United States
Environmental Protection
Agency
Toxic Substances
Washington DC 20460
EPA-560/13-79-014
December, 1979
Toxic Substances
-------
ANALYSIS OF EPA PESTICIDE MONITORING NETWORKS
by
Tyler Hartwell
Philip Piserchia
S. B. White
Nancy Gustafson
Albert Sherdon
Robert Lucas
Donna Lucas
David Myers
James Batts
Robert Handy
Steve Williams
Project Officer
Dr. John Smith
Office of Toxic Substances
Washington, D.C. 20460
U. S. ENVIRONMENTAL PROTECTION AGENCY
OFFICE OF TOXIC SUBSTANCES
WASHINGTON, D.C. 20460
-------
ii
DISCLAIMER
This report has been reviewed by the Surveillance and Analysis
Division of the Office of Toxic Substances, U. S. Environmental Protection
Agency, and approved for publication. Approval does not signify that
the contents necessarily reflect the views and policies of the U. S.
Environmental Protection Agency.
-------
iii
ABSTRACT
This report describes a brief investigation of six pesticide
networks run by EPA. In this investigation, an attempt was made to
answer several specific questions about each network including (i) the
sampling procedures used, (ii) selected summary statistics and tests of
significance appropriate for summarizing the data, and (iii) an exam-
ination of trends over time and differences in geographic areas for
particular pesticides.
-------
iv
ACKNOWLEDGMENTS
The authors would like to thank Frederick Kutz, Sandy Strassman,
Ann Carey, Thomas Dixon, and Henry Yank for their helpful comments in
preparing this report. The authors also greatly appreciate the con-
siderable effort by Ms. Carol Mitchell in typing this report under
severe time constraints.
-------
TABLE OF CONTENTS
1. INTRODUCTION 1
2. DISCUSSION OF QUESTIONS PERTINENT TO ALL NETWORKS,
Summary Statistics 6
Tests of Significance 8
Monitoring Procedures 11
Matching of Pesticide Networks 14
Practical versus Statistical Significance 16
3. NATIONAL HUMAN ADIPOSE TISSUE NETWORK 18
3.1 Detailed Description of the Network 18
3.1.1 Sampling Procedures 18.'
3.1.2 Data Available on Data File 23
3.1.3 Assumptions Made in Analysis of the Network.... 24
3.2 Data Analysis 26
3.2.1 Overview 26
3.2.2 Tests 27
3.2.3 Summary Statistics 28
3.2.4 Analysis Results 28
3.2.5 Final Remarks 30
4. NATIONAL SOIL RESIDUE NETWORK 43
4.1 Detailed Description of Network 43
4.1.1 Sampling Procedures 43 '
4.1.2 Data Analysis 46
References 54
5. NATIONAL SURFACE WATER AND SEDIMENT RESIDUE NETWORK 55
5.1 Description of the Network 55
5.1.1 Sampling Procedures 55
5.1.2 Data Available On Data File 56
5.1.3 Assumptions Made in Analysis of the Network.... 57
5.2 Data Analysis 57
5.2.1 Summary Statistics 57
5.2.2 Tests for Trends Over Time 58
5.2.3 Observed Trends 59
5.2.4 Geographic Differences 59
References 68
6. AIR NETWORK 69
6.1 Detailed Description of Network 69
6.1.1 Sampling Procedures 69
6.1.2 Data Available on Date File 70
6.1.3 Assumptions Made in Analysis of the Network.... 71
-------
vi
TABLE OF CONTENTS (cont'd) p Q
rage
6.2 Data Analysis 72
6.2.1 Summary Statistics 72
6.2.2 Tests 72
6.2.3 Trends 72
6.2.4 Geographic Differences 72
References 78
7 . NATIONAL SOIL APPLICATION NETWORK 79
7.1 Detailed Description of Network. 79
7.1.1 Sampling Procedures 79
7.1.2 Data Analysis 79
References 83
8 . NATIONAL URBAN SOIL NETWORK 84
8.1 Detailed Description of Network 84
8.2 Data Analysis 86
9. GENERAL ACCURACY OF PESTICIDE CHEMICAL ANALYSIS 98
10. RECOMMENDATIONS AND SUMMARY 100
APPENDIX 1: Data Collection Forms
APPENDIX 2: (see Addendum)
APPENDIX 3.1: Census Breakdown of the United States
APPENDIX 3.2: Materials for Selecting Sample Sites for Middle
Atlantic Census Division for FY 1973
APPENDIX 3.3: Originally Selected Cities FY70, FY73, FY77, and
Selected Cities Including Alternates FY73-FY78
APPENDIX 3.4: Survey and Site Quotas for FY73-FY76
APPENDIX 3.5: Survey and Site Quotas for FY7277+
APPENDIX 3.6: Copy of "Guidelines and General Information About
Collecting Adipose Tissue for the National Human
Monitoring Program for Pesticides"
-------
Vll
LIST OF TABLES
Number Title . Page
3.la Sample Allocations for Each Stratum FY72-FY78 19
3.1b Breakdown of Data by Design and Supplementary Files 19
3.2 Summary Statistics for Rarely Observed (« 1%) 33
Residues 33
3.3 Summary Statistics for Uncommonly Observed (« 10%)
Residues 33
3.4 Summary Statistics for PCB 34
3.5 Summary Statistics for Total DDT Equivalent - Total
DDT Equivalent = [pp'DDR+opfDDT+1.114(pp'DDE+
op'DDE+pp'DDD)] 35
3.6 Summary Statistics for 3-BHC 36
3.7 Summary Statistics for Dieldrin 37
3.8 Summary Statistics for Heptachlor Epoxide 38
3.9 Summary Statistics for Oxychlordane 39
3.10 Summary Statistics for Trans Nonachlor 40
3.11 Summary Statistics for Hexachlorobenzene 41
4.1 Summary Statistics and Tests of Significance 50 -
5.1 Periods for Which Data is Missing 57
5.2 Summary Statistics by Compound by Year and Season -
Water - Over All Sampling Sites 60
5.3 Summary Statistics by Compound by Year and Season -
Sediment - Over All Sampling Sites 61
5.4 Tests for Trends Over Time - Water and Sediment.... 62
5.5 Summary Statistics by Compound by Water Basin -
Water (1976 - 1979 Combined) 63
5.6 Summary Statistics by Compound by Water Basin -
Sediment - (1976 - 1979 combined) 64
6.1 Periods of Air Sampling by Site 71
-------
viii
6.2 Summary Statistics - Pesticides in Air - April-June
1975 73
6.3 Summary Statistics - Pesticides in Air - August
1975 - March 1976 74
6.4 Summary Statistics - Pesticides in Air - November
1976 - September 1977 75
6.5 Summary Statistics - Pesticides in Air - 1978 76
7.1 Number of Sites With Wheat and Pesticides Applied
to at Least Five Sites in One or More Years 81
7.2 Number of Sites With Soybeans and Pesticides
Applied to at Least Ten Sites in One or More Years. 81
7.3 Number of Sites With Field Corn and Pesticides
Applied to at Least Ten Sites in One or More Years. 82
8.1 Number of Positive Detections Found in Urban
Soil by Year by Pesticide 87
8.2 Number and Percent of Positive Detections of Major
Pesticides Found in Urban Soil During 1971-1977,
by City by Year 88
8.3 Number and Percentage of Positive Detections Found
in Urban Soil Samples from Ten Cities Sampled in
1971 and Again in 1977 91
8.4 Geometric Means Based on Samples Selected From Same
Locations in 1971 and 1977 92
8.5 Tests of Significance - J^DDT 94
8.6 Summary Statistics for J^DDT 95
8.7 Tests of Significance - Chlordane 96
8.8 Summary Statistics for Chlordane 97
-------
IX
LIST OF FIGURES
Number Title Page
3.1 Percent Positive Occurrences and Geometric Means by
Year for Total DDT Equivalent 31
3.2 Percent Positive Occurrences and Geometric Means by
Year for Dieldrin 31
3.3 Percent Positive Occurrences and Geometric Means by
Year for Hexachlorobenzene 32
3.4 Percent Positive Occurrences and Percent >3ppm by
Year for PCB 32
4.1 States Where Agricultural Soil and Crops Sampled
for Pesticides During 1968-73 - National Soils
Monitoring Program, U.S. Environment Protection
Agency 49
4.2 Percentage of Occurrence and Geometric Mean by
Years 51
5.1 Percentage of Samples with Positive Detections and
Geometric Means Over Time - Water - All Sampling
Sites 65
5.2 Percentage of Samples with Positive Detections and
Geometric Means Over Time - Sediment - All
Sampling Sites 66
6.1 (a) Concentration of Chlordane in Air 4/75-4/76.... 77
(b) Concentration of Dieldrin in Air 4/75-4/76 77
(c) Concentration of Malathion in Air - Greenville,
Miss. - 1977 & 1978 77
(d) Concentration of DDT/DDD in Air - 1977 77
-------
1. INTRODUCTION
In this report the Research Triangle Institute (RTI) will describe
a brief investigation of six pesticide networks run by the Environmental
Protection Agency. The investigation of each network is necessarily
brief due to the fact that the time frame for this report was rather
limited. The network data files used by RTI for analysis were prepared
from tape files and documentation provided by EPA.
