a
M
0)
M
i-
A)
(.
ft
g
3
U
n
a
CM
(O
(O
r-
00
0)
O)
o
a
o
8
o
in
CM
CM
UJ
Z
CM
r-
CM
8
00
Q
^
in
CM
(0
Ooir-,-c»cococMtn~ocor-*-ino*
SSISSSPgJSrtS^SPSSSIC?
oncMO coco m cotoo
*^-r» in oi o in CM uj
Or;'- 2" N " " UJ
at
co(or.cooocncoo^«r.ininoo(ocn««
oop-inenr-inooTfo>r«cMcoeM*cM*ooin
»- CM - COCM
b bbbb ^ .^ . ^~
oinoo oo o ooo
atatatataatataaoOKaxxxaa
MMMHia.Htl-ia.O.O.O.HlO.HIHIHia.Q.
r-OencnooocncDOOcncMinoioin^'in
S«*SeOUJ«S§8^*r?Scor??£
^
ooooomoootinooooMOco
ino(oooo«(oooin «t o •-
'"*"*"'" *" "~ *"
CM^^^CO^COCOCMCOCMCOCOCOCO^"^CM
ZZWM^COZ«flZZM»^WZZ^3
r-CM'-'-eM'-COCMCM'-COCM'-CMM'-'-'-
incoootM«-*0(0(oocor-o>(0(0'*»
CNN^*^(p(00)O)is>u)is>oocoooo)^^r^
^ O C*) ^ O ^ ^* O ^ N ^ W O ^ O ^ ^5 W
§o o o o o o o o o o o ^f ^? ® ® ^^ ®
^^ O O ^5 O O ^S Q O ^J O Q Q Q Q ^J Q
aaoaaaaoaaQooooaoa
ouuooooooooouoouou
SvcMCO^c^cocnOc^^inwr*-
CDG3(0(OCO(O(OP^f^r*f^ir"f*"
CMCMCMCMWCMCMCMCMCMCMCMCMCMCMCMCMCM
-
CM
r-
CM
b
HH
ao
r-
O
a
10
0
en
O
0
CO
UJ
CM
CO
in
eo
»•
8
oo
a
<
on
r-
CM
ID
CO
«
en
r-
in
^
(O
^m
0
oo
CO
en
O
o
CO
s
ID
CM
cn
CM
8
oo
a
<
o
oo
CM
ID
CO
r-
10
r-
ID
r-
O
**
00
ID
O
CM
00
0
0
CM
I/I
en
V—
^.
in
CO
o
o
ao
0
u
^
00
M
ID
CO
10
^
r-
00
r-
in
CM
^f
a
r-
a>
O
8
CM
0
CM
^
CO
qf
8
ao
o
u
CM
00
CM
ID
CO
A-36
-------
<£
Q U
-I
2 (/)
< ?
I/) <
0£
u a
H- U.
M
cn a
21
8
< °
x a
a
j-i
U. K
OO
(3 T
5o
t- ox
(/I O
u
o
u
ft
00
f-
to
a "*
o co"
X CM CM
i -O
00 »- .
P- CM
580
CO II (Q
- a
**-«! c
"•55
n C
A -F
•g +* u
C c/) 6)
OCA
f^1"
u n
£
c
A
°
n
c
t.
01
c
l-l
01
a o
a
0
c •-
0 -H
It (9
a
c
o
A -H
+• U
A •-
Q L.
V)
01
a
. ,— V
g o»
82
Q O)
O-v
3
01
A <*•
ii
U TO
3
c «
01 S
a.
+«
C 0)
u "A
01
a.
a
8L3
0»O
< c
I/I C
3 O
I/I ...
C B)
U Q)
u a
fQ
o
a
01
i/i
>•
O 1.
•H 01
A 3
O 3
A
£
U
m
Ucor-wcooo^cococN^o
lDttcn(Ncnr-r-O»-cnco
r-r-r-(0oooor-inc~r»r-io
cn in (o
O ID V
CM O CM
CM CM
>^ X^
i-i o a M M
0 O
o m co -- in in 10 * »-co
* o r- t ID o co ao (oao
CO (0 CO *
inr-r<-cfl ^-(oen ID
•^ooioat cnoor-cocM
•^bbb bb^^^
o o o
aacraaaaaoaoioi
coo^r-omOcMOcor-o
cnOoooooooQOooOoor^co
OOO*-OOOCMCOOOO
08°w8°S5lS08°
^ ^> ^.
«»-Cl>WCMn'-COCON04'-
U U O U W U
3ZVISZ22MZZMZ
^ ^ *"" CH ** *~ CN ^ ^ CS
coino>
oo9ooi^iDr>>kOrooin^*f
nwww^-ocM'-'-'-co*"
888888888888
oooococooooooocooooooooo
QQDQQQQQQQQQ
*
o^(Oooo)^™co^u)r*oiO
OOOOCOOOOOO)O)O)CnO)O)O
C4C4MC4MC4C4C4MC4CICO
COCOCOOfOCO^^^^^^
o in
o
CM
CO
8
00
a
o
CO
(O
in
oo
cn
(0
CM
^f
^
hH
00
in
0
a
CM
in
00
o
o
CM
^\
2
r.
Ifl
00
*-
CO
(0
in
^
ID
r-
^.
CM
CM
i^
^
ID
CO
o
a
en
oo
00
0
8
CO
u
z
oo
CM
in
i
ao
a
CO
CO
10
in
ID
CM
(D
00
(0
in
00
b
a
°"
CM
;
o
8
*~
VI
^
r-
CM
ro
O
O
00
o
^
CO
(O
in
,-
ca
P~
^
^
u>
i^
CM
CM
^
O
O
8
o
o
CO
^
2
cn
CO
^
8
00
o
in
CO
(0
in
A-37
-------
< H
rf O
0 U
ui
i
a.
3 Z
I/) ^
a
UJ 3
H U.
tH
i/> a
o z
aS
82
IH
(/) X
H O
x a
o
r- uj
CO H
> <
u. Z
I-H
°3
z o
H at
(/I O
M U.
-1
a
u
^
t
CO
•
.
(0
u. «•
a -
o o
Q.
I W N
CO »- .
r» M
IB (J
-n o
^ f -^
M II (0
- OC
o
a I
in
01
at
. ~*.
U O)
C "^
o m
o a
"^
,-^
O 0)
^^
L.
01
(0 4-
+< —
fl —
o a
§
^
C 0)
01 -H
o —
a
C 0)
01 —
u ra
(. X
0)
a
ft*
0)0
4 ^
(9
vi c
3 O
if
E C
0 §
v u
a
0
0)
>N
U
o s.
V 0)
(Q £1
U e
El
(0
J
c.
u
4->
«
00
M
CO
(O
co
(^
*~
M
en
CO
O
O
a.
^
f^
^
o
8
04
00
(s.
^
870O336
Q
%
IB
m
IB
in
^-
in
CO
r-
9)
M
1-1
en
co
m
o
K
1-1
o
6
a
o
8
m
00
in
8700407
Q
^
^
m
IB
in
N
a
o
o
_i oo at
< H U
5 2 a
*3« ?
M -1 O
1- < -1 H
oo z <
w < 0 Z 0
5 -.5 S*
H < UJ I/I U.
00 O X O
M (J t- M
Z Z UJ ui
M UJ Z Z>- C3
5M oo Z
H <
uj a o a at
00 Z UJ Z UJ
a M i- HZ
Q < < 5
zZHZwa:
o o z a: < H
I-H *•* UJ ^ UJ Q. Q
H H I/I < H UJ Z
O < U UJ M t/J
1-1 at at a. a >
oe H a. u < K
HZ a xui
I/I UJ Z UJ O w >
UJ O O >- >- X O
a z »-c z H < o
O H UJ < Z IU
< u =) z a a
H -I 3 X
< II UJ < Z < (/)
a o a: o uj o
-------
t-
o
u in
o!z
x a
(/) U.
i/i
25
§x
o
l-l
i/i a
H-
• K
U. O
o5
zS
l-l U.
1-
M
J
Q
U
2^
i
00
r-
(0
u. »
tJ CO
I CM CM
1 -O
en •*• ^
to CM
• v u
t- O
• co o
+rf
CO II IS
- CL
^" (Q O
?M
(Q 4*
TJ 4J U
C (/) 0)
3 (•
OCX
o +•
o «
s
3
i
Q)
c
M
0)
00
" u
w
a:
o
c —
0 V
M n
a
c
0
ra v
•M U
« —
Q t.
VI
0)
at
. ~*
y n
c ^
82
0 O)
o ^
^r
t.
0)
(0 C
O *(«
C fl)
gu
tS
Q.
^
C 0)
8 «
s
0) O
II
U 0£
C
I1
o
1-4
0)
(0
I/I
o t-
•f $
ra 5
L E
?l
^^
£
U
10
CO
CM
S
CM
"S^
o
o
••J
U)
*-
§
o
8
o
in
CM
CM
z
CM
r-
^
CM
§
CO
a
S
^
in
(0
Ocn^-tncoocM
SSSSSPSIS
CM CM CM CM CM CM CM
**•* ^""^ ^^ "*x '^«, ^s ^s,
ozoooaaoa
o o o o o o o
r« r- *- in
O CM r> en n in in
0-0-00--
OOOOQOQQ
r-OencnOeoeniD
^tncMCMinniocn
doooonoo
°°°S8588
" "
riT-^T-n^nm
Z Z « „ Z « S „
*
2S^?82coS
inc*)ooc4Oi^vo
CSN^"^tptDO)O)
^ O CO ^ O ^ ^ CO
8S888888
oooooooooooooooo
aaaaaaaa
inr^oooiO^c^co
lfltfltOU)(0(0(0(0
C4MC4CMC4CNC4C4
v^on^ + ^uo*-
SS?Sf:5SS"??!
CMCMCMCMCMCM CMCMCMCM
"\ *^ \. ^^ >s«» ^^ *vXi >sx ^X X^
oaooao aoaa
oooooo oooo
r-
00
00
o
CM r- r- co to
coen co i^to^cocoin
00"0'-00'----W
ooooooacoooo
OOcncMinoeointinu)
OOi-en
-------
< H
< O
Q U
a.
Z (/)
1/1 <
t- u.
M
CO Q
O Z
^
oio
u z
to x
1- O
^ M
*=
fx tlj
Ei
u. ae
o o
f9 T
So
»H
h- Of
co o
IH U.
0
u
OL
00
*"•.
10
U. f
o co
Z CM CM
1 -O
O> »- •
CO CM
• *• II
r- O
-co o
co u re
CM'TJ
v* A ft
„?"
ra +*
TJ *• 01
C CO 0)
3 t.
o c ra
o +*
u ra
£
c
ra
3
,_
n
(.
0)
c
>.
L.
0)
a o
~£
_ o
c •»-
O -H
*H (Q
a
c
o
ra +*
** o
ra ••-
o c.
VI
01
a:
. *-*
U 0)
82
a 01
O v.
-i ai
a
t.
01
ra it-
ra >—
Q n
o
C 0)
L £
01 X
a.
44
S "
ura
0)
a.
§ii
< L
U
M C
3 O
01 —
01 01
U K
C
aj +>
g ^
/* X
U O
rt) Q
a
o
M
0)
M
§•
ra
1/1
i_
O L
•H 01
ra j3
(. E
O 3
£
U
ra
oa
to
to
^
CM
a
o
i
10
Q
Z
CO
0>
o
o
CO
X
in
en
s
8
CO
8
<
CO
00
CM
to
CO
00
v
r"
CM
a
o
s
^
i
o
8
o
8
UJ
Z
CO
00
CO
CM
8
8
<
00
CM
10
CO
f-
q.
^
CM
a
o
(O
en
6
§
„
CM
00
O
o
CO
1/1
P«
00
CO
8
CO
0
<
ID
00
CM
10
CO
in
en
(0
«
a
i
CM
*"
a
r-
in
00
••-
in
CM
*
«
o
CM
8
co
O
<
00
CO
ID
CO
CO
CM
oo
(O
in
CO
M
CO
oa
o
at
o
in
oo
o
8
CM
UJ
Z
O
CM
10
8
CO
<
en
00
CM
ID
CO
CO
ai
co
*
-
CO
*"
at
t-
U)
r-
00
O
o
M
z
10
09
r-
8
00
O
<
en
CM
(O
CO
'-cococM^ooin(ocoin»r-aj«»0co
-------
< K
Q CJ
UJ
J
Q.
I/) <
DC
UJ 3
K U.
IH
I/I Q
0 Z
2
o 1/1
u z
IH
(A X
52
5°
Q
p*. m
25
u. Z
U. Of.
o o
i5
IH
t- a
I/) O
M U_
O
^
^
00
-
r«
10
U. *f
a •
U CO
a -
Z CM CM
i -O
o> »• •
- 1 ^m
to CM
- i- II
1- O
-co o
CO II (S
CM'TJ a
- u c
••-wo
•DM
ii C
9 +>
C IA Ol
3 L.
0 C «
Q.OI-
5 +J
u
a o
QL
O
C —
0 -H
M (Q
a
c
o
ra v
V U
(0 •-
O i.
1/1
0)
00,
82
a O)
-i ~n
a
t.
01
n <*•
^ .^
(8 >-
O (Q
3
C 01
«1
a
4^
c o
8«
OJ
&
a
M 3
8) O
< C.
u) c
3 O
M —
if
h
Si
0) U
a
g
0)
1
i/>
X
i.
o c.
^ 5
O 3
j2 2
n
£
u
m
0}
CM
cn
(0
CM
^^
a
o
r-
CM
i"
a
^
»
o
8
CM
Vt
r-
eo
8
i
(0
n
(O
in
^
in
oo
CM
•x^
a
o
in
in
T*
§
o
b
00
o
§
PI
I/I
in
s
i
r-
00
u
r-
n
(0
in
CM
\
o
o
VI
M
I/) H
IH Z
I/) UJ
>- U
_i in a
f^4 UJ
i/> a.
^
(/) _J O
-i IH < m
w _l O
V" ^ — J ^
M 3 o z o
h* ^*^ 3 ^
< _IZ U
1- < Ul CO U.
MUZ O
M Oh- CO
Z Z _ UJ UJ
IH u Z Z > O
X IH CO Z
o u t- 2
uj a o o a
w z uj z uj
ZZHi2iM
o o z a < t-
IH M iu X 1" CX Q
1- »-< 1- uj Z
U < UJ UJ IH
a t- a. o < oc
K z a z u
(/) U Z uj O IH >
£g2zK28
a H u < z u
< O 3 Z 0£ a
H -JO X
< ii uj < Z < VI
o o a o uj o
CM u u. M a. M
O Q ii u u u u
U. O
-i u u. IH a ae.
a
UJ
3
UJ
a
8
A-41
-------
t-
^
o
UJ I/I
ii
I/I U.
ui Q
£3
VJ
O I/)
a z
O X
U 0
l-l
i/> a
<£
^S
00 M
X IX
u. a
.J
00
(3 IX
ZO
(-
-J
Q
Q
O
£
00
•
r-
(O
a V
a -
O CO
a. -
Z CM CM
i -O
CO i- •
r- CM
ID O
-co o
eo ii ra
• IX
" L C
ii C
'U +* 01
C I/I 0)
3 i.
o c ra
Q n tmm
O R)
-H
ra
°
ra
c
01
c
£
0)
I/I >
a o
KH U
01
IX
o
c —
o *•
IX
c
o
ra v
•H O
(8 •-
a (.