In particular, the data networks examined here are:
1. National Human Adipose Tissue Network,
2. National Soil Residue Network,
3. National Surface Water and Sediment Residue Network,
4. Air Network,
5. National Soil Application Network, and
6. National Urban Soil Network.
The data for these networks were in various stages of completeness
when work on this task began. Several years of data were either still
in hard copy form or were waiting to be edited. Accordingly, in order
to meet its limited time schedule, RTI contracted Viar and Company to
assist in data editing and converting hard copy data to computer read-
able form.
RTI personnel reviewed the existing software that comprise the
HUMAN and SWEMS systems that are used to edit monitoring data. The
decision was made for this task not to implement any of this software at
the RTI computer facility (Triangle Universities Computation Center,
TUCC). Instead, the Viar Company used the existing software and com-
puting resources of EPA to complete the data editing that has been done
to date.
In brief, the data finally available to RTI from each pesticide
network were the following:
(1) National Human Adipose Tissue Network—Data on pesticide
levels found in human adipose tissue from a national network
of hospitals for the years FY 1970-FY 1977. The number of
individuals and hospitals from which data were collected
decreased over time (approximately 1400 persons in 1970 to 760
-------
-2-
persons in 1977). Appendix Tables 1.1 and 1.2 present the
data collection forms used for reporting the pesticide levels
for this network.
(2) National Soil Residue Network—Data on pesticide levels found
in a national network of agricultural soil sites (levels in
soil and in crops) for the years FY 1969-FY 1974 excluding
FY 1971 (note, data are no longer being collected for this
network). In general, approximately 1,400 soil samples and
730 crop samples were collected once each year. The data col-
lection form for this network is given in Appendix Table 1.3.
(3) National Surface Water and Sediment Residue Network—Data on
pesticide levels found in a national network of water and
sediment samples for the years 1976-1979. These data were
collected quarterly for water and biannually for sediment.
At the present time the first two quarters of 1976 and the
last two quarters of 1979 are missing from RTI's analysis
file. The sample sizes for water include approximately 130
sites and those for sediment approximately 100 sites. Appen-
dix Table 1.4 presents the data collection form for recording
pesticide levels for this network.
(4) Air. Network—Data on pesticide levels in ambient air from three
or four cities for the years 1975-1978. These data are quite
limited and were collected for different cities in different
years; thus, no trend analysis is possible. No data collec-
tion form was available to RTI for this network.
(5) National Soil Application Network—Data on compounds applied
to a national sample of cropland sites for the years FY 1969-
FY 1974 excluding FY 1971 (again, these data are no longer
being collected). These data were collected once each year by
personal interview with the landowner or operator of approxi-
mately 1,200 sampling sites. No data collection form was
available to RTI.
(6) National Urban Soil Network—Data on pesticide levels found in
urban and suburban soils from approximately 2100 sampling sites
in 36 Standard Metropolitan Statistical Areas for the period
FY 1971 - FY 1976. The sites employed in FY 1971 were resampled
in FY 1977 enabling the effect of time to be examined.
-------
-3-
In the original work plan for the pesticide network investigation
described here, RTI was to receive a list of pertinent questions for
each pesticide network from EPA. These questions were to form the basis
for the investigation. Because only a few specific questions were sup-
plied to RTI, RTI developed its own set of questions so that the investi-
gation could be carried out in the required time frame. RTl's list of
questions are the following:
(1) What statistics are appropriate to summarize the pesticide
levels for the network (e.g., arith. means, geom. means, per-
cent detected, mean of positive values, percent trace, percen-
tiles....)?
(2) How should the various summary statistics be tested for
differences over time (e.g., before and after an event such as
DDT use stoppage, years, seasons,....) and for one time period
(e.g., between Census Regions)?
(3) What trends, if any, exist over time in particular pesticide
levels for the network, nationally and for geographic areas
such as Census Regions (using the tests given in 2)?
(4) What geographical differences exist in particular pesticide
levels for a given time period (using the tests given in 2)?
(5) For the Human Tissue network are there age, race, and sex
differences in pesticide levels over time and for a given time
period?
(6) In general, how can the various networks be used for pesticide
monitoring (e.g., a surveillance network that flags unusual
trends and increases in percent detected for particular
pesticides)?
(7) What objectives can be met using the present sampling plan for
the network?
(8) How might the present sampling plan be altered to meet addi-
tional network objectives?
(9) Is there potential for matching network data files (e.g.,
water and tissue data) for analysis purposes (e.g., how long
does it take for increased water levels of a pesticide to be
detected in human tissue)?
-------
-4-
Within the available time frame RTI has attempted in this report to
answer the above questions for each of the six pesticide networks inves-
tigated.
In particular, in this report, Section 2 addresses the questions
listed above for which answers are pertinent to all of the six networks
(e.g., summary statistics, tests of significance,...), Sections 3 through
8 describe each of the six networks in detail and the questions listed
above that apply to each network (e.g., network trends over time for
particular pesticides). Section 9 discusses in general terms the accu-
racy of pesticide chemical analyses, while Section 10 gives recommenda-
tions and a brief summary of the report. In each of Sections 3 through
8 a sampling procedure subsection is first presented, which gives the
sampling procedure used for each network and addresses questions con-
cerning generalizability of the data; then a data analyses subsection is
given which discusses such questions as trends over time and geographic
differences for particular pesticides.
Before proceeding with a description of the analysis of the data,
it is worthwhile to mention that all of the data for the various pesti-
cide networks with the exception of the Air Network, has been loaded on
the TUCC computer system at RTI and is available for additional analysis.
In general, the original formats (see Appendix Tables 1.5 and 1.6)
were retained for these files, except for a few modifications which were
necessary before analysis could be undertaken. The original records for
the HUMAN System were written as variable length but were in fact fixed
length; hence, they were rewritten as fixed. In some cases data were
repositioned in the file for a particular network to preserve consis-
tency. For example, the Human Tissue files did not contain the pesti-
cide levels in the same position on each record and it was therefore
necessary to reconstruct some records. The size of the Soils Residue
file was reduced by decreasing the number of pesticide residues from 39
to 13. This was done to create a smaller, more efficient file for
analysis. The new record length is 316 bytes as opposed to the original
536 bytes. At the same time, a decimal point in the Soil residue measure-
ment was explicitly entered. In order to accomplish this, the least
significant digit was dropped in some cases. Similarly, the Water
Residue file was reduced from a length of 536 bytes to a length of 390
-------
-5-
bytes. The list of residues included in these analysis files is pre-
sented in Appendix Table 1.7.
-------
-6-
2. DISCUSSION OF QUESTIONS PERTINENT TO ALL NETWORKS
Summary Statistics
The first question examined by RTI for the various pesticide net-
works was what summary statistics seemed appropriate to describe the
levels of a particular pesticide at one point in time. After examining
the data, it was found that in many cases the distribution of particular
pesticide levels had a large number of zero values and a few positive
values. For example, for the soil residue data the following distribu-
tions were found for aldrin (A), dieldrin (D), and heptachlor (H) in
1972 (fiscal year):
Value (ppb)
0-10
10 - 100
100 - 200
200 - 500
> 500
Frequency
A D H
1239 983 1305
91 213 54
17 83 10
17 78 6
14 21 3
Cumulative
Frequency
A D H
1239 983 1305
1330 1196 1359
1347 1279 1369
1364 1357 1375
1378 1378 1378
Cumulative
Percent
A D H
89.9 71.3 94.7
96.5 86.8 98.6
97.8 92.8 99.3
99.0 98.5 99.8
100.0 100.0 100.0
Frequency
455
8
9
6
6
6
Cum. Freq.
455
463
472
478
484
490
Cum. Percent
92.9
94.5
96.3
97.5
98.8
100.0
Similarly for the water data, the following frequencies describe
atrazine levels over the four quarters of 1977.
Value (ppb)
0
0 - .400
.400 - .600
.600 - .800
.800 - 1.100
> 1.100
Accordingly, it is quite clear that the pesticide levels do not
follow a normal distribution. In fact, in many cases the distributions
appear to be a mixture of a discrete distribution (a large number of
zero or less than detectable values) and a continuous distribution (the
positive values). In addition, in many cases the positive values are
skewed to the right indicating a lognormal or exponential distribution.