VI
0)
IX
. ***.
00,
o fn
3 a
a o>
3 w
a
01
.^B
ra *•
^ •••
<9 *-
Q (0
§
S2
a.
^
CO)
O fl
OJ
a.
Sie
M c
3 O
IA -
C 0,
0) 01
oa:
E C
|cS
o
01
f«
1
I/)
X
c.
O (.
•H 0)
ra a
i. e
ra
c
u
A
oa
in O O en
CO CO CM CO
en O o en
CM •-
cn
CM
M IX
oo *- O en
(O ^* (D oo
oo in in «
•^ co CM
co r* oo
O 0 0
o
a a a o
o r~ o 01
§t in M
en oo »
o o o o
in o o o
CM
CM CM *- *-
uj O O
Z Z Z faO
CM en o cnnO(O*
oocoooMoioooor--«TOin(ooor--(Ocno)
»- ^^ ^^M^^CM »- »- ,
(O <- co cn CM (0
••-oO'-O'-'-O'-oO'-wcocM'-oo
aooooaaooaaoaooooo
OcooiiooocncMinocDin-o-iniDoocMr-
oaincniocnoc-r^r~cooor>^oocncnoocn
oinooovtnoooocMoaooooo
§'-OOOVCMOOlOO'-OOOOOOO
(oo5o«(OOOin t o >- o
CO'-COCONCOCMCOCOIOCO'-'-CMCOCOeMCM
O UJ UIUJ O OOO UJ O
Z(/>Zl/>ZZf>3Zl/IZZZ2Z3(/)Z
ococMCOOoocM^r^inco^ootocMiocn^
CM'-CO«eM'-COeM'-CMCM'---'-CO«-«-eM
en'-*O(D(OOcor-o>(DiDt*incM*»t
U)(DAOtts>LOr**oooocoo)^H'^^coo>uico
iiiiiiilillliliiil
Soocococococococococoeoooeococoooco
o888S||88§gSS888SS
o^cNn^tt^coOTON^rifltDr^oio^cN
(0(0(0(0(0(OCD(DrVk^*r^ts*t>>^t^r^ooooaQ
NWWMWO4CSC4WC4P4WC4WNNNCS
{0(O(0(0(0(D(0(0(0(0(0(OCO(0(0(0(O(O
A-42
-------
—N
< H
K Z
28
u w
u "
-i
too
z
oo 3
oc
i- £
^
MO
%
82
t-i
K O
< M
T Q
2
Q
fs, j^j
u. Z
1-1
u. a
o o
ax
|5
K Ct
VI O
^ u»
Q
a
u
Q.
X
CO
t*
(0
Q *
O CO
X CM CM
i -O
ao »- •
- 1 1-"
- ^ It
ID U
-co o
co u ra
a
CM -a
-'•I
55
M C
o
(0
*~
^
CM
(0
O
a
o
in
00
o
o
CO
oo
o
N
CM
r-
n
8
CO
a
o
CO
(0
in
o
(0
o
2
CO
on
o
or
CM
in
00
o
o
CM
„
*
r»
N
O
8
i
^
^
CO
ID
in
0
^
0
r-
^
CO
en
«o
O
o
en
00
CO
o
8
CO
y
*
00
CM
in
8
§
<
CO
CO
(0
in
en
r-
cn
en
0
in
in
,_
a
CM
J
o
8
•^
to
r»
r-
CM
CO
8
i
<
*
CO
(0
in
0
8
co
in
^
^
r.
a
o
8
o
o
CO
y
*
on
CO
*
8
oo
O
in
CO
(O
in
A-43
-------
^
< H
Si
ou
Ul
Z co
Ii
a
Ml ^
1- U.
tH
co a
oz
a. <
^
O CO
oz
cox
t- O
x a
a
t- UJ
00 h-
U. Z
M
u. a
o o
_i
a x
Z o
t- at
CO O
JjMj ^
J
a
a
u
i
oo
pj
ID
a t"
a -
0. -
X CM CM
i -O
co ••- .
Is* M
• ^ II
CO U
• co o
n n «
- a
CM TJ
- t- c
'-MO
"21-1
ii C
M +*
^ +* 01
c co 0)
§^
£ M
ao (-
§£
U ffJ
+rf
•^
c
o
M
C
{.
01
•f
c
X
L
Q)
CO >
o o
1-1 IT!
a:
o
c —
O -H
l-l («
at
c
o
•H O
(8 —
a L.
VI
01
U O)
c\
52
•*_••
Q O)
O \
Q.
.*_^
L.
ID
- o
-1 CO OZ
< l-l UJ
Z - 3
M _i a
H < _l h-
CO Z < CO
p$y is
< -If U
t- < U CO U.
CO U X O
l-l O t- CO
Z Z UJ u
M UJ Z Z > C9
X i-* 03 Z
a u i- <
u o Q o at
CO Z UJ Z UJ
3 w 1- 1- Z
Q « i-i
§z K z •- a
oz a < K
M M UJ X UJ 0. Q
(- K CO < 1- UJ Z
(J < UJ UJ w CO
11 ox ac a. oc >
ce i- a u < oz
H- Z Q Z UJ
CO Ul Z CU O M >
£z-2zr38
O 1- u < Z lu
< o D z a oz
< ii ui < z < co
a o oz o u o
CM u u, M a. M
OZ "~s
O Q ii u u n ii
U. 0
_i o u. ti a. oz
Q
Ul
CO
3
CO
Ul
Q
8
A-44
-------
H
Q
UI I/I
ol 5
5 ?
(/I U. Q
O P-
U O O 00
^m Z O *
w 3 i o
I/I CM
O I/I »- PI
Q.Z U
X i-i co o
° 2 ii «
i/) o u. a
i- a •o
< a o «- c
^P ^| Q M ft
*fc 2"
< n C
P- Z It 4*
00 w TJ V ffl
>- a c t/i 9)
U-0 31-
-j o c ra
U- Z o ft u-
00 I;
eg a u n
z o +*
M U. ••-
1- +*
I/I C
_I 3
_
*«
C
OJ
«
OJ
o o
» u
OJ
a
o
c —
o +•
l-l fl
a
§
M V
•H U
O L
VI
HI
a
0 Ol
c ^
801
a
^•k
a a
o^
J5
t.
0)
« c
3 O
VI —
C Ol
0) 0)
0«
Sv
•- 9
U 0
0) U
a
o
0)
I/I
X
L.
O «-
•H 01
a n
Q 2
ns
j
U
ra
CO
P- O O
oo en CM
en co CM
CM CM
•N^ -^
Q a o
a o
•J wJ
(0
00 (O *
ID (D »-
O O *-
§ § §
O P> O
§* in
en oo
»-
O O O
in d o
M
CM CM »-
* z1 8
N en o
^" v* ^
* CM CM
CM * O
888
00 00 00
< <: <
* in P-
in in in
CM CM CM
(O (0 (0
OOOOfo(DOCMOfoom»-en**
§*-CO*-QinCM(0*-CMCMCO*-'-CO*-
T-^Oencoop-»-enors-p-cneoen
CM P-
U> in
0 b
CMCMCMCMCMCM CMCM
x^ x^ x^ N^ %^ x^ >^ x^
0 O Q Q Q w 0 Q w 0
OOOOOO 00
co oo en CM cocnoo *-
*- oo »- co p-enr- CM
00 *- T- CM »-CMT- M
in n r~ ID CM en
oo(Do>p-»- en*- '-cococoin CM
ineooocM^tvcncn oo o co en (O oo
~O»-OCM'-OO »•»- o O '----'•
in in
cnenooocn(OOOencMinoa)in«in
SS££$S£8;:£££££?£
ooowooowinoooocMOco
°S858885S88S°582
-"CO'-COCOeMCOeMCOCOCOCO'-'-CM
..^coz^SSai^^.^^^*
»*OCOCMCOOWCM«P-inCOP-00
O
U
CO
jj
(D
O
01
O
o
CO
^
CM
CO
in
00
o
O
00
a
<
en
CM
ID
CO
0
in
CO
T_
^
b
u.
HH
P-
CM
O
at
00
CO
01
o
o
CO
*
(0
CM
01
CM
8
oo
<
0
oo
CM
ID
CO
O
CM
CO
^
P-
O
^
l-l
p-
«*
CM
O
CM
00
O
°
CM
(/I
en
in
CO
O
O
oo
a
u
<
»-
co
CM
ID
0
,_
*~
00
^
^
hH
^
f*fc
CO
o
at
IH
P>
n
o
8
CM
*
*
CM
CO
i
oo
<
CM
CO
CM
10
CO
A-45
-------
< H
< O
O O
Ul
5 *"
to <
at
Ul D
t- U. Q
00 Q (J 00
O Z 0 •
Z CM
O I/) ^H
IH no
x »- •-
1- O *-
< W MIQ
I o u. at
Q O L. C
00 H ID H
U. Z « -M
n ft£ ? yj fij
oo 5 u
(IT rt O U>
2 5 JS K
\- at o «
o a
Q 0)
O \
ft
^
t_
01
ra iZ
« —
a m
g
C 0)
u *^
01 X
a.
^«
C 0)
u 'ra
L Z
Q)
(L
Q.
Oi O
8
§
8
CO
00
CM,
(0
CO
o
in
*•
CM
Q
O
(0
^
^
Q
Z
O
8
^
0
I
^
Ul
Z
CO
00
CO
CM
§
oo
a
«
00
CM
CO
ooooooooooooooooooooo
enocoeov^r-fMO>o»p-oo*-incoo*cooinocMOcor-OcononinoocMoicMO
oooaoooooooooor>(orooocoO«roir)or-oor-o>
-oe-oo^onooinencM-wo-iflNOeMi-co
weM<0ocM222co*oocMco2
-------
-
< o
a o
Ul
l|
ui D
H u. a
I/I O O co
C3 Z O *
a. < i o
Z CM
o i/» «- »
u z u
M CO 0
I/I X *- —
1- O •*
< M ii n
z a u. cc
2 a u
a o c. c
i- ui o a o
00 H O W
> < u C
"" M r * g!
u. DC C I/I 8)
O O 3 (-
-1 OCA
w 5 £
t- os u a
I/I O -H
H U. •"
""* c
i
f_
A
C
i.
01
c
X
01
M >
o o
H U
01
a
0
c -
o +-
H (fl
CC
§
It +•>
•H U
(9 —
a (.
i/i
0)
"2
L
01
(0 *
o *n
3
5** 0)
+J
o •-
L £
0) 3
a
4^
S-
o m
0)
d
a
ul o
o
«i c
3 O
£5,
oi a
u a
ic
I1
a
Sample
X
L.
O C.
•f 0)
n 5
L 6
El
<0
-1
U
4*
n
CO
O
00
T"
^
CM
X^
Q
O
OI
CM
i
-
{J
O
8
CM
I/)
¥••
O
ACD8700
o
-1 C/I OC
Z i/i a.
M -I O
-J M < in
< viz -
u > <
PS!., e
I/I Z < I/I
w < O Z O
< -1 Z U
K < HI I/I U.
MUZ O
n O 1- CO
Z Z UJ Ul
M Ul Z Z > (9
51-1 ca Z
i- 9
ui o g o a
I/I Z ui Z u
D M H HZ
§< < 11
§nK
o z at < t-
n M ui x u a. o
(- (- 01 3 t- u z
U < UJ Ul M (/I
w a ce a. a >-
ce H a. o < ce
h- z a z ui
I/I ui Z ui o n >
U CJ O »- M X O
Ct Z M Z K- < U
81- UJ < Z UJ
3 z a a
i- -i 3 x
< n ui < z < i/i
Q O OC O Ul O
CM u u. M a. 11
o a i i
u. o
_i u u- w a. cc
a
Ul
M
I/I
Ul
Q
8
A-47
-------
i
U (/J
_j z
z K
< 3
co u. a
a r-
Hi O O co
^ z o *
•-•<. i O
CO CM
2 CO »- II
z u
z IH co o
IH II (V
co a Q at
i- a TJ
< a o L c
In °|5
P- Z (9 *"
oo M 'D +* 0)
> a c co a
U. O 3 L.
^ 0 C (Q
It I n n tmm
°° §£
ZO °5
l-l U. ...
1- +>
2 S
a
_
ra
c
L
01
C
HH
L
0)
CO >
a o
ac.
o
O -H
i-i ro
a:
c
o
ra -H
•H O
flj -
a L
U)
01
a
U D)
82
a a»
O ^x
— 1 B)
a
i.
0)
n> M-
a 15
s
S3
o •-
l.£
V 2
a
44
Si
U (0
i- Z
01
a
O) O
< L
(9
Ul C
3 O
I/) •-
C D)
01 01
U K
C
i C
"™ 3
88
a
f^
9)
f»
A
CO
L
O L
•H 01
n 5
O 3
u
n
ca
r~OOOOOOCoinOCMOCooin'-en o
co a
00 O
00 O
CM CO
in
2 Z
(O CM
*- CO
« in
r» oo
CM «-
8 8
00 00
8 8
< <
r» en
r- r-
CM CM
ID U)
O
in
CO
"
O
(D
CO
n
en
a
oo
CO
en
o
o
CO
3
(O
CM
on
CM
8
00
a
o
00
CM
ID
CO
o
CM
CO
"
O
1-
r~
CO
CO
a
M
OO
O
o
CM
CO
en
^.
in
CO
8
ao
a
u
^
00
CM
(D
00
o
^
"
en
in
CM
O
CO
0
r.
en
0
§
"
CM
O
Z
^
CM
^,
CO
1
00
a
u
CM
oo
CM
(D
A-48
-------
^^
< K
<0
°§
U
iz
to 3
a
U 3
Hll Q
Ik U
25 ? o
Z CM
^J Z O
•H WO
(SI 4^ ^- •^«
VI ^ ^ ^
HO **
< H n n
z° IT,"
Q 0 C C
Is* U O (Q O
oo 1- •gw
>- < n C
U. Z fl -H
IH TJ V 01
u. or c to 01
OO 3 C.
-1 OCA
«" I"1"
h™ Q£ U (Q
wo *•
_l •<->
c
I
a
L
0)
c
M
X
L
01
a o
a
0
O -H
HI (0
a
c
0
•M U
a £
V)
Q)
(C
— »
o cn
82
^^
a en
o\
ff
•~
9)
(0 **•
^ '•"
(9 —
Qffi
Ml
S
C »
Si
a.
^
s w
u "n
i. X
01
a.
« §•
Oi O
< L
3 O
i?
oa
i C
5E
0
1-4
«
ra
CO
X
O (.