Thus, it appears for this type of data that primarily three summary
statistics should be computed; namely, (1) the proportion of positive
values, (2) the geometric mean of the positive values (or if almost all
-------
-7-
values are positive, the geometric mean of all values), and (3) the pro-
portion of values greater than some "meaningful" value. That is, since
the distributions are a mixture of discrete and continuous, more than
one summary statistic is needed to describe them. Of course, other
summary statistics may also be used (e.g., medians and percentiles of
the positive values), but in general, the proportion of positive values,
the geometric mean, and the proportion greater than some "meaningful"
value seem to be the most appropriate. It should be noted that if the
proportion of positive values is large for a particular pesticide (this
is true for some of the human tissue data) that the geometric mean of
the positive values will be close to the geometric mean of all values
(including zeros) which appears to be one of the primary statistics used
currently in EPA reports to summarize the pesticide levels (e.g., see
[2.1] and [2.2]). (Note, when computing geometric means of data with
zero values, a constant must be added to each data point, see Appendix 2)
As an illustration of the proportion greater than a particular
value, consider for the soil residues data aldrin and dieldrin propor-
tions found to be greater than .05 and .10 ppm over time.
Aldrin Dieldrin
Year (FY)
1969
1970
1972
1973
1974
The preceeding type of table is an extremely useful procedure for track-
ing pesticide levels over time for all of the monitoring networks. Of
course, defining "meaningful" levels for each pesticide for the various
networks is a complex problem. For the current report, meaningful
levels were not available to RTI; and therefore, when the proportion
greater than particular levels are given they simply were selected for
convenience. However, it would appear to be worthwhile for EPA to
consider defining "meaningful" levels whenever possible since a summary
statistic such as the geometric mean over time may be of little value if
the observed means are of little practical interest (e.g., relatively
small).
. 05 ppm
.058
.070
.150
.040
.019
. 10 ppm
.042
.045
.035
.029
.014
.05 ppm
.163
.200
.187
.184
.128
. 10 ppm
.099
.138
.132
.128
.078
-------
-8-
Tests of Significance
Assuming that the proportion of positive values, geometric means,
and the proportion of values greater than a particular level are used as
summary statistics, then appropriate statistical tests to use to detect
pesticide level differences; such as, differences over time or between
geographic areas are:
(1) For the proportion positive or greater than some "meaningful"
2
value a x test may be used to test if the proportions for
several years or several geographic regions are equal - i.e.,
suppose the proportion of positive values for soil levels for
a particular pesticide are Pn , P0, ..., P, from 1969 through
1 Z D
1974 over the entire soil network. Then to test if these six
proportions are equal, the following test statistics applies:
- P~)2 n(P - P~)2
x
v '
P Q P Q P Q
6
E n-P.
— =
where P =
Q = 1 - P; n. = the number of observations that each P.
2 2
is based upon and X/c\ is the x statistic with 5
degrees of freedom.
2
A standard x table may be used to determine if the test is signi-
2
ficant (e.g., at the .05 level of significance X/r\ =11.1). If the
test is significant then it may be of interest to determine which of the
proportions are significantly different (see [2.5]).
(2) For the geometric means a one-way analysis of variance (ANOVA)
may be used to test if a series of geometric means are equal.
This test may be carried out by taking the natural logarithms
of the pesticide levels and using an ANOVA on the transformed
values (note, it may be advisable to add the minimum positive
value to each number before taking logs or to transform the
data by a simple scale factor (ppm to ppb) to avoid obtaining
large negative numbers after taking logs {see Appendix 2]).
-------
-9-
For example, suppose the mean positive DDT levels for several
years are to be tested from the following data matrix:
1970 1971
1974
X
11
X.
X
al
12
a2
X.
15
X
a5
where X.. are the DDT levels in year j for sampling site
i.
After taking the natural logarithms of the data, the
following ANOVA table may be generated (e.g., see [2.3]):
ANOVA TABLE
Source of
Variation
Between
Years (T)
Within
Years
Total
Degrees of
Freedom
4
5 (a-1)
5 a-1
Sum of Squares
2
v 2 , Y
CC/'T'S — \ v/-i J...
bb(.l; - /, i. ./a - _
3 J 5a
ss(w) = E Yij - E YVa
Mean Square
SS(T)
ti^VJ-'/ 4
F-Test
MS(T)/MS(W)
where Y
±j
Y... =
„
EE
i J
The test for equality of the geometric means is the F-test given in
the last column of the ANOVA table. A standard F-table may be used to
determine significance (e.g., at the .05 level of significance the F-
test with 5 and 120 degrees of freedom is significant if the calculated
F-yalue exceeds 2.29). If the F-test is significant, it implies that
the geometric means for 1970 through 1974 are statistically significant.
In this case, it may be of interest to examine which geometric means are
significantly different by using a procedure such as the Duncan Multiple-
Range Test (see 12.3J). In addition, if a trend over time is evident,
-------
-10-
one might fit a regression equation with pesticide levels as the depen-
dent variable and time as the independent variable. This would then
quantify the observed trend.
It should be noted here that the one-way ANOVA given above reduces
to a simple two- tailed Student t-test if only two geometric means are
being compared (e.g., only two time periods such as before and after DDT
use was suspended) . It is also important to note that the above statis-
tical tests assume independent observations in determining the propor-
tion of positive values, proportion greater than some level, or the
geometric means. This assumption is probably not true for testing
different time periods (e.g., the water samples from the water network
are collected at the same locations from season to season). In general,
ignoring this lack of independence may result in declaring too few
significant differences over time. Unfortunately, the data available on
the pesticide data files are for very limited time periods; and therefore,
time series analysis which would take into account correlations between
samples over time cannot be used effectively. Hopefully, in the future
when several years of pesticide data are available on the data files, it
will be possible using appropriate techniques, to take into account
correlations over time in making statistical tests among time periods.
Finally, it is important to note that the above analyses have not
considered more than one factor at a time in analyzing pesticide levels
(i.e., only time or geographic region were considered).
However, when analyzing pesticide levels from a data base such as
the human tissue network, it is necessary to consider several factors at
a time (e.g., age, race, sex, ...) in the analysis. To handle this
problem, models such as the following may be needed:
Ai + Sj + \ + T* + Lm + £ijk£mn
where
.., „ = pesticide level in tissue for the n individual, in the
ijk&m y h th
m location, for the £ time period, in the (ijk) age-
race-sex group;
H = mean pesticide level;
A. - age effect for the i age group;
S. = sex effect for the j sex group;
-------
-11-
R, = race effect for the k race group;
T = time effect for the £ year;
L = location effect of the m location;
m
random error'
In the above model, testing for age, sex, race, time and location effects
corresponds to testing the equality of the A., S., R^ , T and L .
i j K. Jo m
Standard statistical computer software may be used to carry out these
tests for unweighted data and specialized software is available at RTI
for carrying out the analyses for weighted data (e.g., [2.4]). In Section
3 a model similar to the above is used for testing the proportion of
positive values and the geometric means of the positive values for
several pesticides found in human tissue.
Monitoring Procedures
One of the primary purposes of the various pesticide networks
should be to determine if unusual or unexpected trends are occurring in
the various pesticide levels over time. This might be a national trend
or a trend for a smaller geographic area such as a Census region or a
State. In order to detect unusual or unexpected trends in the various
networks, several approaches might be investigated including:
1. control charts on the geometric means of the positive values
for each pesticide,
2. control charts on the percent positive or the percent greater
than some level for each pesticide,
-3. comparisons of trends for percent positive or percent greater
than a particular level, or geometric means of positive values
from past data and current data.
The control charts would simply involve computing the means and standard
deviations of the statistic of interest (e.g., percent positive) for a
particular pesticide in a geographic area from past data and then deter-
mining if current data were within +_ 2 standard deviations (say) of the
historical mean. Values outside the +^ 2 standard deviation would be
flagged. (.Note, + 2 standard deviation represents 95 percent confidence
limits for the normal distribution). For example, for human tissue
levels of DDT from 1971 through 1977 the following control chart could
be used.
-------
-12-
CONTROL CHART FOR MEAN LOG (DDT) LEVELS OVER HUMAN TISSUE NETWORK
2 15
*• • -L-/
2.10
2.05
2.00
1.95
1.90
1.85
Mean 1.80
Ln (DDT) 1>75
1.70
(ppm)
1.65
1.60
1.5
1.3
1.1
X
X
X
X
X
X
71 72 73 74 75 76 77
Control limits
Control limits
In constructing the above control chart, the center line was the mean of
the natural logs of DDT levels in 1970 and 1971, and the dashed control
lines are + 2 standard deviations from this mean (center line).
The comparisons of trends might involve computing regressions over
time of a pesticide level from historical data and then predicting the
current value of the pesticide level and its confidence limits from the
regression equations. The current value of the pesticide levels could
then be compared with the predicted value and its confidence limits. Of
course, if the slope of the regression line is not significantly diffe-
rent from zero (i,.e., no trend is evident) then control chart analysis
is probably sufficient to monitor the data.
-------
-13-
Furthermore for trends over time, it would also seem very important
to be able to detect pesticides that had not been detected in the past
at any sampling site but then are detected at a few sites. This phenom-
enon would not be detected by a control chart on percent positive since
all historical data is zero. Therefore, it would seem appropriate when
a few sites (e.g., 1 or 2%) begin to show positive values to flag these
sites and to investigate in detail where the percent positive sites are
located. For example, if the positive sites are all in one area, this
might be cause for some specific action such as oversampling of this
area (e.g., the special 1976 Mirex study in southern States) to deter-
mine the extent of the problem caused by the particular pesticide. As
an example of this, consider PCB percent positive in the approximately
1400 agriculture soil sites in the Soil Network from 1972 to 1974.
% of Sites
Year Sample Size* No. Pos. Sites Positive
1972 1401 0 0
1973 1403 2 .14
1974 1388 24 1.73
* Number of sampling sites
The table indicates that in 1972 no sites had detectable PCB levels, but
in 1973 and 1974, there is an indication that PCB levels are beginning
to be detected at a few sites. RTI examined the location of the 24
positive PCB sites in 1974 and found four of them to be in New York,
three of them in Michigan, and three in Nebraska. No other state had
more than two positive sites.
In addition to trends over time for a particular geographic area,
it may also be instructive to examine trends over time for several
geographic areas simultaneously. This will indicate how different areas
of the country compare over time with regard to percent positive, per-
cent greater than a particular level, or geometric means for a particu-
lar pesticide. For example the following plot indicates for sediment
samples the percent of sites from four water basins where PCB levels
were detected in 1976, 1977 and 1978:
-------
-14-
90 -
80 -
70
13
2 60
o
-------
-15-
phically close. Possible crude linkage by state or census region is
possible, but the value of this level of matching is questionable (e.g.,
note, there are limited urban soil and human tissue data from the same
cities).
The ability to relate observations of one network to another could
have many advantages. For example, if the soil application, soil and
crops residue, water residue, and human tissue networks were linked
appropriately, increasing concentrations of specific pesticides might be
traced from the application on crops to the water supply and into the
human population. Questions about the rate at which pesticides move
through the environment and sources of contamination could be investi-
gated.
Linkage of the networks would require a special sampling plan that
considered the various networks simultaneously. One feasible plan might
be to randomly sample watersheds then sub-sample the city(ies) and water
supply(ies), the soils (and crops) in the watershed, and the ambient air
levels. Other factors that could effect the concentration of pesticide
in adipose tissues might also be sampled such as pesticide levels in
commonly consumed food (e.g., milk).
Of course, it should be emphasized, that even if the monitoring net-
works were matched in some manner, the prediction of tissue levels from
soil, water, air, and food levels might be impractical due to the rela-
tively large number of uncontrolled factors in the individuals sampled
(e.g., smoking habits, weight, diet, occupation, where time is spent,
general health history, ..., etc.). However, averages over geographic
areas might prove to be extremely useful in examining the sources of
pesticide residues in humans. For example, with properly designed
pesticide networks, plots such as the following could be examined for a
given geographic area.
Mean
Pesticide
x
Levels in
Human
x
Tissue
(Time 1)
Pesticide Levels in
Corresponding Soil Samples
(Time 2).
-------
-16-
Thus, it would seem worthwhile to at least investigate the possibility
of matching data from the various networks. However, it is important to
state that designing sampling procedures so that data from the various
networks could be matched would probably sacrifice geographic coverage
of a particular network for a fixed amount of funds. That is, if the
soil network were tied to the human tissue network, then for a given
amount of funds the soil network would not be able to cover as many geo-
graphic areas as it would if its sampling procedures were independent of
the human tissue network.
Practical versus Statistical Significance
Before preceeding with the analyses of the various pesticide net-
work data files it is important to discuss, briefly, practical versus
statistical significance. To facilate the discussion, consider the
following example from the Human Tissue Network that involves the
geometric means of hexachlorobenzene levels by sex over years.
Sample Size Geometric Mean (ppm)
Male 1562 .044
Female 1564 .039
A test of significance of the two geometric means is statistically sig-
nificant at the .01 level of significance because of the large sample
sizes. However, from a practical standpoint, the two geometric means
are probably not significantly different; particularly since the levels
of hexachlorobenzene are only recorded to two significant digits. Thus,
it is very important to keep in mind that an investigator should not
indiscriminately run and report the results of statistical tests of
significance without keeping in mind what differences in his data are
different from a practical standpoint. Tnis is analogous to the dis-
cussion given above on defining "meaningful" levels when summarizing the
data for a particular pesticide. Simply performing statistical tests,
particularly with large sample sizes, is not sufficient to determine
what differences are important. The investigator should incorporate his
knowledge and experience of his particular data in interpreting what
differences are really meaningful.
Running several statistical test on a data base is fine, but it
should not be done in a vacuum where intimate knowledge of the data by
the investigator is ignored when results are reported.
-------
-17-
REFERENCES
[2.1] Carey, A.E., J.A. Gowen, H. Tai, W.G. Mitchell and G.B.
Wiersma: "Pesticide Residue Levels in Soils and Crops in 1972,
National Soils Monitoring Program", Technical Service Division,
Office of Pesticide Programs, U.S. Environmental Protection
Agency, Washington, D.C., 1972.
[2.2] Kutz, F.W., A.R. Yobs and S.C. Strassman: "Effects of
Reducing DDT Usage on Total DDT Storage in Humans", Pesticides
Monitoring Journal, Vol. II, pp. 61-63, September, 1977.
[2.3] Steel, R.G.D., and J.H. Torrie; Principles and Procedures
of Statistics, McGraw-Hill, New York, 1960.
[2.4] SAS User's Guide, SAS Institute Inc., Post Office Box 10066,
Raleigh, North Carolina 27605.
[2.5] Goodman, L.A., "Simultaneous Confidence Intervals for
Contrasts Among Multinomial Populations", The Annuls of
Mathematical Statistics, 1964, Vol. 35, pp. 716-725.
-------
-18-
3. NATIONAL HUMAN ADIPOSE TISSUE NETWORK
3.1 Detailed Description of the Network
3.1.1 Sampling Procedures
a. Sample Design
The statistical design used to collect data for the
Adipose Tissue Survey of the National Human Monitoring Program (NHMP)
has several stages. The contiguous 48 states were stratified into
several regions. Sampling sites were selected from a list of eligible
places proportional to the population. Within the sampling sites, the
subsampling was performed by cooperating pathologists and medical exam-
iners .
In FY70-FY72 the contiguous 48 states were stratified by census
region. Beginning in FY73 the strata were changed to census divisions.
The details of the census regions and divisions by States is given in
Appendix 3.1. In the earlier period, the number of sites selected
within each stratum is determined by its population as given in the 1960
Census. The allocation was: Northeast 11, (28%); North Central 12
(30%); South 9 (24%) and West 7 (17%). In FY73, the allocations were
revised based on 1970 Census data. The strata were also changed to
coincide with the nine Census divisions. The allocations are summarized
in Table 3-la.
In FY77, the sample sites where changed from cities of greater than
25,000 persons to Standard Metropolitan Statistical Areas (SMSA). The
allocation of the 40 sample sites is also summarized in Table 3-la.
In FY70-72, the eligible places were cities with populations great-
er than 25,000 persons based on the 1960 Census. In FY73-76, the eli-
gible places were cities greater than 25,000 based on the 1970 Census.
In FY77, the sample sites were SMSA's with size also being based on the
1970 Census. For FY70, FY73, and FY77 the sample sites were indepen-
dently selected for each stratum with probability proportional to size.
This was done by listing the sites in random order along with their
cumulative totals. An interval for each stratum was calculated by
dividing the total population for all sites listed by the number to be
selected in the stratum. (An example of this process is given in Appen-
dix 3.2). A random number was obtained between 0 and the length of the
-------
-19-
TABLE 3-la
Number of Sample Sites for Each Stratum FY73 - FY77
Census Division
New England
Middle Atlantic
East North Central
West North Central
South Atlantic
East South Central
West South Central
Mountain
Pacific
FY 73-76
4
14
15
6
11
5
7
3
10
75
FY 77
2
7
8
3
6
3
4
2
5
40
Percent of
Allocation
6
18
20
8
15
6
10
4
13
100
TABLE 3-lb
Sample Size by Design and Surplus Records
Year
1970
1971
1972
1973
1974
1975
1976
1977
Total number
of records
2919
3379
4351
1276
1050
910
785
907
Number of
design records
1436
1595
1922
1117
924
793
689
773
Number of
surplus records
1483
1784
2429
159
126
117
96
134
Number used
for weights
1456
1624
2034
1213
1050
910
785
907
* This number includes all "D" records and surplus "S" records from
cities selected in sample design.