*> Q
fl 2
^ e
Ei
5
£
U
m
m
o o o o o o o
in in en o co oo v
CM ^ ^ cn ^~ ^ oo
at x at ct
* U) CO S ^ CM 0
^ *^
CO
0> *t O O CO ^ O)
r*«, in v^ f^ ^ ^ *^
r^ w ^ r< ^ ^ v
o o o o o o o
co o <«• r- o in o
0) O 00 00 CO 00 O
^ t"
O 'O O *- O O O
Q O O t*- O O O
o in o oo
^" V
111 lit 1 \
x z cn 2 z z z
"
in co r- * o to in
^" "!•" ^ ^| ^"
e» oo »- o c~ oo •-
CO CM CO CM »- O CM
8888888
00 GO 00 00 00 00 00
Q O O O a Q Q
u o u u u u u
<<<<<<<
co n to co 01 *- co
00 00 00 00 CO O) O)
W W O* W W W
^" ^
O UJ UJ O O
F^^OOOlOOOCDO^fCO
OCOCOWWN^OJCO^
cooin^^-co^cstOM
^ ^ ^ CO ^ O O CS CO *™
8888888888
oouoouuuou
^•inc^cno^-M^tDco
tn
-------
„ ,
< H
< a
a u
*— . •
UJ
I/I <
Q^
UJ 3
H u. a
M a r-
1/1 Q U CO
o. 5 i b
Z CM
O I/I ^n
CJ Z O
p-l CO O
I/I X *- •-
HO +*
< w u ig
5° ST:"
a o c. c
r- ui o n O
CO H TJ M
>- < n C
u. Z <0 *•
M O *• 01
U. K C I/I ft
o o g t.
-i o c ra
* ° § £
H DC u n
t/l O +>
M U. •<-
C
n
i
r—
E
0)
c
hH
X
01
i/i >
0 0
M fl*
ox
c-2
o +-
11 ra
at
c
o
(9 -H
*• O
a L
M
0)
a
^^^
00,
0 O)
u a
00,
a ^
-1 0)
a
L
01
(Q *^
•H —
Ifl —
Q n
Qf
C 01
8*
-
t. .C
01 3
Q.
4^
§ -
om
01
a.
01 3
710
< (.
M C
3 O
I/I —
01 01
u at
c
i£
•53
$ u
a
a
n
Sample
X
t-
o c.
•H 01
(0 Q
^ g
0 3
(0
J
J.
u
as
m
o
co
*•
J
^
M
O
CN
in
O
a
M
[V
^f
o
8
(N
C/)
f*B
(O
o
-i i/i a
< M UJ
9 >•
I/I -1 O
-i M < in
< i/i z »-
o>- 3
w -J O
H < -1 H
i/l Z < i/>
•i < U Z 0
1- n < *
< -1 Z U
t- < UJ I/I U.
MUZ 0
11 U t- CO
Z Z UJ uj
M UJ Z Z >• (3
5n oa Z
H 5
uj Q o a at
(/> Z UJ Z U
3 M 1- HZ
ooz z S < H
11 M uj x uj a o
H H !/>< H U Z
u < iu u w i/i
M a a: a. a >-
a H Q. u < a
HZ a z u
i/i ui z u o M >
U U O H M X O
a: z M z H < u
O H UJ 4 Z UJ
s
o a u t a
u. o
_i uu. M a at
0
UJ
I/I
D
UJ
a
8
A-50
-------
H
Q
UJ I/)
s!3
*. <*
< a
I/I U.
IU O
|_ 2
** 5
CO
O I/I
Q- Z
O X
uo
IN
00 Q
• K
U. O
u. X
OO
(9 QC
2 O
JHU.
(/I
IH
~*
Q
Q
O
X
oo
r-
10
o m
Q .
U CM CM
X -CM
I I i-
00 CM
• ^m II
r*" O
• CO 0
4"1
co ii m
- K
V JM Q
'•as
u C
A -H
C I/I 9)
OCA
Q.O (-
0 +•
U A
£
C
A
§
A
£
0)
C
L
0)
O O
a
sS
M A
at
c
o
A -H
•M O
A —
O t-
V)
01
a
^
go,
82
a^>
O *"*-
*^
i.
0)
A 14-
A —
a A
C 0
"i
01 2
a.
+j
c. o
SA
01
a.
0)3
a) 0
< L.
(9
3§
M —
c ai
5V
g
•^ 3
U O
S U
S
VI
Q
HI
01
A
I/I
X
C.
o t-
A &
o 2
A
j
£
U
A
CD
o t to co
* in en co
UI (O ("*• 00
a
o "t n
CM CO 00
00
in
CM
§
c- en o en
* CM in (D
en * oo en
O 0 O O
° ° 8 8
«• «•
CM «- CO CO
z! i/> ^ z
en * o CM
»-»-««
in oo en *
CM •" u en
§888
00 CO 00 00
§ n §
in oo o CM
in in 10 10
CM CM CM CM
(O 10 (0 (O
^t^CMTfCMUJOCMCOaD^-OOUJ*
s
in
IH
CM*- »-CMOOOO'-CO »-
cooo'-M iD'-roeotooiiniot
incoencM IOCOOI^CMUCMCOID
" " * *~
(Dtoocooinocor-cocMCMO'-
Ol ^f CO CO U) ^ O CO CO *f If) ^ O ^
(DtOlOiOOcoOooi^lOOOnO^'
t» v-
oooooooooooooo
88dci8ciSci8cic;8ci8
^ ^" ^ ^ ^ ^
CO»-COCOCMCO'-CMCMCMCM'-COCM
bOZ2XZ^3ZOOOO^OO^OO
coaoa)inOioinooen<«r>-r-enr~
CM»-»-^'CM»- ^^^N^^T-
OfCMenr-ao^ooincoor^cott)
en<-cno(Or-Oin«(OincM«n
COOCMCO^OCM^COCOOCO^CO
88888888888888
OOOOCOOQCOOOOOOOOOCOCOaOOOOO
QQQQOQQQOQQQQQ
(Ot'-COCOOOWWCDOO^^^^
WWCJWMWWMWCOCOCOCOCO
(0(OCD(OC0(0(0(0(0(0(O(D(0(0
^™ w co co co co ^r ^ ^ tfl in in u* in
CM
^
a
o
(/I
M
M Z
^^ £
to iu
> o
-1 00 0£
< 1-H UJ
^ ^
00 _l O
-i IH < m
M _l O
t- < _i i-
« 2 o z o
K IH < *
< -1 Z U
t- < IU I/I U.
OIL) I 0
•H o 1- CO
Z Z iu iu
IH iu Z Z >• <3
5>H 00 Z
i- <
iu o oo at
I/I Z iu Z iu
3 « (- 1- Z
§< < IH
IH a
o o z a < i-
h- t- I/I < H IU Z
U < UJ IU «H bO
IH a a a. a: >•
a: H a. o < at
K Z O Z IU
I/I IU Z IU O M >
IU U O h- M X O
a z IH z »- < o
^m ^ 9 ^
< n iu < Z <
-------
5
a
u en
«S 3
CO U.
Hy
^
CO
& Z
§x
o
CO Q
is
2£
1** Z
00 11
>- ce
U. O
It T
00
ts at
Z 0
M U.
1-
co
HI
~*
a
o
u
X
1
00
r»
ID
a co
a -
U CM CM
X -CM
i i »-
CO CM
* *- II
1- O
-co o
(0 t- .-
CO u (9
- at
w- «. c
7i~
« -P
U +• 01
c oi
R o "
Q£
U «s
£
c
ID
i
*_
(0
c
0)
c
>»
(I)
00 >
a o
M U
a
0 -H
HI m
a
c
o
« -H
•H O
a L
in
01
at
^^
00,
82
a 9>
o ^
^
^^>
0)
•VB
4^ *^
a n
o
C 01
U -^
"r
01 3
a.
s *
0 «
a.
01 3
010
< s.
(9
(A C
3 O
M —
SO)
01
u «
c
i C
oi
0)U
a
a
^^
01
i/)
X
L
O L
ra a
L E
W
i
£
U
n
m
OTT(aoQ^r--CM«cMtt)OcMncn(0'-co(o
^incnrooncocMtotorocMvooi-'-aoro
^*
(0 r-
^ ,_
QC 1-4 l-|
^ w c^ o oi r^ oo ^ c*5 in ^
^-cooo««(oco(0 r- o> to O) r- r~ in ^r
»- »• »- »- v
j^
(^
O
o
r^a>oo>(o«cocooinO(orocDt^Oifl^c')(oinc^^f
^ O O ^ fO O ^ CO ^™ O ^ *~ 0? O f) O CO ^
888888888888888888
aooooooocooooooooocoooooooooooooooco
8QQQQQQQQQQOQQOQQO
UOOOUUOOOOOUOOUUU
<<<<<<<<<<<<<<<<<<.
inin(Oioo>o>a>oO'- — ••-
CMCMCMCMCMCMCMMCMCMCMCMtMntOPJCOW
(0(0(0(O(O(0(0(0(0(Dffl(0(0(0(0(O(O(0
*
co
(O
^.
s
1
CM
\
Q
o
I/)
»- CO
5 v
-1 CO
< HI
1 SH
O CO -1
§< CO Z
o >• ^
^ I- < _l
co z <
H. 3 o
t- HI
CM < -11
CO 01
H O (-
Z Z UJ
HI uj Z Z
X HI
co o u (-
uj a o
CO Z UJ Z
3 HI t-
o <
o o z 5
^- *-»-«/><>-
U < UJ UJ HI
M a a o. a
a i- a. o
i- 2 a
(0 CO Ul Z U O
(O LU CJ O ^ Hi
CO Q( Z HI Z ^*
§O H UJ <
< u n z a
§< II UJ < Z
a o a o
U CM O U. HI
< at ^
o a » u ii
u. O
-j U U. HI
Q
(0 LU
»• CO
CO 3
(O
UJ
a
o
U
A-52
-------
^
1- in
< Z.
a <
_l U.
if
M
vi x
< 2
I Q
O
r- u
> <
u. Z
IH
u. at
0 0
_l
" a
en o
i^
a
u
i-
00
^
CO (0
CM •
U. 1 O
a CM
O »- ii
i m o
00 f —
r- n w
- a
CO V
-c. c
CM n o
•b w
ii C
p ** u
C «/» ft
3 L.
o c n
Q.O H
6 V
u n
c
I
C
0)
c
M
in >
*•* u
a
o
c •-
O -H
"1 IS
OL
C
o
Ifl -H
+j U
fl .-
Q C.
(A
01
a
. S-*.
gm
82
O)
Q >sv
32
•»*•
(.
m *.
Q lo
<§
a
0)
f"
i/i
^"
O L
«E
J?
£
U
n
oo
enr~<0incMP>oo«ooinm(ffTr-«
(DO — '-'root'-cnoooiocN'-'-oiro'-a)
p-oooot-(or-t^ooin^r>(0(o^r-t>.inr-r~(o
l» *- oo
CM CM O
^ ^ ^»
CM, CM N CM
**^ ^^ >HO.z>-ia.a.Zi-iaa.
iuii ^i uiuiiu,
j U ^ ^ U ^ «J U ^ ^ ^ U «J ^ U
oa^H oa ^ M CQ x MCO <-> x a i-> x
w a _i _IMO. M_ia -IO.M _I&M
ao.(/ioaoa.i/)ao.oi/iaoi/)aooina.
L:^SiwS§S&sS§&S§3wiSs
i-c^cotoincD'-cocoinr^ocHCooocoinr-cNoo
C4C>4C4C4C4NC4CMCie>«C4nOC*)COO
CO(D(0(D(OU)(0(0(D(D(0(0(OOD(D(OU)(0(0(D
^^"*"*" ^vvininuim
A-53
-------
<
1- .
HI
(/) >
a o
M O
0
c —
0 -H
HI (Q
a
c
5*
(Q •r-
0 L
W
0)
a
o'o)
c \
82
>«r
0$
o n
3 a
t.
V
« •-
a n
3
o
l-l
(U
B-
a)
i/)
X
t_
o «-
A a
1. E
C
u
(Q
00
O CO
O CM
r- r-
N
a
o
"^
CO
^
CM
(O CO
an in
i- O
0 0
i .2
^
§
J uj
oa ^
a a
O VI
t- X
UJ O
Z -1
•
»- CM
in in
CM CM
(0 10
r- v
(D ^ ^ *" O (O
r> CD CM co en co
r* (0 to r^ (O r*
CM
Q
O
J
CO
CO (0 •>- CM 00
r~ o en r» to
in <• ^ in
CM oo * in oo *
r» oo * co «- (0
O *• CM «- CM '-
o o o o o o
a a a a a a
a. a z a. a. a.
^
UJ < UJ
•^ J UJ ^
M 00 ^ M
a. -i -i M a.
1/1 o a o a. i/i
at o at vi
M O UJ O O HI
z o z o -i z
co to in ID — co
in m co 10 r-> r-
CM CM CM CM CM CM
(0 CO (0 (0 CO (O
*- «- N CM CM CM
CO
en
(0
CM
a
0
J
00
O)
o
o
§
^
5
j
03
a
i-
UJ
Z
oo
r-
^
O
CM
O
a
a.
UJ
N/
a
in
s
o
_i
in
oo
(0
«
(Oco'-enmOcM'-io
»-r»coenoiin^oo)
r-r-p~r~r»O(ooip-
CM CM
O a
o o
J J
CO ID V
r^ co (0 en *~ co (o
Ol CO 00 ^ ^ -CO - CO
in in •*• in
«cooo»- cncop-'-
in«(0(0in*>*-on
'-'-'-^•inint'-'-
ooooooooo
ooaoooaao
Q.Q.ZQ.Q.Q.ZQ.Q.
^ ¥
UJ ^ UJ 9 UJ
^£ «J ^ UJ «J ^
11 00 H ^ 03 l-l
_j a. _i a >-" -j a
Si/iaoc/iaaoi/i
o a: en o a
§MUJOi-iOUJOi-i
IZCJI-I3EOI
r~o«M(Doocoinr-cM
oocnenenaiOOO^
CMCMCMCMCMCOCOCOCO
(OCDCDIOCOCOCOCOCO
cico**«t«tininin
P.
r~
r»
p-
^*
m
CO
0
a
a.
UJ
\s
a.
o
_i
oa
CO
(0
in
A-54
-------
<
K I/I
< z.
Q <
at
UJ 3
_l U.
Q.
5 a
z
1/1 <
*8
(/) X
1- 0
x a
a
r- ta
to H-
£2
u. at
o o
_i
S5
i- a
I/) O
j
L.
8
u
a.
00
1^
n
U. CM U)
a -in
o *- •
UJ I »-
a. CM
co U
-n o
js» ^ .^
nun
OS
N-?c
•-IOO
1,1"
(8 V
•D +* B
C «/> 81
3 (.
o c n
Q.O K
5 t!
+•
+•
c
ra
3
1
0)
C
c.
(/) >
o o
Q
c •-
o +•
a
Q
Q ,jj
Q i.
4^
M
01
DC
i •— >,
O O)
C ""s.
82
^J
Q ^s.
32
s_^
L
Q)
Q 4
Sample
X
L.
O (.
n a
\1
n)
_i .
r
u
R)
CO
en »- in to co co in
ID oo in in co t t-
oo co en oo ^ en co
CM
CM CM CM CM
Q a Q a IH
o o a o
^ ^j «j ^
^_
en r- (O
CM « en
»- in
§™ CM 8 - W
»• » " O 0
606600
o
§ 1 2 i § i 2
¥ ¥
3 uj 3
-I UJ it -I IU
CO ^ M 69 ¥
•H O. -1 _l M
Q a. (/> o a o a.
Si/) a o ae «/)
H * 5 Z ? Z *
SI 3 S 8 £ 8 S
*- w co (0 in (0 *-
in in in in CD to r-
C4 N C4 C4 M M C4
(D (D (0 (0 (0 (0 tf)
i- CM *-
CM n i^.
01 r- co
CM CM
a a
0 O
-i -i
CM
6
in
»- co
co r- »
(MOW
o o o
2 § i
^
u 3
^ J
M CO
85 I g
s
O CM (O
en en en
CM CM CM
ID ID ID
m * *
o o o
n en in
O * CO
CM
a
o
J
oo V
on »-
^ ^
CM r- n
»- i- en
(0(0'-
O O 0
2 a a
a. z
x
ui 3
iC ui -i
w * 00
a. IH
«/» a. a
I/) O
S » 5
£ a uj
oo « in
en o o
CM co n
ID 10 ID
*f t in
r-
(O
en
CM
a
0
co
O
O
O
CONTROL
t«-
§
CD
in
o o
§n
0
O ID
ID 6
in CM
CO O
CM t
o o
o o
a a.
IU
^ UJ
IH ^
a. M
I/) O.
1/1
§ *
S 3
CM CO
CO ID
in in
A-55
-------
<
t- V)
at
Ul 3
-J a.
a
*l
I/I <
°2
VI X
1- O
x a
a
r- ui
25
u. Z
M
u. a
oo
_i
M
i- at
I/I O
_j
u.
a
u
Ul
a.
i
oo
•
.
.
CO
.
CM
II
|
|
g
u
Q_
Q
u
Ul
1
00
r»
CO
CM in
-in
f
t i~
CM
*- II
U
co o
+«
n ra
at
TJ
L C
?o
M
ra 4*
•H 0)
i/i 5)
L
S"
-H
(Q
+rf
+4
C
CO
3
C
L
01
c
*+
a
M
C
o
w
ns
v
fll
Q
H
8
Q
_l
(0
i
X
C.
n
c.
n
X
0)
o
u
01
a:
o
+j
c
0
!H
U
L
«
01
a:
In
2
ai
_
01
t*.
"n
B
O.
n
01
a
re
1/1
«
O
E
u
+J
ra
00
tn ^ in ID eo
ID oo in in oo
co co cn oo r**
CM CM
^ \
Q O
O O
J
r> » CM
O v- 00
*• r> CM
(0 CM 10 ID
in to oo en *-
CM T- CO O O
o o o o o
a a a o a
z a a a z
^ ^
2 uj 3
_l Ul ^ U
CO ^ n 00
M a J
a a. i/) o a
o 1/1 a o
ui O M o Ul
Z -r x o z
•^ CM co to in
in in in in o a. o
at w o 1/1 at
z s 5 i- s z
O O M Ul O O
0 -1 X Z -1 O
(0 *- co oo in r»
CM CM CM CM CM CM
CM CM CM CO CO CO
en CM ^ o O o
CM co r^ co cn in
cn t^ co o « co
CM CM
a O
O O
J J
* en o O
CO •* f CO
(0 CM I- CM
(0 (0 en co cn cn O O
CM CM CM CM CO CO
ID ID <0 (0 (D ID
co » » * » in
p~
(O
cn
7
CM
CO
r-
co
0
2
_l
§
o
u
r-
0
CO
(0
in
o
o
o
00
r>
CO
CM
<«•
CM
O
a
a
UJ
n
a
V)
i
n
X
CM
CO
(0
in
o
CO
o
0
CM
CO
ID
CO
O
2
UJ
^
h4
a.
c/i
o
_j
oo
CO
(0
in
A-56
-------
^
1- I/I
< Z
a 5
a
Ul 3
u u.
a.
t/1 V
V) ^
1- 0
< 11
r- ui
00 I-
^B ^
u. Z
°l
w
H a
CO O
M U.
-1
§
U
Ul
a.
i
oo
i-
n
a CM in
o -in
o — .
Q. CM
1 »- II
oo U
-co o
["* ^ -^
CO II (Q
at
CM TJ
•fe S
*• flJ O
"5 M
ii C
(Q +*
TJ +• 01
c 1/1 8)
O C (8
Q.O h-
o n
.^
4.*
C
(Q
1
0)
c
M
>,
L.
A)
l/l >
a o
o:
c.2
o *>
a:
§
n -H
*• o
n —
a s.
VI
at
a
. y—*.
00.
32
Q SN.
32
,_^
u
.^
m M-
Q (Q
CT
O
0)
&
n
CO
x
3 v
(Q 2
m
j
*
c
u
m
CO
oooencnocoOcM
OOCO»-CM'-'-CM'-«
r»O»-enoiOenor»
18
•
N CM CM
^*X ^"^ "^N,
gu a a
0 0
en i- in en o co
co en en ^ ^ ^
CM in *- CM co r-
(0 oo oo in to
*• in (p oo o en in CM
^ cnot^iocof^oo
O O"OO'-OO
o
iS5221222z
^ v ^
2 Z 2
3 uj 3 "i ?
J UJ ^ _J Ul X -J
a) * IH CD ^ w oa
HH a. -i j w a
aaioaaoacna
luoN^oujoo^HUj
Z-IZUZU-IZZ
T-cxcotDinto-^ooo
ininminujcor-r-r-
OICNMCSCNMMCMW
(0(O(D(O(O(O(0(0(0
^^«-^wwwww
0 O 0
CO CO O
00-
r- o) r~
- 6 -
CO CM (O
f-
10 00 00
•0- C1 (D
CM - O
a o o
a. a. a.
Ul
U| ^
^ M
M -i a.
a o
-------
<
^m W
UJ 3
-1 U.
Q.
co5
^
n
CO X
H 0
< M
i°
a
r- u
CO t-
X
U. Z
n
u. ae
0 0
_!
(j x
z o
i- at
CO O
"
u.
a
u
x
X
00
»;
u. co"
Q -
U CM CM
00 CM
- ^ II
r- o
-co o
*^ ^ <*•
*•
co u n
- at
CM 1]
- C. C
'-MO
•B~
u C
It) 4^
TJ *• 0)
c co o)
3 t
0 C 19
O (0
+J
<->
c
a
a
01
1
>»
(U
>
a o
0)
a
o
c -~
O -H
c
o
(Q 4*
4* O
O L
+t
Vt
(D
U O)
C \
32
O)
O 0)
-i a.
N.v'
C.
0)
(0 *4-
Q (0
§
0
H4
Sample
x
0 «-
•H 01
m a
A * 1
£2
ra
£
U
(Q
00
in en co r* o
* en in in t*
oo CD r-. r- r~
CM CM
•"S^ -S^
o u u u a
0 0
~J ^
r- en
S* CM
CO •*
in ID
*- *
— o
d d
000
a '-»-'- a
Z 0 0 O Z
¥ ^
< in 5
_l LJ *; _l
oa * M oa
a & co o a
O CO (X 0
i- 3 a z H
£ 3 S 8 £
«- CM co ID in
in in in in at
CM CM CM CM CM
ID ID ID ID ID
^ v »- »- CM
«-
CO
CO
u
CM
en
CM
0
o
CONTROL
(0
ID
CM
(0
CM
<-
00
O
ID
in
in
o
o
UJ
n
a
CO
§
,_
r*
CM
ID
CM
-
en
U
co
CM
O
0
UJ
^
pjH
a
CO
5
CO
r-
CM
(0
CM
CO
en
00
CM
a
0
-i
CO
o
d
a
X
5
_i
oa
METHOD
CO
r»
CM
(0
CO
ncn«incnincncn
CMinoooinmiooo
cor~uimiDior-u3
CM CM
U O 0 O 0 U Q
o o
in r~ CM o
» co in r-e»*-
in v- o m f in
CO 00
o co o
in »- CM
o d o
o o 32°
OO.OZOOOZ
)g )L
UJ < IU 9
UJ ^ — 1 ^ UJ _l
^ w oa i x oa
M _i a. _i a M
aocoaocoaa
: z § ,1 z 5 : g
oo^^iuOi^ouj
-IUXZUX-IZ
inr-OcM(Ooocoin
oooocncncnenoo
CMCMCMCMCMCMCOCO
(O(0(D(D(0(0
-------
<
1- I/I
21
tt
UJ 3
-liL
g2
a -
U CM CM
X "CM
X " ^
00 CM
- ^ II
r- o
-co o
- £
co n n
- a
"fe§
a IM
n C
18 •¥>
?+> V
i/) a)
3 L
U IS
v
V
c
n
M
£
C
IHl
t
i/) >
a o
a
0
o •*•>
a
§
(0 £
•H U
IB ...
a c.
VI
u
at
(JTO
52
9^
3a
i.
01
nj M.
v —
1
Q
Of
I
X
O L
4-> at
(Q A
sj
IQ
-1
U
1 Q
0 0
_J -I -1
in to in CM en co
to en CM CM co oo
co en »- »- CM en
r~ v (0 in « co CM
r* to •*•*-* at o>*
•-»•«» O n coo
oooooo oo
O
SaaoooeeoQ
o.a.a.za.i-ia.z
< u < ui <
-1 IU ^ -1 UJ ¥ _l
a ^ H ca ^ >-> OQ
w a. -i _/ 1-1 a.
aa.i/)oooa.i/ia
Sc/i a o of 1/1 o
.-*§£?£*§£
£3i8£83££
^
"
«-WCO(OUl(0'-C*)00
iflioinia K Q K 1/1 Q
, £ 5 £ £ 3 * ?
3S£^8£3SI
inr>OcMiooocoin
eoooenenenoioo
CMCMCMCMCMCMCOCO
(0(0(0(0(0(0(010
cococo^^v^fin
co ^ in
r- on *-
ID r^ r^
r> CM
CM in CM
»- o co
^
»- r» oo
SCM 00
in oo
o o o
a o o
a. a. a.
in
^ UJ
M X.
_i a. M
§1/1 a.
V)
U X -1
f- CM 00
O - "
CO CO CO
ID ID ID
in in in
A-59
-------
<
o
-co o
ID *• ••-
+*
* ii ra
CO B
- L. c
CM (0 O
u C *"
ra -H
TJ •* «
C I/) 0)
Oc a
U ra
*••
!£
c
(Q
3
ra
c
Q)
c
X
01
O 0
hH U
at
ox
o
Q "~
M ra
*
c
o
(0 -H
•H O
(0 •-
a L.
+*
i/i
V
Of
00,
32
OJ
32
0)
ra *
a ra
3
a
01
I
n
1/1
x
0 (.
** 9
ra n
A ^
r
£
O
00
in o) co r-
v 01 in in
oo uJ r- r-
CO
CM
O
CM
O H
0
ID
tO CM 0)
O (0 «-
co en
00 (0 00
r^ en oo
CM »- *
odd
in
a a o £
z a. a. •-
jg
3 iu
-1 IU ^
m \/ fc^
M a. _)
a a u) o
o i/) ac
K s 3 z
iu O M a
Z J I O
^- CM CO (D
in in in in
CM CM CM CM
ID ID ID ID
^ r. f. r-
0
r-
r-
CM
a
o
CO
in
o
d
z
^
5
i
oa
METHOD
in
ID
CM
(O
CM
»-
en
oo
CO
en
,_)
01
ID
6
2
_l
CONTRO
CM
(0
CM
»- *- co « en « in
*- »- en CM in oo o
co oi co oo r- in in
CM CM
a u u u a
o o
in ID oi r- o
d oi i- « «
CM en co »- oi
Sis s
odd d
0 0 D ° 2 ° Q
a a. z o o o z
¥ ^
IU § UJ 3
IU ^ — 1 IU X -1
X •-> 00 * 11 00
Ha M _i a
a i/i a a o i/> Q
I/) O I/I K O
X O H 3 Z J3 t-
O *H IU O O H IU
-I I Z -1 O I Z
*- co oo in i~ o CM
i^ r. is. co oo en en
CM CM CM CM CM CM CM
ID
O
CO
(0
in
•»
en
r»
U
en
O
o
Ul
^
H
a
1/1
i
1-4
I
CN*
^
CO
10
in
in
r-
u
t^
CM
CO
o
^
o
IU
V
»H
ft
O
00
^
CO
(0
in
A-60
-------
<
si
at
u D
_l U.
li
CO <
82
t/\ ^f
VI ^
1- O
< M
5°
Q
f- IU
00 I-
5- <
u. Z
u. Z
O O
_l
H DC
1/1 O
_j
u.
Q
U
^
|
00
£
»
U. CO
U CM CM
X -CM
I i *—
en CM
- »• n
CO U
-co o
.'~ ^
co n n
C.-TJ*
~&g
•Q H
ii C
« -H
TJ +* fl)
C CO O)
OCR
u n
*>
+<
c
ra
|
OJ
c
X
L.
Ill
(/) >
O O
*H U
U
tf
O
i^
IH n
K
§
IS •£
•M U
ra •-
Q t-
VI
4)
or
U 9)
32
0^
32
^^f
L
0)
n <*•
19-
u (0
a
Q
V
1
X
L
5fe
n 2
II
n
£
U
A
00
in
9
oo
CM
Q
0
CO
CO
o
0
s
^
§
J
CO
a
£
t-
UJ
s
in
CM
10
^
en co r- o
en in in r-
CM CM
a a
0 0
en
* CO
CM O
CO CO CO
»- co * in
CM in in o
o o o o
o o a Q
a. a. z z
^
IU <
IU ^ _1
^ i-l 00
M a -i
Si «" § i
.55?
0 H 0 UJ
CM co (0 in
in in in ID
CM CM CM CM
(0 (0 (0 (O
«-»-»- CM
<-'-'-co«'O>*incninencn
cn^t-encMincooinin(Oco
ooooenoocot^ininioioc^u)
CM CM CM CM N CM
"""S, "^ **V. "N^ ^"Si "*»^
Q Q Q Q Q Q
o o o o o o
_l -I _) -I _J _I
r» o CM *r co ^
oo co O ** r» oo
v- en CM r- en »•
CMC9'-* »-CM CO CMin
»-CM»-incO«OI^CMt~COtB
inioinocoio(0^en«r~cM
OOOO-OOOO'-OO
Sa?°§°s2§So?s:§
^ ^ ¥
IU < IU $ IU <
U ^ J 1U ^ ^1 ^ U J
^MCO^ MOO l-l ^ 03
-IHQ. w _i a. _i Q. i-i
Sai/iaaoMOovtaa
co a to a: 9 a coo
z*3t-az5i-z5xt-
SOt^uJOOnluo^iOUJ
-jzz-iuzaeuz-iz
S^-cocoinr-ocMiooocoin
r^r^r^oocooncncncnoo
CMCMCMCMCMCMCMCMWCMCOCO
(0(DCO(0(OCO(0(DIO(OIO(D
cMCMCMConcoco«<#««in
CO
r-
(O
CM
Q
O
--
CO
r~
O
s
_l
CONTRO
o
CO
ID
m
*
en
t-
^
en
in
CO
(0
O
0
a
IU
^^
i^
a.
CO
o
n
z
CM
CO
(0
in
in
^
r»
«t
in
,_
o
Q.
IU
^^
a.
1/1
1
00
CO
(0
in
A-61
-------
1- 10
< z
a <
at
IU 3
_l U.