-------
-20-
interval to give a starting point. The sites were selected by matching
their cumulative totals with the starting point or integer multiples of
the interval plus the starting point. A listing of the originally
selected sites can be found in Appendix 3.3.
When no cooperative pathologist or medical examiner could be
found, alternate sites were selected. These sites were chosen by posi-
tion in the listing with respect to the nonrespondent site. The first
alternate was the site immediately below the originally selected site,
the second alternate the one immediately above, the third alternate was
the second site following and so on. A listing for each year of the
sites where samples were supposed to be taken is given in Appendix 3.3.
The subsampling within each site was done with the aid of coope-
rating pathologist and medical examiners. Each site was assigned a
quota based on the demographic characteristic of age, sex and race. The
ages were grouped into three ranges 0-14, 15-44, and greater than 44.
The races were classified as white and nonwhite. The quotas for FY70-
FY72 were based on the National demographic characteristics according to
the 1960 Census [ref. 3.1J. The quotas for FY73-FY76 and FY77 + were
determined by the appropriate demographic characteristics for each
census division. Appendices 3.4 and 3.5 contain the quotas for each
site and census division.
Below the sample city (or SMSA) level there was no probability
structure for obtaining the participants in the survey. The hospitals,
pathologist and medical examiners were selected subjectively. Also, the
tissue samples from cadavers and surgical patients were selected subjec-
tively by the cooperating pathologists, etc. Guidelines for selecting
patients were distributed by EPA. Appendix 3.6 contains the letter
sent to each cooperating professional containing these guidelines.
b. Computation of Sample Weights
Even if one assumes the selection process within sample
sites is approximately random, the sample design of the NHMP adipose
tissue network does not assign equal "probabilities" of selection to all
elements in the sample.. In this situation a sample weight for each
observation should be calculated that reflects its approximate proba-
bility of selection. Including weights in the analysis should reduce
-------
-21-
bias in estimating means or proportions (for probability sampling this
results in strictly unbiased estimations for such statistics). In the
following paragraphs procedures are given for computing weights for the
NHMP (true sampling weights cannot be calculated because some stages in-
volve non-probability sampling).
With the information on hand, RTI was not able to calculate the
probability of selecting individuals within each sampling site (city or
SMSA). Hence, equal probability of selection was assumed for the within
site stage of sampling. This assumption, within the given time, was re-
quired for RTI to calculate the probability of selecting an individual.
This method involved dividing the number of individuals selected by the
population of the city. The probability of a city being selected is
calculated by dividing the population of the city by the length of the
selection interval (see Section 3.1.la, above). Hence the weight is
given by
. , _ length of interval population of city
population of city number selected in city"
Note that the population of city terms cancel giving the simplier ex-
pression
. , ^ length of interval
weight = —a ~, :—: : •
number selected in city
For several of the larger cities, the population was larger than the
interval, hence these cities were selected with probability one.
Quotas were assigned for each site depending on the population
characteristics of age, sex and race. The quotas were based on stratum
or national characteristics which most likely did not reflect the demo-
graphic characteristics of the individual sites. The time limitations
of this task and other design factors did not permit taking these fac-
tors into consideration in computing weights.
Only a subset of the data for each year is used in calculating the
weights. The data for each year is classified into two groups. The
data record is classified as a design "D" record or a surplus "S"
record. "D" records are usually from cities selected in the sample
design and included in the quota for the site. "S" records are all
samples from cities not selected in the design and also surplus samples
from design cities not included in the quota.
-------
-22-
For purposes of weight calculation, there appears no reason to
treat "D" and "S" data differently from the same sample site. Hence, in
calculating the weights, RTI includes both "D" and "S" records into the
number selected from each sample site. All data from cities not included
in the sample are excluded in calculating the weights (see Table 3-lb).
The weights are also adjusted for "nonresponse" depending on the confi-
dence codes.
Analysis of the data files using the weights calculated above is
continuing at RTI. The results will be reported at a future date. For
this report, because of limitations; however, it was necessary to carry
out some analysis with unweighted data.
c. Evaluation of NHMP Sampling Plan
The present sampling plan for the NHMP adipose tissue
network can be used to answer various questions of concern. The basic
sampling plan has many good features but could be improved. Some gene-
ral comments and recommendations are given below. Since several stages
of the sample plan are done judgmentally, estimates of the variance of
means and proportions cannot be done using an exact sampling distribu-
tion. One must make assumptions on the unknown effects of certain stages
of the sample design to make statistical inference.
Several important topics can be addressed with the NHMP data files.
With the files the concentration of many pesticides for a relatively
large subset of the U.S. population can be recorded. Changes over
time (trends), differences between various subpopulations based on age,
sex, and race can be investigated using these files. These topics can
be investigated nationally or for various geographic regions.
If a true surveillance network is desired, then the capability is
needed to subsample quickly suspected hot spots. The NHMP could arrange
in advance to have additional pathologists and laboratories available to
select and analyze samples. The additional information would be helpful
in discovering any causes of the unusual readings detected by the over-
all program. To make this effective, the lagtime (presently more than
a year) from obtaining samples to reporting results would have to be
reduced.
Finally, the following are a few general comments and recommenda-
tions on the current sampling design:
-------
-23-
The purposive exclusion of cities less than 25,000 persons or non-
SMSA's in the sampling frame should be modified. This results in a
significant fraction of the total population having no chance of being
selected in the sample. Allowing the selection of sampling sites from
places greater than 2,500 persons would essentially eliminate this
difficulty. The techniques of selecting alternate sites need improve-
ment. Repeating the process by which the original sites were selected
would be satisfactory (3.1.1.a above).
The purposive method of selecting hospitals within sample sites is
undesirable. The method of subsampling within cities needs to be modi-
fied so that the selection probabilities for individuals may be calcu-
lated. Sampling frames for hospitals or pathologist could be constructed
for selected cities. The pathologist could then be randomly selected
from these frames with a known probability. A simple protocol describ-
ing how surgical patients and cadavers are to be sampled would allow for
calculations of the selection probability for this stage of sampling.
Invoking the assumptions that patients live in the same geographic area
as the hospital from which the sample was obtained could be improved on
by recording the zip code o£ residence. The site from which the tissue
sample is taken could be recorded. This would help in investigating
and accounting for the difference in tissue concentrations of residues.
These alterations are offered for consideration, recognizing that some,
while theoretically desirable, may not be practical.
3.1.2 Data Available on Data File
Pesticide and chemical residue data collected through the
National Human Monitoring Program were available in machine readable
form for 15,577 individuals. For each individual, residue amounts in
adipose tissue were determined for nineteen (19) different pesticides
and polychlorinated biphenyls (PCB's) for fiscal years through 1977. In
addition to the 20 different residue amounts, each individual's record
contained information on the fiscal year the sample was taken, the
census division in which the sample was taken, the individual's age,
race and sex, and whether or not the sample was taken as part of the
National Human Monitoring Program's designed sample discussed in Section
3.1.1. Although not complete for all samples, information was also
obtained on diagnosis, height, weight, and occupation for each indivi-
dual. Of the 15,577 samples collected, 9,249 (.60%) were from the de-
signed sample.
-------
-24-
3.1.3 Assumptions Made in Analysis of the Network
It was decided to attempt analysis of all twenty residue
amounts; however, six of the DDT derivatives were combined to form a
total DDT equivalent by the formula:
Total DDT = pp'DDT+op'DDT+l.UAtpp'DDE+op'DDE+pp'DDD+op 'ODD) .
Hence, analysis was confined to the following fifteen residues:
Total DDT Equivalent
a-BHC
B-BHC
Lindane
6-BHC
Aldrin
- Dieldrin
Endrin
- Heptachlor
Heptachlor Epoxide
Polychlorinated Biphenyls (PCB)
- Oxychlordane
Mir ex
Trans Nonachlor
- Hexachlorobenzene
Factors appearing on the individual records were categorized into
the following levels:
Factor Levels
Time FY 1970
1971
1972
1973
1974
1975
1976
1977
Age 0-14
15 - 44
> 45
Race White
Black
Other
Sex Male
Female
-------
-25-
Census Divisions New England
Middle Atlantic
South Atlantic
East South Central
East North Central
West South Central
West North Central
Mountain
Pacific
For each record, a decision was made with respect to its inclusion
in the analyses and for each residue on that record certain decisions
and data transformations were made. On an individual record basis the
following decisions were made:
If the observation was not from the designed sample it was not
considered for analysis (only "D" samples considered, recall
that the analysis in this section is on unweighted data),
- If a residue amount had unsuccessful confirmation, then the
residue amount was made equal to zero (confidence code I or
J),
- If the record indicated that a technical error had been made
for a particular residue amount, that residue amount was con-
sidered missing (confidence code K),
- If the confidence code for a particular residue amount was
blank, that residue amount was considered missing,
If the record indicated that there was less than 10% Lipid Ex-
tractable Material, all residue amounts were considered miss-
ing for that record,
- If a residue amount was a trace amount, then that residue
amount was set equal to zero,
- All residue amounts except PCB's were divided by the proportion
of Lipid Ex-tractable Material and were the basis of analysis,
For all positive, non-PCB residue amounts, the natural loga-
rithm of the amount was calculated,
- For all residue amounts, an indicator variable was created
which, was one if the residue amount was positive and zero if
the residue amount was zero. If the residue amount was miss-
ing, the indicator variable was considered missing,
-------
-26-
- For PCB an additional indicator variable was created which was
one if the PCB residue amount was greater than three parts per
million (3ppm) and zero if less than 3ppm. The indicator
variable was considered missing if the PCB residue amount was
missing.
3.2 Data Analysis
3.2.1 Overview
With the data set created pursuant to the rules given in
Section 3.1.3, two analyses were conducted for each of the fifteen resi-
dues.
For the non-PCB residues, an analysis of variance was conducted on
the natural logarithms of the positive residue amounts to test for
differences among the factor levels (also given in Section 3.1.3). In
addition, for each residue (including PCB), a chi-squared analysis was
performed to test for differences among the same factor levels in the
percent of positive detections of the pesticide. For PCB's, the propor-
tion of individuals with greater than 3ppm PCB was analyzed by chi-
squared techniques to test for differences among the factor levels. In
all the analyses conducted, the contribution of a specific factor to the
variable (% positive or log {amount]) being analyzed was assessed after
statistically adjusting all other factors. This method of analysis is
equivalent to comparing the levels (e.g., fiscal years) of a factor
(time) after adjusting for all other factors.
The use of the analysis of variance and chi-squared techniques
given above presumes data present in sufficient quantity to allow sta-
tistically valid comparisons among the factor levels. Therefore, the
following rules were made to allow the comparisons:
If a residue had many fewer than 10% positive detections or
many more than 90% positive detections only difference among
the years were tested using the statistic percent positive
detections; however, tests employing the geometric mean were
performed where appropriate.
If a residue had fewer than 1% positive detections, no tests
were performed.
If a residue had more than 99% positive detections, no tests
were performed using the statistic percent positive detections;
-------
-27-
however, other tests using the statistic geometric mean were
performed where appropriate.
- If fewer than 10% of the amounts were positive for a particu-
lar residue, then no tests were performed using the statistic
geometric mean.
For Oxychlordane, Trans Nonachlor and Hexachlorobenzene, the
percent positive detections could not be distinguished from
blanks for all years. For these pesticides, tests employing
the statistic percent positive detections are based only on
the data from years where the percent positive detections
could be unambiguously determined.
The above rules led to the following tests for the different resi-
dues:
Residue
Total DDT Equivalent
a-BHC
3-BHC
Lindane
5-BHC
Aldrin
Dieldrin
Endrin
Heptachlor
Heptachlor Epoxide
PCB's
Oxychlordane
Mir ex
Trans Nonachlor
Hexachlorobenzene
3.2.2 Tests
% Positive Detections
no test
years only
years only
years only
no test
no test
years only
no test
no test
years only
Geometric Mean
all factors
no test
all factors
no test
no test
no test
all factors
no test
no test
all factors
all factors (includes % >3ppm)
years only all factors
no test no test
years only all factors
years only all factors
The statistical tests done assume simple random sampling
from the U.S. population. Insofar as the actual sampling procedures are
not simple, the indicated significance levels may be somewhat too high.
However, while the statistical tests used do not precisely accomodate
the sampling procedures, still some important patterns are recognizable.
-------
-28-
In addition, because the sample sizes are so large virtually all diffe-
rences are statistically significant. Perhaps what is more important
than statistical significance is the actual magnitude of a statistically
significant difference (i.e., practical significance).
3.2.3 Summary Statistics
Summary statistics for the residues appear both graphically
and in tabular form for Total DDT Equivalent, Dieldrin, Hexachloroben-
zene and PCB's (Figures 3.1 - 3.4).
For rarely observed positive residue amounts (<1%) only the percent
positive detections (over all factors) are given. These pesticides are
6-BHC, Endrin, Heptachlor, and Mirex and are shown in Table 3.2.
For the uncommly observed (<10%) positive residue amounts, the per-
cent positive detections are given by year, and differences among the
years are tested for significance. These pesticides are ct-BHC and Lin-
dane, and are shown in Table 3.3. Also given in Table 3.3 is the per-
cent positive detections for Aldrin which was not tested because in the
years when it was apparently detected it could not be distinguished from
PCB's.
For all other pesticides, summary statistics are given for each
level of each factor given in Section 3.1.3, with statistical testing
performed where the sample requirements of 3.2.1 are met.
For PCB's, the summary statistics are percent positive detections
and percent >3ppm by factor level (Table 3.4).
Tables 3.5 - 3.11 give summary statistics in terms of percent posi-
tive detections, geometric means and the standard error of the geometric
mean by factor level for the remaining pesticides.
3.2.4 Analysis Results
a. Time Trends
For each residue tested, time (in years) was an impor-
tant source of variation. However, a particular pattern of trend was
not discernible for all pesticides.
The following lists and denotes by X the residue and statistic with
discernible time patterns:
-------
-29-
Residue
a-BHC
PCB's
Total DDT
6-BHC
Dieldrin
Oxychlordane
Time Pattern Type
Decreasing
Increasing
Decreasing
Decreasing
Decreasing
Decreasing
% Positive
Geometric Mean Table
X
X
X
(also % >3ppm)
X
X
X
X
3.3
3.4
3.5
3.6
3.7
3.9
b. Age Differences
Age differences are present for all residue amounts, the
oldest age group has the greatest amount of residue as measured by the
geometric mean. The percent positive detections also vary by age but
not as systematically as the geometric means.
c. Racial Differences
Except for Hexachlorobenzene, whenever there were suffi-
cient data to allow testing, Blacks had greater amounts of pesticide
residue then Whites. Again, the percent positive varied between Black
and Whites but not in the systematic way as did the geometric means.
d. Sex Differences
With one exception, males showed greater amounts of
pesticide residue for all pesticides for which there were sufficient
data. The exception is 3-BHC, where females showed slightly greater
amounts of residue than males.
e. Census Divisions Differences
Again, for the pesticides for which there were suffi-
cient data, census divisions were a significant source of variation.
Attempting to summarize this variation for each pesticide for which a
test was performed, the following gives, by pesticide, those census
divisions with geometric mean amounts 25% greater than the national
geometric mean:
Pesticide
PCB (>3ppm)
Total DDT
e-BHC
Dieldrin
Heptachlor
Oxychlorodane
Trans Nonachlor
Hexachlorobenzene
Census Division With GM >1.25 GM National
Mid Altantic, New England
South Atlantic
East South Central, South Atlantic,
West South Central
None
None
West South Central
East South Central, West South Central
Pacific
-------
-30-
3.2.5 Final Remarks
The analyses presented in this section represent a first
step in the simultaneous assessment and testing of the importance of
various factors on human adipose pesticide/chemical residues.
Future analyses of these data should examine more subtle hypotheses
than have been examined in this report. For example, knowing that there
is significant yearly variation in the data may not be sufficient infor-
mation to make policy decisions. More subtle hypotheses would uncover,
where they exist, orderly time trends and, perhaps, differential time
trends for different population subgroups.
Also, future analyses of the data should be performed using tech-
niques which accomodate the complex selection procedure of the indivi-
dual samples. The complexity of selection derives from both the unequal
selection probabilities of the individual samples and the clustering of
sample elements that occurs due to the hospital based collection of
tissue samples. While the hospital based collection procedure is,
perhaps, the only reasonable way to obtain tissue samples, it does have
an impact on the data variability and should be accounted for in subse-
quent analyses.
Finally, the pesticide/chemical residue data analyzed in this re-
port represent all the residues for which information is routinely
gathered by the National Human Monitoring Program's data collection
instrument "Tissue Pesticide Residue Analysis Report" (Appendix 1).
Unfortunately, this instrument does not contain information concern-
ing pesticide substitutes. For instance, recommended substitutes for
DDT (e.g., parathion, malathion, etc.) do not appear on this data file
and questions concerning exposure to the substitutes cannot be answered.
-------
Figure 3.1.