5 §
co 3
°z
(/) X
t- 0
4 t"4
T Q
2
a
r- LU
co t-
2:1
u. a
oo
§5
t- a
CO O
j
a o
a o
o u
X X
X X
1 1
co co
r- t-
to to
CO M
CM CM CM
- -CM
00 CM
r-"o "
-eo o
•H
n ii ra
• at
CM 13
- L. c
'-WO
•nil
n C
(9 •*•
TJ -H CJ
C CO O)
o c ro
ao»-
a *•
u *^
u «
£
c
n
3
O
s
L
c
L
(U
>
O O
HH U
a
o
0£
c
0
(8 •?
•H U
IS •-
a (.
<^J
U)
0)
ae
. ^— ^
y oi
C *>Si*
O O)
u a
0)
Q "*s.
Ofli
Wl
-i a
>^-
t.
IS *
n —
o n
a
o
M
0)
1
CO
L
(Q J3
i. B
m
£
U
ra
co
05 en « oo to
« to in oo in
CM CM
Q O
0 0
in in «
en ^ co
t- « '-
o oo «- r- in
oo cn co oo p*
CM CM •* 10 CO
O O O O O
a o o o a
z a. a a. z
¥ ¥
§ IU <
j |U ^ |
CD ^ >-l CO
w O. _l
a a. co o a
O CO Of O
IU O l^ O IU
Z -J I 0 Z
»- CM co to in
in in in in to
CM CM CM CM CM
to to to to to
f- i- f -^ CM
*cM**incnM*toes»-oin<
r~ui'-ooooiniflcnTf^r-coen*
to
CM CM CM
a o o
O 0 0
or-o CB»-«- to in in <»>•«
**o r-incM cMt~t^ COCM
»-'-«• »-'-'- '-co
» en r- en * »-
^•looinn cn o> m i- *r to *t
^encoi**cn*'oooo^cooocn^tco
i-OOO'-'-OO'-'-OO'-'-
aooooooooooooo
aQ.Q.ZQ.Q.O.ZQ.Q.Q.ZQ.a.
¥ ^ ^
lu 3 iu < iu < ui
lU^^IU ^ -J ^IU_I it
^ 11 CO X MCO M^CO M
-IMO. M_IQ. _IQ.I-I _i a
go.coaa.ocoaocoa.aoi/)
co Q in at ? a ^ ? a T
Z»5t-Sz5t-ZU3l-ZC3
OOi-ilUOOt-iliJOi-iOUJOt-"
U-IZS-IUZZUX-IZUZ
to^-nooint^ocMiooocoinc^cM
tDt^r-t~coooo)cno)cnOOO'-
CMCMCMCMCMCMCMCMCMCMCICOCOCO
(OtO(OtO(OtO(0(0(OtOtOfOtOtO
cMCMCMOncoco**«nininin
(0
to
^
^
*"
en
in
,_
O
a.
IU
^
n
a.
1/1
o
oo
m
to
in
A-62
-------
«£
<5
a <
at
-1 U.
a.
3 a
z
co <
85
(- O
< l->
x a
a
r- ui
oo t-
£2
^ Q£
o o
_l
M
H Of
CO O
M U_
_i
a
a
u
X
X
CO
"-.
(O
a co
a -
U CM CM
?7~
0) «
- ^- tt
"-So
p. »- —
+<
CO II «
"Si
u?M
W'H
5 ** 9L
C I/I 0)
3 L
O C 19
C
I
«
L.
0)
c
M
>»
l/» >
a o
a
0
0 v
M (Q
a
§
r*in
CM^CO «CMU) »-CM*
» V h-
•w 01 n oo -" x
_l w Q. M_IO. -10.11
§a.i/iaaoi/iaoi/ia.
1/1 g M oc o ae 1/1
zxi5i-xz(5i-z5s
OO*iuiOO*iUIO*iO
U-JIZ-IOIZOXJ
(O^-coooinr-ocMCoooco
cor-r-r-oooocncneneno
CMCMCMCMCMCMCMCMPMCMCO
lOCOIOCOCDCDCOCOCOCOCO
CMCMCMCOCOCOCO*«»«'»
0
CO
p-
CM
a
o
en
o
§
^
5
_i
CO
METHOD
in
o
CO
(0
in
in
en
(D
v-
-CM
CM
en
v-
^
a
a.
CONTROL
p-
0
CO
co
in
^
q.
00
in
CM
00
CO
0
Ul
n
I/I
§
1-4
X
CM
CO
CD
in
CO
CO
p-
CM
en
CO
CD
2
Ul
V
a.
3
co
CO
CO
in
A-63
-------
(-
a 2
IX
UI 3
_l U.
za
t/l <
U M
az
1— O
< M
T* Q
9
a
P- UI
oo t-
u. Z
M
u. at
oo
-i
H4
i- a
I/) O
w U.
u.
Q
O
a
i
oo
-
p*
(O
LL V
a -
(J CO
Z M CM
i -O
oo ••- •
P- co
p- «-
O O
2 §
^
ui 5
n CD
v) a
s ^
M UI
X Z
O CM
e> en
CM CM
ID ID
CO ^
•w en « co ^ in (o
in to en in CM co CM
(O (0 p- P- p- co P"
CM
*^
a
o
"*
»- in »- "
o in co -»- - P- o
co oo in co in (0
CM en ID CM in
in co co ^ oo 10 ID
en o oo in (o (0 CM
o »- o O o O «-
20 o Q a o a
a. a. z a. a. a.
x
UI 5 UI
^ UI -I ¥ UI
11 * 00 11 *
_i a. w _i a. M
O I/) O. O O V) 0-
K u> g at 1/1
z 15 s i- z 5 *
O n o UI O Hi O
0 I -1 Z 0 I -1
10 00 CO UI P- CM 00
en en o O o — *-
CM
-------
u.
Q
U
Q.
X
1
oo
K - < ll C
u- Z ra -H
M TJ V 0)
u. a c 1/1 5)
o o 3 t-
-1 0 C «
11 O •H
i- a u ra
1/1 o -M
_i £
c
ns
15
L.
01
c
11
L
41
I/I >
O O
M U
01
at
0
c
o
« V
•H U
Si
44
in
01
a
(j!^n
82
^-^
91
Q ^s.
On
wi
-i a
L
m <*.
a 15
S
a
0)
X
L
0 t-
v 01
n n
(- e
£
U
(fl
CO
CM * co o in
co ^ ^ oo co
Is- r* r^ P* (D
CM CM
^ V,
a Q
0 0
V ID
•*• CO i-
CO CM
in co in en
I** in ^ (D t**
*- in v- en CM
O O - O O
§2221
^ ^
< IU <
_l UJ * _l
CO ^ 11 CD
M 0. _l
a a. 1/1 o a
i s s 1 i
*- CM co (o in
in in in in ID
CM CM CM CM CM
(D (0 (D (0 (0
,-,-,-,- CM
(0 M r-
,- ,J. ,-
r- co co
CN
""^
a
o
^
(O
en r-
CM CO
in
i- 00 ID
en co oo
O «- "
§22
IU
IU ^
-1 M CL
o a. i/i
DC I/I
o o u
U -1 X
S^ CO
r- r>
CM CM CM
(0 ID ID
CM CM CM
i^ooooajfcn^co^ina)
*CM*cor-in(oenincMCOcM
(DoocnooiOfOCDr^r^r^oor^
CM CM CM CM CM
x^ v^, ^^ ^^ "»^
o a a a a
0 O O 00
_j _j -j -i _j
CM CM CM ID
oo«-co in co --CMOS
CM CO » CM i- CM
CM CM *- * •*•
n m in n eo ID to oo t- in ^
incM*»-'-coweMr»enenco
OCMO'-O'-'-'-OOO'-
§2?2§i22§§22
^ ¥ ^
^t u 5 u 5 u
_JIU X -I ^IU_I ^ UJ
oa^ w 03 n^oa M^:
n _i a. _i Q. I -10.11
aoLOi/iaot/io-Qoi/io.
o i/> at o ot 1/1 o at 1/1
IszSl^S^ll^S^
UJOOnllJOi-iOllJOi-iO
s-iOxxux-isux-i
ooint~OcMioaocoint^cMoo
r>ooooenenenenOOO»-'-
CMCMCMCMCMCNCMCOCOOCOCO
(0(OIO(0(O(OIOCO(OIOIO(0
coconco«5r«
-------
1- • <
u.Z
n
u. a
00
M
t- a
I/I O
M U.
_l
a
8
1
oo
P.'
(0
a V
a
O CO
a -
Z CM CM
i • o
00 T- .
• 1 f"
f~ CM
- ^ II
(D L)
-co o
+*
co ii m
- at
CM "O
- (- C
«- 19 O
II ?*'
19 +•
TJ -H 01
§"?
O C 19
a o t-
O •H
U <9
+*
+*
c
19
O
c
l_
01
c
n
C
0)
I/I >
0 0
*i U
1
0
C '^
0 *
11 <9
or
c
o
(9 4*
M •-
a L
VI
01
a
O O)
O O)
N^
^•^
O)
a \
On
wl
-i a
^
.2
19 " CM
(O CM U) (0
o o o o
§o o a
a. a. a.
< Ul
-1 Ul ^
OB ^ i-l
i_i ^ j
O Q. VI O
g V, X
H » O Z
uiOnO
z -r z u
" CM CO ID
in in in in
CM CM CM CM
(O (D ID ID
T- »• ^ **
ID
r-
CM
0
o
ID
t*
O
O
i
i
m
METHOD
in
ID
CM
(D
CM
too
10 co o
O) _ „
n oo t
•«• (D in
<- •- CM
r* r»
»- co «
CM »- »-
a a a
a. a. a.
Ul
Ul ^
^ n
-i 11 a
So. to
i/i
£ » IS
O O M
0 -1 Z
§ — CO
r- r-
CM CM CM
ID ID ID
CM CM CM
(0
(0
r-
CM
a
o
00
o
CM
o
§
1
_l
CO
Ul
oo
CM
(0
CO
o
t-
o>
^
^
*•
(D
^
o
a
u
V
n
a
I/I
3
in
00
CM
ID
CO
0
CO
in
CM
^
en
in
O
o
a.
CONTROL
r«
CO
CM
ID
CO
O 0
co r»
CO CO
CM
a
3
^
(0
CM
en
CM CM
(D CM
o o
o a
a. Z
Ul <
X -1
M CD
i/i a
M Ul
Z Z
O CM
en en
CM CM
10 ffl
CO »
«- o o o o
en oo CM co *
en — *• »- o
CM
a
0
O in co -co
* CM ID CO
*™ CM ^ ^
CM en
in r- CD r*
*• r* oo »- oo
" O O " O
a a o a o
a a. a. z a.
Ul <
^ u _i
w ^ 03
-i a M -i
§1/1 a. a o
oo o cc
Z 5 » t- z
O M O Ul O
0 Z -I Z 0
(o oo co in t^
en en O O O
CM CM CO CO CO
ID ID ID ID ID
*t ^ ^f in in
0 0
O en
CM O
.
en ID
CM ^
r» r-
01 en
r» en
O O
a a
a. a.
Ul
X Ul
M ¥
a M
vi a.
-------
H VI
o 3
at
UJ 3
_l U.
Q.
3§
i/i <
O I/I
OZ
M
(/I X
t- 0
< 11
i°
o
!«••. m
00 h-
> <
u. Z
TING OF
R CHLORI
I/) O
M (^
_J
^
00
O •
i O
CM
CO 0
*l
II in
U. 0£
a TJ
o c. c
o m o
,,SM
« -f
TJ +• 9
C 10 O)
3 L.
o c n
to t-
I
ra
+4
>^
v
c
Qf
c
0)
£
M
to >
28
8
DC
§2
at
c
o
« -H
•H O
«j —
a L.
^
VI
01
O ^)
2 v^
0 01
u a
o nt
32
,^f
01
(Q *4»
^ •«"
Q.
O
g-
«
1/1
X
o c.
+* u
n a
c. E
?3
"
£
U
ra
aa
O CM
oo co
oo oo
8
CM
a M
0
_j
^
in
CM
CO
in
d
o
§ 2
V
2
^
_1 U
oa ^
o a
a -i a. M _i
SO. I/I Q Q. O
I/I Q I/I Of
$ * 5 f * z
8 3 £ 31 3 8
10 i- co oo in t-
CM CM CM CM CM CM
ID (D (D (D (D CO
CM CM CM CO CO CO
O
O
w
a:
CM
CM
CM
CO
in
in
O
S
u
IH
a
3
1-4
O
en
CM
(0
CO
0
en
in
at
CO
«
CO
o
§
y
2
^
-J
aa
METHOD
CM
en
CM
(D
^
o
r-
co
^
§
^.
2
CONTROL
(D
en
CM
(0
^f
0 0
in to
in *
at
co
00 CO
(O CD
CM
CO ID
0 0
_ ,_
o a
Q. 0.
UJ
* UJ
85 ft
$ *
x 3
00 CO
en o
CM CO
(D (0
* *
0 0
oo «-
CM CO
CM
•^
Q
O
_l
en
CO
CM
(O
* (O
CO O
§ 85
v
z
^
_J
ffl
METHOD
CONTROL
in r>.
O 0
CO CO
(D (O
in in
o o
in CM
ID (0
CH O£
t^
r~ CM
O *
CM
*- «
in co
7. ,_
O O
& a.
Ul
^/ Ul
M ^
i/i a.
1/1
§ >
"-" O
X -1
CM 00
^ ^
CO CO
(D (O
in in
A-67
-------
<
H- I/)
< 2
Q <
Ot
IU 3
_1 u.
l|
i/l <
(J I/I
OZ
CO X
t- o
Q
r- uj
oo 1-
> <
u.2
u. at
S5
i- a
I/I O
j
§,,
U 00
O •
i O
CM
*- II
o
co o
+J
H ra
§T3
C. C
o ra o
•fan
II C
E-H 0
I/I 0)
o c ra
Q. O t-
u ra
+>
£
c
ra
§
ra
c.
0)
c
>>
01
00 >
O 0
0)
at
o
c •-
0 -H
M (Q
DC
c
o
fl V
•M O
ra •-
a L
VI
. -—.,
00,
0 O)
••M-
Q o>
o o,
-i a
*—
0)
ra v
CM in o> in *- *
* * in co r> in
* r« »- o oo r«
co CM in «• co »-
O
a o a a a o a
0* 0* CL Z CL fl. w
^
UJ 3 IU
U ^ «J U ^
^ M at x. M
-i n a. M _i a
o a. co o a. o i/i
a 1/1 a to at
z » 5 t- * z S
O O M Ul O O w
0 -i X X -1 0 X
to «- co co in r- o
03 t» r» r» co oo en
CM CM CM CM CM CM CM
CM CM CM CO CO CO CO
O
Ol
in
at
CO
en
en
d
a
z
^
§
_i
00
METHOD
CM
en
CM
^
O
r»
^
d
CO
00
in
O
CONTROL
(0
en
CM
(0
^
O
in
in
at
6
0)
r-
00
00
^
a
IU
^
n
Q.
(/I
5
M
z
CO
en
CM
10
^
O
ID
V
o
CM
in
en
en
ID
a
IU
a!