-31-
Percent Positive Occurrences and Geometric Means by Year for Total
DDT Equivalent
100% H
99%-J
0)
o
S
O
CJ
O
4J
•rl
CO
o
-------
u
c
01
a
u
o
01
>
CO
0
a*
4J
c
01
u
-32-
Figure 3.3. Percent Positive Occurrence and Geometric Mean by Year for
Hexachlorobenzene
100% -
75% -
50% -
25% -
Percent Positive
Geometric Mean
-.050
-.045
r.040
-.035
-.030
1970 1971 1972 1973 1974 1975 1976 1977
Figure 3.4. Percent Positive Occurrence and Percent >3ppm by Year for PCB
I
ex
PO
A
g
QJ
PH
01
O
c
01
3
U
u
o
CO
0
PH
a
II
u
3ppm
-15.00%
-11.25%
- 7.50%
- 3.75%
S
ex
ex
(U
u
01
CM
- 0%
1970 1971 1972 1973 1974 1975 1976 1977
-------
-33-
Table 3.2
Summary Statistics for Rarely Observed ( < 1%) Residues
Residue
6-BHC
Mirex
Endrin
Heptachlor
N
9041
9050
9041
9051
Percent
Positive
.02%
.09.2
.01%
.02%
No tests performed
No tests performed
No test performed
No test performed
Table 3.3
Summary Statistics for Uncommonly Observed ( < 10%) Residues
Years
ALL YEARS
1970
1971
1972
1973
1974
1975
1976
1977
N
9042
1380
1559
1885
1092
901
779
682
764
Residue
a-BHC
% Positive
1.10
2.10 **
0.12
2.01
1.64
0.44
0.12
0.29
0.78
Lindane
% Positive
1.12
1.81**
1.47
0.26
1.55
0.55
0.51
0.43
2.62
Aldrin
% Positive
2.11
0.86 NT
11.48
0
0
0
0
0
0
**=Differences among the years significant at the 1% level (99% confidence)
NT = No tests performed because in 1970-71 Aldrin was not distinguishable
from PCB's.
-------
-34-
Table 3.4
Summary Statistics for PCB
Factor
ALL
YEARS
AGE
RAGE
SEX
CENSUS
DIVISIONS
Factor
Levels
1970
1971
1972
1973
1974
1975
1976
1977
0-14
15-44
> 45
White
Black
Other
Male
Female
E. No.
Central
E. So.
Central
Mid-
Atlantic
Mountain
New
England
Pacific
So.
Altantic
W. No.
Central
W. So.
Central
N
6058
0
0
1765
1114
924
793
689
773
1057
2000
3001
5147
851
60
3101
2957
1332
396
1384
204
228
656
688
434
736
Percent
Positive
90.6
__— _
89.7**
78.5
90.9
94.2
97.9
99.7
81.9**
92.3
92.5
91.3**
86.9
85.0
90.6NS
90.6
88 . 8**
82.8
88.4
97.6
99.1
91.9
94.7
92.8
91.4
Percent
> 3 ppm
7.5
_«»_.
5.8**
5.6
5.0
10.4
9.3
12.3
1.6**
6.5
10.2
6.5**
13.0
10.0
9.0**
5.9
6.6**
9.3
10.4
3.4
16.6
4.7
8.3
5.2
3.9
NT = Factor not tested because of sample size (percent positive)
requirements given in section 3.2.
Not significant at the 5% level (less than 95% confidence).
Factor significant at the 5% level (95% confidence level).
Factor significant at the 1% level (99% confidence level).
NS
*
-------
-35-
TABLE 3.5
Summary Statistics for Total DDT Equivalent
Total DDT Equivalent = [pp'DDT+op'DDT+1.114(pp'DDE+op'DDE+pp'DDD)]
Factor
ALL
YEARS
AGE
RACE
SEX
CENSUS
DIVISIONS
Factor
Levels
1970
1971
1972
1973
1974
1975
1976
1977
0-14
15-44
> 45
White
Black
Other
Male
Female
E. No.
Central
E. So.
Central
Mid-
Atlantic
Mountain
New
England
Pacific
So.
Atlantic
W. No.
Central
W. So.
Central
N
9036
1380
1557
1884
1092
901
779
682
761
1285
2737
5014
7611
1335
90
4709
4327
1912
694
2152
263
320
1025
1055
676
939
Percent
Positive
99.9
NT
99.9iNi
99.7
99.9
100.0
99.8
100.0
100.0
99.8
NT
100. 0 L
99.8
99.9
NT
99.8Wi
100.0
100.0
NT
99.9Wi
99.8
NT
99.9Ni
99.8
99.9
99.6
100.0
100.0
100.0
99.4
100.0
Geometric
Mean (ppm)
6.04
7.90**
8.14
6.89
5.87
5.06
4.63
4.34
3.25
2.47**
5.89
7.69
5.47**
10.39
7.18
6.24**
5.50
4.22**
7.07
6.17
5.55
4.98
7.19
7.82
5.59
7.52
Standard —
Error (ppm)
< .009 GM
< .025 GM
< .025
< .020
< .025
< .030
< .030
< .035
< .030
< .025
< .020
< .015
< .010 GM
< .025
< .015 GM
< .015
< .020 GM
< .035
< .020
< .055
< .050
< .030
< .030
< .035
< .030
NT = Factor not tested because of sample size (percent positive)
requirements given in section 3.2.
NS = Not significant at the 5% level (less than 95% confidence).
* = Factor significant at the 5% level (95% confidence level).
** = Factor significant at the 1% level (99% confidence level).
I/ Standard Error of the Geometric Mean given as its proportion
of the mean. For examnle, if the geometric mean is .236 ppm
s SE < (.03)(.236 ppm) = .00708 ppm.
-------
-36-
TABLE 3.6
Summary Statistics for g-BHC
Factor
ALL
YEARS
AGE
RACE
SEX
CENSUS
DIVISIONS
Factor
Levels
1970
1971
1972
1973
1974
1975
1976
1977
0-14
15-44
> 45
White
Black
Other
Male
Female
E. No.
Central
E. So.
Central
Mid-
Atlantic
Mountain
New
England
Pacific
So.
Atlantic
W. No.
Central
W. So.
Central
N
9075
1387
1570
1889
1094
905
784
683
763
1306
2745
5024
7642
1343
90
4735
4340
1915
710
2156
263
321
1029
1058
682
941
Percent
Positive
96.6
98.6**
98.7
91.9
98.9
98.1
97.1
96.7
94.3
NT
89.3
97.7
97.8
NT
96. 7m
96.2
94.4
NT
96.0 ^
97.2
NT
96.1
94.2
97.2
98.0
99.0
94.6
98.7
95.8
,96.9
Geometric
Mean (ppm)
.272
.381**
.353
.302
.264
.221
.201
.197
.161
.116**
.222
.372
.267**
.367
.242
.268*
.276
.208**
.382
.237
.258
.187
.244
.383
.273
.441
Standard —
Error
< .001
< .020
< .020
< .020
< .025
< .025
< .030
< .030
< .030
< .020
< .015
< .010
< .010
< .020
< .075
< .015
< .015
< .020
< .030
< .020
< .050
< .040
< .025
< .025
< .030
< .025
(ppm)
GM
GM
GM
GM
GM
NT
NS
*
**
I/
Factor not tested because of sample size (percent positive)
requirements given in section 3.2.
Not significant at the 5% level (less than 95% confidence).
Factor significant at the 5% level (95% confidence level).
Factor significant at the 1% level (99% confidence level).
Standard Error of the Geometric Mean given as its proportion
of the mean. For example, if the geometric mean is .236 ppm
then SE < .03 GM implies SE < (.03)(.236 ppm) = .00708 ppm.
-------
-37-
TABLE 3.7
• Summary Statistics for Dieldrin
Factor
ALL
YEARS
AGE
RACE
•
SEX
CENSUS
DIVISIONS
Factor
Levels
1970
1971
1972
1973
1974
1975
1976
1977
0-14
15-44
> 45
White
Black
Other
Male
Female
E. No.
Central
E. So.
Central
Mid-
Atlantic
Mountain
New
England
Pacific
So.
Atlantic
W. No.
Central
W. So.
Central
'
N
9100
1394
1575
1899
1097
904
783
683
765
1315
2753
5032
7662
1347
91
4751
4349
1925
713
2156
264
331
1033
1064
679
945
Percent
Positive
96.8
95.6**
98.2
97.5
98.5
98.5
95.0
93.4
95.1
NT
91.5Wi
97.3
98.0
NT
96.9Ni
96.5
93.4
96 7NT
97.0
97.0NT
95.0
97.6
98.1
97.8
96.4
96.2
96.3
96.8
Geometric
Mean (ppm)
.169
.197**
.222
.189
.181
.150
.141
.116
.099
.106**
.156
.199
.163**
.214
.158
.184**
.155
.166**
.201
' .171
.130
.133
.138
.199
.184
.178
Standard —
Error (ppm)
< .008 GM
< .020 GM
< .020
< .020
< .025
< .025
< .030
< .030
< .030
< .025 GM
< .015
< .015
< .009 GM
< .025
< .080
< .015 GM
< .015
< .020 GM
< .030
< .020
< .050
< .045
< .025
< .025
< .030
< .025
NT = Factor not tested because of sample size (percent positive)
requirements given in section 3.2.