1/1
o
CO
o
CO
(0
«•
0 0
co *-
CM CO
U.
oo
CM
CO O
en
*-
co
CO
in
CM
a o
^
3
j
00
METHOD
CONTROL
S S
C*> CO
(D 03
in in
0 0
in CM
(0 (O
ac at
d d
in t
(O *•
CM CM
co to
O 0
IU
¥ IU
M *
a. M
oo a.
l/I
§ s
Z -1
CM CO
CO CO
03 03
in in
A-68
-------
H I/)
23
a
Ul 3
-1 u.
o
2
I/) <
O Z
tH
I/I X
< 2
I S
a
fx m
cot-
2:2
si
85
M
i- a
I/) O
M Ik
X
Z
1
00
r«
(0
a m
o -
OWN
X -CX
z ••- •
a tv
-^ H
r- O
-o o
•f
- at
N- L c
*- « 0
•DM
ii C
?** 9L
^1 0)
o c m
t^
u m
£
c
a
^
A
E
01
C
X
t_
01
a o
<-i U
0)
a
0
c -
w m
a
S
•H U
ra —
a c.
M
0)
a
o«
82
O)
O--S
32
C
0)
4 "*•
W «-
o m
a
0
0)
0
14
X
o t.
a a
-i .
c
u
m
cn^^cH^incocN^in
.
rSr-r--t~air-St--t---(O
(O (0
o <- o o NO
q- q- <-(ODOMCx
^ ^
-------
t- CO
< z.
a 2
CL
UJ 3
-I U.
n
i a
(/> <
U I/I
2
cox
t- o
•C Q
2
a
r- uj
00 H
u. Z
M
U. K
OO
-1
25
I- K
CO O
w u.
§
U
X
X
1
00
f»"
(0*
a •
U CM CM
f 7 ^
OD N
- ^* M
* co o
<0.'" J
M II UJ
K
(STI
- (. C
*•
0 0
at
o
c •-
O 4-
>-l fl
a
§
fl 4^
4-« O
M ••-
a L.
VI
01
at
ua,
U Q.
N_^
""•"
O \
32
>«.
«.
01
ID (or-r-(D
^(vi^^ls.'codeoCDtfl
uicocowo^w^^w
*"C^^^"N^ (/> (/> a
^3zs3z0<3S2
QMOQMQMMOQ
-IXU-JI-IXI-JU
SCO O Is-
O) O O
C4C4C4tSCSCSCSC4CO(*)
(0(0(0(0(0(0(0(0(0(0
^^{NC^C^COO^^lfl
A-70
-------
APPENDIX B
ANALYTICAL PROTOCOL
B-l
-------
1.0 SAMPLE EXTRACTION
1.1 Extraction of Adipose Tissue
Addition of internal quantitation standards — Allow the adipose
tissue composite to reach room temperature and then add the carbon-13 internal
quantitation spiking solution such that it delivers 500 to 2,500 pg of each of
the l3C-labeled surrogates.
Add 10 ml of methylene chloride and homogenize the mixture for
approximately 1 min with a Tekmar Tissuemizer®.
Allow the mixture to separate and decant the methylene chloride
extract from the residual solid material using a disposable pipette. The
methylene chloride is eluted through a filter funnel containing a plug of
clean glass wool and 5 to 10 g of anhydrous sodium sulfate. The dried extract
is collected in a 100-mL volumetric flask.
Add to the sample a second 10-mL aliquot of methylene chloride and
homogenized for 1 min. The methylene chloride is decanted, dried, and trans-
ferred to the 100-mL volumetric flask.
Rinse culture tube with at least two additional aliquots (10 mL
each) of methylene chloride, and transfer the entire contents to the filter
funnel containing the anhydrous sodium sulfate. The filter funnel and con-
tents are rinsed with additional methylene chloride (20 to 40 mL). The total
eluant from the filter funnel is collected in the 100-mL volumetric flask.
Discard the sodium sulfate.
Adjust the final volume of the extract for each sample to 100 mL in
the volumetric flask using methylene chloride.
1.2 Lipid Determination
Preweigh a clean 1-dram glass vial to the nearest 0.0001 g using an
analytical balance tared to zero.
Accurately transfer 1.0 mL of the final extract (100 mL) to the
1-dram vial. Reduce the volume of methylene chloride from the extract using a
water bath (50-60°C) gentle stream of purified nitrogen until an oil residue
remai ns.
Accurately weigh the 1-dram vial and residue to the nearest 0.001 g
and calculate the weight of lipid present in the vial based on difference.
Nitrogen blowdown is continued until a constant weight is achieved.
Calculate the percent lipid content of the original sample to the
nearest 0.1/K as shown in Equation B-l.
B-3
-------
Lipid content, LC (%) = ULR v ,.EXT x 100* Eq. B-l
WAT x VAL
where: WLR = weight of the lipid residue to the nearest 0.0001 g;
VEXT = total volume of the extract in mi"11 iliters;
WAT = weight of the original adipose tissue composite to the
nearest 0.01 g; and
VAT = volume of the aliquot of the final extract in ml used for the
quantitative measure of the lipid residue (1.0 ml).
1.3 Extract Concentration
Quantitatively transfer the remaining extract volume (99.0 ml) to a
500-mL Erlenmeyer flask. Rinse the volumetric flask with 20 to 30 ml of addi-
tional methylene chloride to ensure quantitative transfer.
Concentrate the extract to an oily residue using rotary evaporation.
2.0 CLEANUP PROCEDURES
2.1 Bulk Lipid Removal
Add a total of 200 ml of n-hexane to the spiked lipid residue in the
500-mL Erlenmeyer flask. Slowly add, with stirring, 100 g of the 40% w/w sul-
furic acid impregnated silica gel. Stir with a magnetic stir-plate for 2 h.
Allow solids to settle and decant liquid through a powder funnel
containing 20 g of anhydrous sodium sulfate and collect in a 500-mL sample
bottle. Rinse solids with two 50-mL portions of hexane. Stir each rinse for
15 min, decant, and dry by elution through sodium sulfate combining the hexane
extracts.
After the rinses have gone through the sodium sulfate, rinse the
sodium sulfate with an additional 25 mL of hexane and combine with the hexane
extracts.
Prepare an acidic silica and a neutral alumina column as follows:
Pack a 1 cm x 10 cm chromatographic column with a glass wool plug, add
approximately 25 mL of hexane, add 1.0 g of silica gel, and then add 4.0 q of
40% w/w suIfuric acid impregnated silica gel and allow to settle. Elute the
excess hexane from the column until the solvent level reaches the top of the
chromatographic packing. Verify that the column does not contain any bubbles
or channels. Pack a second chromatographic column (1 cm x 30 cm) with a glass
wool plug, add 4 g of sodium sulfate, add 4.0 g of neutral alumina and allow
to settle, and then top with a 4-g layer of sodium sulfate. Elute the column
with 10 mL of hexane until the solvent level reaches the top of the chromato-
graphic packing. Inspect the column to ensure it is free of channels and air
bubbles.
B-4
-------
Quantitatively transfer the hexane extract from the Erlenmeyer flask
to the silica gel column reservoir. Allow the hexane extract to percolate
through the column and collect in a KD concentrator.
Complete the elution of the extract from the silica gel column with
50 ml of hexane in the KD concentrator. Concentrate the eluate to approxi-
mately 1.0 ml, using nitrogen blowdown as necessary.
Note: If the 40ft sulfuric acid/silica gel in noted to be highly
discolored throughout the length of the adsorbent bed, it is necessary to
repeat the cleaning procedure beginning with the acidic silica gel slurry
procedure.
2.2 Separation of Chemical Interferences
Transfer the concentrate (1.0 ml) to the top of the alumina
column. Rinse the K-0 concentrator with two 1.0-mL portions of hexane and
transfer the rinses to the top of the alumina column. Elute the alumina
column with 10 ml of 8% (v/v) methylene chloride in hexane until the solvent
level is just below the top of the sodium sulfate. Archive this eluate.
Columns must not be allowed to reach dryness (i.e., a solvent "head" must be
maintained).
Elute the column with 15 ml of 60% (v/v) methylene chloride in
hexane and collect this fraction containing the dioxins and furans in a 50-mL
culture tube. Concentrate the fraction to a volume of approximately 2 ml.
Prepare an AX-21/silica gel mixture by thoroughly mixing 1 g of
AX-21 (100/325 mesh) and 19 g of silica gel in a 40-mL vial. Activate at
130°C for 6 h. Store in a desiccator. Cut off a clean 5-mL disposable glass
pipet (6 to 7 mm ID) at the 4-mL mark. Insert a plug of glass wool and push
to the 2-mL mark. Add 1 g of the activated AX-21/silica gel mixture followed
by another glass wool plug. Using two glass rods, push both glass wool plugs
simultaneously towards the AX-21/silica gel mixture and gently compress the
AX-21/silica gel plug to a length of 4 cm. Preelute the column with 4 ml of
toluene followed by 2 mL of 75:20:5 methylene chloride/methanol/benzene, and
4 ml of 1:1 cyclohexane in methylene chloride. The flow rate should be less
than 0.5 mL/min. While the column is still wet, add the concentrated eluate
from the alumina column to the top of the AX-21/silica gel column. Rinse the
culture tube which contained the extract twice with 1 ml of hexane and add the
rinsates to the top of the column. Elute the column sequentially with two
0.5-mL aliquots of hexane, 10 ml of 1:1 cyclohexane in methylene chloride, and
5 ml of 75:20:5 methylene chloride/methanol/benzene. Combine and archive the
first three eluates. Turn the column upside down and elute the PCDO/PCDF
fraction with 20 ml of toluene into a 6-dram vial.
Using a stream of nitrogen, reduce the toluene volume to approxi-
mately 1 ml. Carefully transfer the concentrate into a 1-mL minivial and
reduce the volume to about 200 yL using a stream of nitrogen.
Rinse the concentrator tube with three washings using 500 uL of 1%
toluene in methylene chloride. Concentrate to 200-500 uL and add 10 uL of the
B-5
-------
tridecane solution containing the internal recovery standard and store the
sample in a refrigerator until HRGC/MS analysis.
Prior to analysis, using a gentle stream of nitrogen at room
temperature, remove toluene and methylene chloride. Submit sample to HRGC/MS
once a stable 10-uL volume of tridecane is attained.
2.3 HRGC/HRMS Analysis for PCDD/PCDF
Once initial and routine calibration criteria are met, the instru-
ment is ready for sample analysis. Prior to the first sample, an injection of
tridecane will be analyzed to document system cleanliness. If any evidence of
system contamination is found, corrective action must be taken and another
tridecane blank analyzed.
Note: Syringe Technique — Congeners of PCDD/PCDF can carry-over
between injections in the syringes used for HRGC/MS analysis unless the
syringes are properly cleaned between samples. The following procedure has
been found to be very effective for PCDD/PCDF removal from contaminated
syringes and will be used throughout these analyses.
Rinse the syringe 10 times with isooctane.
• Fill the syringe with toluene and sonicate syringe and plunger in
toluene for 5 min and repeat at least twice.
• Rinse the syringe 10 times with tridecane and pull up 1 uL of clean
tridecane.
Syringe is ready for use.
At no time should air be introduced into the HRGC column by using an air plug
in the syringe. The oxygen present in the air plug will quickly degrade a
nonbonded GC phase.
Inject a 1-yL aliquot of the extract into the GC, operated under the
conditions previously used to produce acceptable results with the performance
check solution.
Acquire SIM data according to the same acquisition and MS operating
conditions previously used to determine the relative response factors.
Instrument performance will be monitored by examining and recording
the peak areas for the recovery standard, 13C12-1,2,3,4-TCDD. If this area
should decrease to less than 50% of the calibration standard, sample analyses
will be stopped until the problem is found and corrected.
2.4 HRGC/MS Analysis for PBDD/PBDF
Procedures for the analysis of the brominated species will be
similar to the procedures for the PCDD/PCDF analysis, with the following
exceptions:
B-6
-------
1. There is no column performance window-defining mix available.
2. Sensitivity and daily RRF check will be established by injecting the
CSS level calibration standard.
SIM data will be acquired using the same chromatographic and MS
operating conditions used to determine the relative response factors. Instru-
ment performance will be monitored by examining and recording the peak areas
for the recovery standard, 13C12-l,2,3,4,7,8-HxCDD. If this area should
decrease to less than 50% of the daily calibration standard, sample analysis
will be stopped until the problem is found and corrected.
3.0 CALCULATIONS
In this section, the procedures for the data reduction are outlined
for the analysis of data from the HRGC/HRMS method for PCDD/PCDF and the
HRGC/MS method for PBOD/PBOF. Figure 8-1 presents a schematic of the qual-
itative criteria for identifying PCDDs and PCDFs. Identical calculations and
qualitative criteria will be applied to the data generated for the brominated
analogs of the dioxins and furans.
3.1 Qualitative Identification
The ion current responses for each mass for a particular analyte
must be within ±1 s to attain positive identification of that analyte. For
example, m/z 338 and m/z 340 must have maximum peak responses that are within
±1 s to be positively identified as a pentachlorodibenzofuran.
The ion current intensities for a particular PCDD/PCDF or PBDO/PBDF
must be > 2.5 times the noise level (S/N > 2.5) for positive identification of
that isomer. The integrated ion current ratios of the analytical masses for a
particular PCDD/PCOF must fall within the ranges shown in Tables B-l and B-2.
3.2 Quantitative Calculations
Relative response factors for native PCDD and PCDF analytes (RRF)
are calculated from the data obtained during the analysis of concentration
calibration solutions using the following formula:
RRF = STD ' IS Eq. B-2
AIS * USTD
where: ASTD = the sum of the areas of the integrated ion abundances for the
analyte in question. For example, for TCDD, A^jn would be
the sum of the integrated ion abundances for m/z J2o and 322;
B-7
-------
HRGC/MS-SIM Data
Quantitate Specific Isomer as per Protocol
Response to
Characteristic Molecular
Ions within the Appropriate
Homolog Retention
Window?"-
Characteristic
Ion Ratios within ±20%
Theoretical ?
Response
Corresponds-to Specific
Isomer Retention
Time?
Report Compounds cs
Not Detected (ND)
Calculate Sample LOP
Response Due to
Coextracted Interference
Quantitate Compound
as Per Protocol
Report as Isomer Unknown
Figure B-l. Qualitative criteria for identifying PCDOs and PCDFs.
B-8
-------
Table B-l. Ion Ratios for HRGC/HRMS Analysis of PCDD/PCDF
Compound
TCDF
i3Ci2-TCDF
TCDD
i3C12-TCDD
PeCDF
*3C12-PeCDF
PeCDO
i3C12-PeCDO
HxCDF
i3Cl2-HxCDF
HxCDO
i3C12-HxCDF
HpCDF
i3Cl2-HpCDF
HpCOO
i3Cl2-HpCDO
OCDF
OCDD
i3Cl2-OCDD
Ions monitored
304/306
316/318
320/322
332/334
340/342
352/354
356/358
368/370
374/376
384/386
390/392
402/404
408/410
418/420
424/426
436/438
442/444
458/460
470/472
Theoretical ratio
0.76
0.76
0.76
0.76
1.55
1.55
1.55
1.55
1.22
0.50
1.22
1.22
1.02
0.44
1.02
1.02
0.87
0.87
0.87
Acceptable range
0.61
0.61
0.61
0.61
1.24
1.24
1.24
1.24
0.93
0.40
0.98
0.98
0.82
0.35
0.82
0.82
0.70
0.70
0.70
- 0.91
- 0.91
- 0.91
- 0.91
- 1.86
- 1.86
- 1.86
- 1.86
- 1.46
- 0.60
- 1.46
- 1.46
- 1.22
- 0.53
- 1.22
- 1.22
- 1.04
- 1.04
- 1.04
B-9
-------
Table B-2. Ion Ratios for HRGC/HRMS Analysis of PBOD/PBDF
Compound
Ions monitored Theoretical ratio Acceptable range
TBDF
i3Cl2-TBDF
TBOO
i3C12-TBOD
PeBDF
i3C12-PeBOF
PeBOO
HxBOF
HxBOO
482/484
494/496
498/500
510/512
562/564
574/576
578/580
642/644
658/660
0.68
0.68
0.68
0.68
1.02
1.02
1.02
1.35
1.35
0.54 -
0.54 -
0.54 -
0.54 -
0.82 -
0.82 -
0.82 -
1.08 -
1.08 -
0.82
0.82
0.82
0.82
1.22
1.22
1.22
1.62
1.62
B-10
-------
ATP a the sum of the area of the integrated 1on abundances for the
labeled PCDO/F used as the internal quantisation standard for
the above analyte. For example, for i3C12-2,3,7,8-TCDO, ATP
would be the sum of the integrated ion abundance for m/z 332
and 334.
concentration of the analyte in pg/uL; and
Cj$ » concentration of the internal quantisation standard in pg/uL.
Table B-3 provides the pairing of target analytes to internal quantitation
standards for determining RRF values for PCDO, PCDF, PBDD, and PBDF compounds.
Relative response factors for the internal quantitation standards
(RRF TP). The RRFjs values are calculated from data obtained during the
analysis of concentration calibration solutions using the following formula.
where Aj^ and Cj^ are defined as given in Eq. B-2 and
^RS ~ concentration of the internal recovery standard in pg/wL; and
ARp * the sum of the areas of the integrated ion abundances for the
labeled PCDD (i3C12-l,2,3,4-TCDD or 13C12-1, 2,3, 7,8,9-
HxCDO). For example, the i3Cl2-l,2,3,4~TCDD, ARS would be the
sum of the integrated ion abundance for m/z 332 and 334.
The RRF values for the 13C12-PBDD/PBDF compounds will be calculated using a
similar equation. Refer to Table B-l for pairing of the internal quantitation
standards with the appropriate internal recovery standard.
3.3 Concentrations of Sample Components
Figure B-2 presents a schematic for quantitation of PCDD and PCDFs
which meet the criteria specific in Section 3. Calculate the concentration of
PCOO/Fs or PBDD/Fs in sample extracts using the formula:
Eq. B.
where: Csam_ie = the Tipid adjusted concentration of PCDD or PCDF congener
in pg/g;
A-ami_ » sum of the integrated ion abundances determined for the
p PCDD/PCDF in question;
B-ll
ample AJS • RRF • WAJ • LC
-------
Table B-3. Target Analyte/Internal Quantitation Standard and Internal
Quantitation Standard/Internal Recovery Standard Pairs
Target analyte
Internal standards
Quantitation
Recovery
Chlorinated
2,3,7,8-TCOO
2,3,7,8-TCDF
1,2,3,7,8-PeCDF
2,3,4,7,8-PeCDF
1,2,3,7,8-PeCDD
1,2,3,4,7,8-HxCDF
1,2,3,6,7,8-HxCDF
2,3,4,6,7,8-HxCDF
1,2,3,7,8,9-HxCDF
1,2,3,4,7,8-HxCDD
1,2,3,6,7,8-HxCDD
1,2,3,7,8,9-HxCOO
1,2,3,4,6,7,8-HpCDF
1,2,3,4,7,8,9-HpCDF
1,2,3,4,6,7,8-HpCDO
OCOF
OCDD
Brominated
2,3,7,8-TBDD
2,3,7,8-TBDF
1,2,3,7,8-PeBOO
1,2,3,7,8-PeBDF
1,2,3,4,7,8-HxBDO
1,2,3,4,7,8-HxBDF
13C12-
13C12-
-2,3,7,8-TCDD
2,3,7,8-TCDF
1,2,3,7,8-PeCDF
1,2,3,7,8-PeCDF
1,2,3,7,8-PeCDO
1,2,3,4,7,8-HxCDF
1,2,3,4,7,8-HxCDF
1,2,3,4,7,8-HxCDF
1,2,3,4,7,8-HxCDF
1,2,3,6,7,8-HxCDD
1,2,3,6,7,8-HxCDO
1,2,3,6,7,8-HxCDO
1,2,3,4,6,7,8-HpCDF
1,2,3,4,6,7,8-HpCDF
1,2,3,4,6,7,8-HpCDD
OCDD
-OCDO
i3Cl2-2,3,7,8-TBDD
i3C12-2,3,7,8-TBDF
i3Cl2-l,2,3,7,8-PeBOF
»3C12-l,2,3,7,8-PeBDF
»3Cl2-l,2,3,7,8-PeBOF
i3Cl2-l,2,3,7,8-PeBOF
13C12-
13C12-1,2,3,4-TCDD
i3C12-l,2,3,4-TCDO
i3Cia-l,2,3,4-TCDD
»3C12-1,2,3,4-TCDD
i3d 2-l,2,3,4-TCDD
i3C12-l,2,3,7,8,9-HxCDD
z-l,2,3,7,8,9-HxCDD
:12-l,2,3,7,8,9-HxCDD
z-l,2,3,7,8,9-HxCDO
13C12-1,2,3,7,8,9-HxCDO
i3C12-l,2,3,7,8,9-HxCDO
»3C12-l,2,3,7,8,9-HxCDD
:I2-1,2,3,7,8,9-HxCDO
2-1,2,3,7,8,9-HxCDD
:12-l,2,3,7,8,9-HxCDD
i3C12-l,2,3,7,8,9-HxCDD
2-l,2,3,7,8,9-HxCDO
i3C12-l,2,3,7,8,9-HxCOO
i3Cia-l,2,3,7,8,9-HxCDD
i3C12-l,2,3,7,8,9-HxCDD
i3C12-l,2,3,7,8,9-HxCDO
i3C12-l,2,3,7,8,9-HxCDO
i3C12-l,2,3,7,8,9-HxCDO
B-12
-------
QUANTUM ION
Report as Not Detected
Calculate Sample LOD
HRGC/MS-SIMData
Response
Meets AI
Qualitative
Criteria •?
Response
times
S/N?
Calculate as per Protocol
Report as Trace (tr) Value
Quantitate as per Protocol
Report as Positive Quantifiable Value
Figure 8-2.
Procedure for quantitation of PCDDs and PCDFs
in human adipose tissue.
B-13
-------
sum of the integrated ion abundances determined for the
labeled PCDO/F used as the internal quantitation standard
for the above analyte;
the amount (total pg) of the labeled internal quantitation
standard added to the sample prior to extraction;
RRF = relative response factor of the above analyte relative to
its labeled internal quantitation standard determined from
the initial triplicate calibration;
WAT = weight (g) of original adipose tissue sample; and
LC = percent extractable lipid determined from Eq. 4-2.
Refer to Table 8-1 for pairing of target analytes with the appropriate
internal quantitation standard.
Quantitative data will be classified to indicate the intensity of
the signal response. Qualifiers will include: not detected, NO (signal-to-
noise ratio is less than 2.5); trace, TR (signal-to-noise ratio is greater
than or equal to 2.5 but less than 10); and positive quantifiable, PQ (signal-
to-noise ratio is greater than or equal to 10).
Because of the lack of analytical standards to demonstrate isomer
specificity of the brominated compound, all PBOD and PBDF responses that elute
at a retention time of a 2,3,7,8-substituted isoraer will be classified as a
maximum possible concentration (MPC) for that specific isomer.
Recovery of internal quantitation standards. Calculate the recovery
of the labeled internal quantitation standards measured in the final extract
using the formula:
Internal Quant. Std. _ IS * ^RS . 1QQ Eq. B-5
Percent Recovery ~ ARS • QIS • RRFJS
where: Aj$ = sum of the integrated ion abundances determined for the
labeled PCDD/PCDF internal quantitation standard in question;
AR5 = sum of the integrated ion abundances determined for m/z 332
and m/z 334 of 13C12-1,2,3,4-TCDD or m/z 390 and m/z 392 of
i3C12-l,2,3,7,8,9-HxCOD (recovery standards);
= amount (pg) of the respective recovery standard, added to the
final extract;
= amount (pg) of the labeled internal quantitation standard
added to the sample prior to extraction; and
B-14
-------
• . If a response for a specific dioxin or furan congener is qualified
as a trace, TR, value (signal-to-noise ratio is greater than or equal to 2.5
but less than 10), the analyst will also provide an estimated method detection
limit. This is accomplished by using the observed signal-to-noise ratio on
either side of the response and calculating as given in Eq. B-6.
B-15
-------
APPENDIX C
METHOD FOR ESTIMATING MEASURES OF UNCERTAINTY
C-l
-------
The additive model for analyzing NHATS composite sample data has been presented and
discussed in Chapter 7. This model assumes that the concentration for an individual specimen
is equal to a linear combination of fixed effects (census region, age group, sex, and race group
effects) plus the sum of two independent random effects. The random effects are the effect due
to random selection of the MSA and the effect due to random selection of the donor from that
MSA. The MSA effect is assumed to have variance o^, and the donor effect has variance a\. In
addition, random measurement error in the observed concentration for an analyzed composite
sample is also present and assumed to have variance oj. This appendix discusses how these three
measures of uncertainty are estimated and interpreted.
The estimate of o* is obtained from the analysis of control QC samples, detailed in Chapter
6. Control samples were unspiked samples taken from a homogenized bulk sample of human
adipose tissue. As a result, the variability in the predicted concentration for unspiked control
samples reflects measurement error exclusively. The estimate of or was obtained from Table 6-3
by multiplying the predicted concentration by the coefficient of variation (as a proportion) for the
control samples. The result is equal to the estimated standard deviation of the measured
concentration in the control samples. The control sample concentration was nearly equal to the
average concentration of the study samples. The estimate of or is given for each of the
compounds in Table C-l, along with the coefficient of variation (in percentage terms) relative to
the estimated national average concentration.
As discussed in Chapter 7, estimating the additive model parameters in the
prediction process requires an iterative weighted generalized least squares procedure. The
procedure is iterative and weighted in nature because the covariance matrix of measured
composite concentrations, used in estimating the fixed effect parameters, depends on the three
unknown variance terms given above and on the unknown fixed parameters themselves. Starting
estimates for these parameters, used to begin the iterative estimation process, are obtained from
a mixed model analysis which assumes constant error variances. This approximate model is fitted
using the P3V procedure in the BMDP statistical software package. The P3V procedure gives
starting values for the fixed effect parameters and estimates for two surrogate variance
components. One of the surrogate variance components is due to random selection of MSAs and
the other is due to the combined effects of random donor sampling and measurement error.
C-3
-------
Table C-l. Estimates of Measurement and Sampling Standard Deviation
(in pg/g), and Coefficients of Variation (in percent), for
the PCDDs and PCDFs Analyzed in the FY87 NHATS
Standard deviation and CV (%) due to ...
Measurement
error
Analyte
2,3,7,8-TCDD
1,2,3,7,8-PeCDD
1,2,3,4,7,8/6,7,8-HxCDD
1,2,3,7,8,9-HxCDD
1,2,3,4,6,7,8-HpCDD
OCDD
2,3,7,8-TCDF
2,3,4,7,8-PeCDF
1,2,3,6,7,8-HxCDF
°,
1.77
1.84
15.8
2.78
9.52
57.5
0.613
4.42
4.49
(CVt)'
(33)
(17)
(21)
(24)
(9)
(8)
(33)
(46)
(78)
Sampling
MSAs
om
0.0
0.0
0.0
0.0
14.7
0.0
0.346
0.0
0.0
(CVJ
(0)
(0)
(0)
(0)
(13)
(0)
(18)
(0)
(0)
Sampling
individuals
o€
0.0
2.93
0.0
0.0
57.3
571
0.0
0.0
0.0
(CVJ
(0)
(54)
(0)
(0)
(52)
(79)
(0)
(0)
(0)
a CVs (coefficients of variation) are with respect to the estimated national average
concentration.
C-4
-------
Orban and Lordo (1989) show that the surrogate variance components are linearly related to the
variance components o2., o^, and o«. The variance component estimates are obtained by solving
a set of simultaneous linear equations.
Table C-l lists the estimated standard deviations and the corresponding coefficients of
variation (percent of national average) for each of the three random effects. For five of the nine
chemicals the sampling standard deviations were estimated to be zero. The formulas for
estimating variance components can produce negative estimates. These are replaced by zero.
This occurs frequently when measurement variation is large relative to the amount of sampling
variation that affects the composite concentrations. However, it does not mean that there is no
sampling variation. In fact, of the four chemicals with positive estimates of sampling standard
deviations, the sampling errors were found to be quite large.
Two chemicals (1,2,3,4,6,7,8-HpCDD and 2,3,7,8-TCDF) had positive estimates of standard
deviations due to random selection of MSAs. The estimates were 13 and 18 percent of their
respective national average concentrations. Also, three chemicals (1,2,3,7,8-PeCDD, 1,2,3,4,6,7,8--
HpCDD, and OCDD) had positive estimates of standard deviations due to random sampling of
individuals. These estimates were between 52 and 79 percent of the national average
concentrations.
The total sampling standard deviation due to both sampling of MSAs and individuals can
be estimated by
For example, the total sampling standard deviation for 1,2,3,4,6,7,8-HpCDD is estimated to be
59.2 = [(14.7)2 + (57.3)2f .
On a relative basis this is 54% of the national average concentration of 1,2,3,4,6,7,8-HpCDD.
The statistical uncertainty of the variance component estimates is difficult to establish
C-5
-------
because of the approximate nature of the estimation procedure and the relatively small number
of composite samples analyzed. Also, the priorities for creating the sampling and compositing
designs were aimed primarily at producing accurate estimates of the national average
concentrations and determining if there are significant demographic and geographic effects on
the average concentrations. To obtain more precise estimates of sampling standard deviations it
would be necessary to increase the number of composites or individual specimens that are
analyzed and to revise the compositing design objectives.
C-6
-------
APPENDIX D
PLOTS OF ESTIMATED CONCENTRATIONS VERSUS
SPIKED LEVEL WITH TOLERANCE BOUNDS FOR
FY87 NHATS QC SAMPLES
D-l
-------
i- O
IO
o
CM
_ O
J- o
8588?
(CyDd) uofjBJiueouoo
0)
D-3
-------
-8
10
_ O
J- o
g 9 8
O o O
CM T-
03
SJ
to
V-l •
o
CJ (0
£H CO
9 oo
O SH
O ^
U O
O CO
*w -O
C
C 9
o o
•H ffl
4J
(0 0)
^ U
4J 0
C 10
a> M
u a)
a rH
o o
U EH
T3 43
4) 4J
4J-H
•H iH
jj 0}
(0 >
O tJ
(U
-p ^
O-H
I
a
0)
^
3
D-4
-------
o
10
5
•5.
(0
_ O
- O
8 S ? 8 8 S
01
S
CD
(Q
0)
eu e
00 CO
Jfi
a, r-
3 00
o >>
u o
u
O CO
a z
o o
•H CQ
.M
(0 0)
H °
C 10
0) ^
U 0)
fi ^H
o o
U EH
•H <-4
4J 0)
(0 >
U
-------
u O
in
O
CM
_ O
I- O
8
OO
(CyOd) uogaqueouoo
I
Q
05
&
•H
D-6
-------
^ o
in
-8
_ o
- o
(D/Cd)
u
IT)
Q
2
D-7
-------
1-8
-8
-S
_ o
-8
-g
-3
i I I I I
_ O
h O
OO
O O O
I
e
-------
o o
CM 21
OOOOOOO
o
CM
T—
o
T—
8
8
- P
i i i
-8
-8
_ O
- O
to
9
(0
U CO
x a)
KrH
i a
00 §
•• (0
r- w
-to
9 eo
O >*
ab
o ^
u o
o to
•w-O
G
a a
o o
•H CQ
4J
0 0)
>-( O
4J (3
C (0
(U M
U 0)
4) -P
4J-H
1 *
•H i— I
4J 0)
03 >
Ed d)
o-d
a)
•P ^
O-H
PH CU
04 (O
I
Q
O
M
S
t^
•H
fc,
(C/Od) uopaaueouoo
D-9
-------
,_ o
h CM
-8
-8
-g
_ o
-g
I I T 1
_ O
h O
(0
3
CO
Q •
U CO
x 0)
SS^H
i a
oo co
PO
•.CO
™ 61
•0
G
co
o M
u o
M-l
^
o (o
M-4 -O
C
a 3
o o
•H 03
JJ
(0 0)
M U
*J fi
a A
0) M
U d)
C ^H
o o
U E-i
0)
4-> (U
u >
u a>
J
M-l
O T3
0)
4J ^
O -H
M a
& to
00
I
a
0)
M
(B/Od)
D-10
-------
CO
8
CM
-8
- O
I
CM
8
(B/Cd)
CO
%
CN
I
Q
9)
^
3
Oi
•H
&4
D-ll
-------
-8
§
^ o
I I i
1 1 I I I 1
o
CO
o
CM
o
10
_ o
CO
_ o
CM
L o
L o
OOO
OO
OO
0]
3
00
a
a •
u w
X flj
i
a,
6
ao en
en
3 00
O !*
I*
U O
4-1
O CO
c
C 3
O O
•H
-------
-8
-8
-8
8
9
8
8
O
I—
- O
8
CO
CD
Q e
U H
*frl
H
M
T3 O
C3 >H
3
O CO
O O
O (V
10 d)
JJ J
to
•H 0)
4J .M
CO-H
Ed CU
O CO
•P CO
I
O
0)
01
•H
(CyOd)
D-13
-------
o
CO
O
O
-8
O
CO
-8
a
10
i I i
OQOQOOO
I I I
O O O
O
CM
h- O
h O
CO
V
r4
fa a,
89
Q-OT
33
o> O
^
co en
•* a
^
ro t*.
•O o
C 4-1
S
O CO
a-o
g a
o P
o o
m o
c
O ^
•H (!)
10 O
tt £j
-P
c x:
flj 4J
O-H
0) 0)
-P J
(0
ST3
•H 0)
-P ^
CO -H
H Cb
O ca
S3
4J CO
O M
rH 0)
ft. >
I
a
(0/t3d) UOflRQUCOUOQ
D-14
-------
-8
-8
o
T-
-8
-8
_ O
^ co
__ o
CM
_ O
k O
CO
0)
rH
P Oi
o e
u 3
Q-W
s
o O
W
vo
(N
•o o
c IN
3
O CO
CXI?
g G
O 3
O O
t-l U
C (0
O M
•H 0)
•P -H
(0 O
0) -P
O-H
C »
O
U r-t
0)
•o >
V 0)
JJ J
(0
•H O
-p >J
CO -H
w a
O co
53
4-) CO
O M
CO
rH
Q
0)
CM
(CyOd)
D-15
-------
\- o
8
CO
w
fa^<
Q O4
O RJ
I CO
91
-u
oo o
fc
r* w
ȣH
vo rtj
*, jg
•* a
%
co r»
*oo
4) 0)
•P J
05
e-o
•H 0)
•P A:
CO -H
u a.
CO
IM
o ca
3
P CO
o ^
^H 0)
A. >
I
O
0)
(CyOd) UOQ&QUOOUOO
D-16
-------
CO
5
£
CM
- o
Q a
88
I W
00
w
-oo
CN (H
•O O
a 4-1
a
o co
ftTJ
g C
O 3
U O
03
M
O 0)
*4-l U
C
C (0
O M
•H fl)
•P M
10 O
M Ei
J->
a js
(U -P
O-H
U ^H
4)
•d >
0) 0)
4-> >4
(0
W-H
U (X
(A
4-1
O CO
3
4J CO
O M
rH O
ft >
m
iH
Q
0)
D-17
-------
,_ o
h co
^ o
h CM
-8
§
(-8
8
O
o
CO
_ O
}- o
8
CM
8
vo
iH
I
Q
0-18
-------
Nr- "•>
1 1 I I I I I I T I
OO
O Q O
OO
,_ o
h CM
-8
-8
S f
8 |
*
s
-3
~ CO
o
CM
_ O
1- o
0)
0)
co
i
00 (0
rn
-
cs
oo
TJ O
C 4-1
3
O CO
O 3
U O
a
o a)
M-I u
c
ti (0
O M
•H
4) 0}
(0
e-o
•H 0)
^j >^>
05 -H
Ed CU
M-I
O CO
-P CO
O M
rH d)
a. >
I
a
&
•H
(CyDd)
D-19
-------
APPENDIX E
SUPPLEMENTARY DESCRIPTIVE STATISTICS
FOR FY87 NHATS PCDDs AND PCDFs
E-l
-------
SUPPLEMENTARY DESCRIPTIVE STATISTICS FOR
FY87 NHATS PCDDs and PCDFs
Tables E-l and E-2 give the average measured concentration, standard error, and number
of unrestricted measurements for each of the marginal populations defined by the different levels
of the four analysis factors. Composites classified as mixed race group or mixed sex group are
composites which contain specimens from both white and non-white donors or both male and
female donors, respectively. All specimens in each composite are from the same census region
and same age group. For example, there were 36 unrestricted measurements of 2,3,7,8-TCDD and
the average concentration was 6.60 pg/g. The average was 1.74 pg/g for the five composites
containing specimens from the youngest age group (0 to 14 years), 4.33 pg/g from the 12
composites in the middle age group (15 to 44 years), and 9.31 from the 19 composites in the oldest
age group (45+ years). The standard errors are all less than 0.6 pg/g. This suggests that the
differences between the age group averages are statistically significant. However, the reported
significance levels in Chapter 8 were determined by a more rigorous statistical approach.
E-3
-------
cS
Q
CH
H
1
B
u.
u
C
_o
"•§
•3
8.
3
u
'£
a.
^
J
1"
J?
o
•'r
g_
|
U
"o
&
e
a
Z
TJ
C
a
1
t
w
T3
E
(0
TJ
1
cn
(X
S
00
jj
'2
§.
U
2
01
^
f2
D
3
§
S
O
3
§
8
OS
"c
ra
V
^
6
Si
§
1
a:
(A
3
w
US
2i
a
w
-Si
§
u.
1
S
u,
m
5
'r^
2:
c
o
Z
"8
X
S
Ic
^
+
in
^
(2
S
o
5
en
U
M
§
•x:
g
i
A3
l-H
»
5
^
1
SS
m
vO
•*
m
C>
S
S
a
8
T*
a
f
8.
Population
tx in m m \c
"& (N CT* so oo m CN CN fx r** O J""1 O
oo'r-Ht-H T* rj oc -^ <-< -^ Lrir-Jfo tx •* m 1Q £3 ^ •* iri oo 0 ro oo
" O* ~i
„
H fx S^ •* ffi odo Sr"J5 S^m ^^m rq^m SrSS 2mS ^^S
^- 00
85 ,n35 ^^ ^5 « *?{q osr.
in^o gr-it^ StN^ S'^'* ocn1" 33"* °°lrir' ^^^
^
m^ «S ^i moo c$ oo OSM S
O\OCT\ oo'o'ift *™iirioo txo'tx inxcoo o»-5in ^otritx *— i so tx
ri«-iTHrji-ii-<»-iJsi-iOJ\C«-ii-i r-<
rq ts ao "^
m cc oo o oo 91 m oo tx op
»M rttX'-lr*r-l^i^fSm ^
S! I 3^ U2 £ 88 oo K Os W S 000
r-! o in m o GO ^o r4 o ^ ^ tx ^o ^ o tx co oo c4 o ^ ^* -^ •»*
« ^ T*I r-t ^ i-< (SI
M in tx o o m
~ °" ^
oSS v^S CoR ^^ ^H SR v^M
\O'~''~' i-iC-lCJ mCh** CN<~<(N O-^'-O v^r^O ^O*O O^^*O
i-t tx r-.
oo co in in tx
^Mr, M^^
cs m ^* o r-t S
inp ^ ^c in oo •** tx tx coin -^ ^
ir^ rn Q\ iQ t*^ oo in in tx o ^ o^ *~* o^ tx m o tx co o r^ o m M
TH tx ^ ^^ m ^ tx tx CT\ r^
r^ S. *-*
o ^ ^ oo o in c^ f^i
\do-o i-Jt-iin ^*cri r4oos ONOOJ o^ tx^ rC cos cov
tx co m c^ m m ro m m
ff? of Oj i-? Of Of M Of Of
vi
3
tn
S
u
T-* ^.
C |
0 %
1 "BO ^
if I
11 I
g iN
•1 |1i
Illl
(0 £» V T3
E-4
-------
&
3
a
1
CL
3
S
?e group (years)
<
_§
c2
3
c
CJ
.§
UJ
i
£
i
*
^
s
i
c
o
Z
•8
.2
IS
+
S
,
1-H
2
.,
f.
2
w
2
.2
c
c
^
in
•
c*
•*
^
'
CO
s
*
Ov
m
e3
S
S
1— «
"U
£
jjp
c
s
01
CL
c
o
"ra
O
G-
to CO ^ r- i-H ON
r-i o' oo o' o co in (*i *•*
r-*OCO O O •**« OO «-H M
vot.
i—> Csl CO O tN ON
MOM <=»<=> so £ ^ £
i i 0 I i O i i O
o^o
*"•• O so O O IX, fsl f— ' CO
CS CO TH CO
ss si «R
csio'tx o o so o ri so
in °l 5 0 IN SI
CNJ o in o o so IN »-H oo
m o M o c-» in
O CO rsl ON
o w ?5 ^ f^ so
CSJO'ON OO*~< MOC^
3$ ^ S&
HC;VO ""^ ^"^
sO OO OO ON C3
NO *-< CO O t*» C>
** ** ** ~
M SO tV
co co co o *r ^
csooo oo^f cocsico
ON CN] CS O C^ CO
i-iOOO OOOO ^COtv.
8S Ss ooK)
CO Tf i— CO
CO
IJj £Ij
Q Q
s y y
y ^ ^
E7 oo oo
IN^ CO" •**•
r"T r-f co"
fst i-? c*T
o in
t-I <>J CO oo' »-* t-< O O' CO
OO T-H < m
o sr> m so co
ixcoco ifi^Hso csio'rj
N g en iS S co
fsltsco ooes ootx
i-i OO vO CO CsJ O
<-HitM rsOO OOOO
co co m co
•* CO tX CO tx
O CO CO U"i r* sO O ' *-H
s s2 SS
F-
rsi m
tx fsi r-i so rH co o ^ in
M
St-i CO OO t> CO
i-H in oo r-j m
tv co co in ^H vo ^
rn t^. in
O OO CO r-< O in
CM •^ tx sO t-x 1-"
CO 1-H
tU El* tU
D D Q
XXX
°°k °°v °^.
c c 2
o »* ^ o
c2 < Co H
« J3 w -O
E-5
-------
50272-101
REPORT DOCUMENTATION 1- REPORT NO- 2.
PAGE EPA 560/5-91 -003
4. Title and Subtitle
Chlorinated dioxins and furans in the general United States.
Population: NHATS FY87 results
'. Author(s)
U.S. Stanley (1) and J. Orban (2)
). Performing Organization Name and Address
1) Midwest Research Institute, 425 Volker Boulevard, Kansas City, Missouri 64110
2) Battelle Columbus Division, 505 King Avenue, Columbus, Ohio 43201
2. Sponsoring Organization Name and Address
Exposure Evaluation Division TS-798, Office of Toxic Substances
U.S. Environmental Protection Agency, 401 M Street, S.W., Washington, DC 20460
J. Remmers, J. Schwemberger
3. Recipient's Accession No.
5. Report Date
November 26, 1 991
6.
8. Performing Organization RepL No.
10. Projectflask/Work Unit No.
9801 -A(27)
11. Contract(C) or Grant(G) No.
(C) 68-DO-0137, WA27
68-02-4292
(G)
13. Type of Report & Period Covered
Final Report
14.
5. Supplementary Notes
6. Abstract (Limit: 200 words)
Population estimates of the average levels of polychlorinated dibenzo-p-dioxins (dioxins or PCDD) and polychlorinated
dibenzofurans (furans or PCDF) were established using 865 human adipose tissue specimens collected in Fiscal Year
1987 through the U.S. Environmental Protection Agency's National Human Adipose Tissue Survey (NHATS). The
specimens were composited into 48 unique samples prior to chemical analysis. Estimates of the national average
concentrations of levels were made among subpopulations defined by the donors' geographic adipose tissue of the
U.S. population was 5.38 pg/g (±0.32); however, the levels increase from 1.98 pg/g (±0.81) in children under 14
years of age to 9.40 pg/g (±0.38) in adults over 45 years old. Significant age effects were determined for all nine of
the compounds which were present at quantifiable levels in greater than 90% of all samples. Statistically significant
differences based on geographic regions were determined for estimated levels of 2,3,4,7,8-PCCDF with the highest
levels in the northeast and the lowest levels in the west. There were no significant differences in the estimated levels
from different sexes or race groups for any of the target analytes.
7. Document Analysis a. Descriptors
Polychlorinated dibenzo-p-dioxins, polychlorinated dibenzofurans, PCDD, PCDF, 2,3,7,8-TCDD, Human Adipose
Tissue, National Human Adipose Tissue Survey (NHATS), Fiscal Year 1987, FY 1987.
b. Identifiers/Open-Ended Terms
Determination, analysis, general United States population, body burden, weighted national averages.
c. COSATI Field/Group
8. Availability Statement
19. Security Class (This Report)
20. Security Class (This Page)
21. No. of Pages
277
22. Price
;ee ANSI-Z39.18)
See Instructions on Reverse
OPTIONAL FORM 272 (4-77)
(Formerly NTIS-35)
Department of Commerce
------- |