Not significant at the 5% level (less than 95% confidence).
Factor significant at the 5% level (95% confidence level).
Factor significant at the 1% level (99% confidence level).
NS
*
**
I/
Standard Error of the Geometric Mean given as its proportion
of the mean. For example, if the geometric mean is .236 ppm
then SE < .03 GM implies SE < (.03)(.236 ppm) = .00708 ppm.
-------
-38-
TABLE 3.8
Summary Statistics for Heptachlor Epoxide
Factor
ALL
YEARS
AGE
RACE
SEX
CENSUS
DIVISIONS
Factor
Levels
1970
1971
1972
1973
1974
1975
1976
1977
0-14
15-44
> 45
White
Black
Other
Male
Female
E. No.
Central
E. So.
Central
Mid-
Atlantic
Mountain
New
England
Pacific
So.
Atlantic
W. No.
Central
W. So.
Central
N
9118
1390
1584
1897
1102
910
785
683
767
1320
2754
5044
7673
1354
91
4764
4354
1921
717
2162
264
322
1032
1069
684
947
Percent
Positive
93.8
94.1**
94.6
89.6
96.8
95.2
93.6
94.1
95.6
82.5NT
95.4
95.8
94.0NT
92.8
87.9
NT
93.5Ni
94.1
95.0NT
90.0
94.9
93.9
96.8
89.3
95.8
94.2
92.7
Geometric
Mean (ppm)
.092
.100**
.095
.100
.096
.081
.091
.085
.077
.061**
.081
.109
.091NS
.100
.078
.099**
.085
.092**
.105
.100
.076
.075
.063
.112
.100
.090
Standard —
Error (ppm)
< .008 GM
< .020 GM
< .020
< .020
< .025
< .022
< .025
< .030
< .025
< .020 GM
< .015
< .010
< .008 GM
< .020
< .080
< .010 GM
< .015
< .020 GM
< .030
< .015
< .050
< .040
< .025
< .025
< .030
< .025
NT = Factor not tested because of sample size (percent positive)
requirements given in section 3.2.
NS = Not significant at the 5% level (less than 95% confidence).
* = Factor significant at the 5% level (95% confidence level).
** = Factor significant at the 1% level (99% confidence level).
I/ Standard Error of the Geometric Mean given as its proportion
of the mean. For example, if the geometric mean is .236 ppm
then SE < .03 GM implies SE < (.03)(.236 ppm) = .00708 ppm.
-------
-39-
TABLE 3.9
Summary Statistics for Oxychlordane
Factor Percent Geometric Standard —
Factor Levels N Positive Mean (ppm) Error (ppm)
ALL 4218 97.2
YEARS 1970 1380 < 1.0"
1971 1571 53.4
1972 1893 91.8
1973 1092 98. 4=
1974 901 98.4
1975 779 96.4
1976 683 96.9
1977 763 95. 3_
AGE 0-14 806 88.5
15-44 -1554 99.3
> 45 1858 99.3
RACE White 3632 97.4
Black 533 96.8
Other 64 73.4
SEX Male .2055 96.6
Female ,2184 97.8
CENSUS E. No.
DIVISIONS Central 1003 96.4
E. So.
Central 283 97.5
Mid-
Atlantic 787 97.8
Mountain 181 88.3
New
England 169 98.8
Pacific 380 96.5
So.
Atlantic 563 97.8
W. No.
Central 319 98.4
W. So.
Central 533 95.8
.123 < .008 GM
NT .134** < .270 GM
.125 < .025
.130 < .015
** .126 < .020
.120 < .025
.120 < .025
.117 < .025
.113 < .025
NT .072** < .020 GM
.115 < .015
.148 < .015
NT .116** < .008 GM
.166 < .020
.126 < .080
NT .134** < .015 GM
.112 < .015 GM
NT .107** < .020 GM
.130 < .030
.137 < .020
.130 < .045
.101 < .040
.096 < .025
.141 < .025
.104 < .030
.158 < .025
NT = Factor not tested because of sample size (percent positive)
requirements given in section 3
NS = Not significant at the 5% level
2.
(less than 95% confidence) .
* = Factor significant at the 5% level (95% confidence level) .
** = Factor significant at the 1% level (99% confidence level) .
I/ Standard Error of the Geometric
Mean given as its proportion
of the mean. For example, if the geometric mean is .236 ppm
then SE < .03 GM implies SE < (.03)(.236 ppm) = .00708 ppm.
-------
-40-
TABLE 3.10
Summary Statistics for Trans Nonachlor
Factor
ALL
YEARS
AGE
RACE
SEX
CENSUS
DIVISIONS
Factor
Levels
1970
1971
1972
1973
1974
1975
1976
1977
0-14
15-44
> 45
White
Black
Other
Male
Female
E. No.
Central
E. So.
Central
Mid-
Atlantic
Mountain
New
England
Pacific
So.
Atlantic
W. No.
Central
W. So.
Central
Percent Geometric
N Positive Mean (ppm)
2225 97.2
1380 0"
1559 0
1885 0
1092 < 1
902 5.2
779 97.2
683 97.2
763 97.2
456 89.7
813 99.1
956 99.2
.1896 97.6
292 96.2
37 83.8
1118 96.9
1107 97.6
. 572 92.1
152 96.2
435 92.9
110 87.3
117 91.6
168 96.2
258 90.6
153 91.9
260 92.2
.119
NT
.106**
.114
INS .121
.135
.104
NT .067**
.112
.158
NT .113**
.158
.146
NT .129**
.109
NT .095**
.150
.133
.097
.093
.088
.135
.124
.172
Standard —
Error (ppm)
< .008 GM
____
< .680 GM
< .010
< .030
< .030
< .025
< .035 GM
< .025
< .025
< .020 GM
< .045
< .125
< .025 GM
< .025
< .030 GM
< .055
< .035
< .070
< .065
< .055
< .045
< .056
< .045
NT = Factor not tested because of sample size (percent positive)
requirements given in section 3.2.
NS = Not significant at the 5% level (less than 95% confidence).
* = Factor significant at the 5% level (95% confidence level).
** = Factor significant at the 1% level (99% confidence level).
I/ Standard Error of the Geometric Mean given as its proportion
~~ of the mean. For example, if the geometric mean is .236 ppm
then SE < .03 GM,implies SE < (.03)(.236 ppm) ..= .00708 ppm.
-------
-41-
TABLE 3.11
Summary Statistics for Hexachlorobenzene
Factor
ALL
YEARS
AGE
RACE
SEX
CENSUS
DIVISIONS
Factor
Levels
1970
1971
1972
1973
1974
1975
1976
1977
0-14
15-44
> 45
White
Black
Other
Male
Female
E. No.
Central
E. So.
Central
Mid-
Atlantic
Mountain
New
England
Pacific
So.
Atlantic
W. No.
Central
W. So.
Central
Percent Geometric
N Positive Mean (ppm)
3126 92.9
1380 0
1563 7.0
1894 42.7
1095 70.6
901 93 . T
779 91.5
683 93.4
763 92. 9_
650 80.5
1148 96.9
1328 95.6
2672 93.7
408 89.2
46 78.3
1562 92.1
1564 93.7
752 92.1
188 96.2
610 92.9
150 87.3
143 91.6
242 96.2
408. 90.6
248 91.9
385 92.2
.041
NT
.045**
.043
.037
NS .039
.044
.042
.042
NT .040**
.039
.043
NT .041*
.041
.047
NT .044**
.039
NT .038**
.036
.045
.047
.036
.077
.030
.034
.041
Standard —
Error
< .009
_ — _
< .060
< .025
< .025
< .025
< .025
< .025
< .025
< .025
< .015
< .015
< .010
< .030
< .090
< .015
< .015
< .020
< .035
< .020
< .045
< .045
< .030
< .030
< .035
< .025
(ppm)
GM
GM
GM
GM
GM
GM
NT = Factor not tested because of sample size (percent positive)
requirements given in section 3.2.
NS = Not significant at the 5% level (less than 95% confidence).
* = Factor significant at the 5% level (95% confidence level).
** = Factor significant at the 1% level (99% confidence level).
_!/ Standard Error of the Geometric Mean given as its proportion
of the mean. For example, if the geometric mean is .236 ppm
then SE < .03 GM implies SE < (.03)(.236 ppm) = .00708 ppm.
-------
-42-
REFERENCES
[3.1] Yobs, A.R., "The National Human Monitoring Program for
Pesticides", Pesticides Monitoring Journal, Vol. 5, No. 1,
pp. 44-46, June 1971.
-------
-43-
4. NATIONAL SOIL RESIDUE NETWORK
4.1 Detailed Description of Network
4.1.1 Sampling Procedures
The National Soil Residue Network is a subsample of the
Conservation Needs Inventory (CNI) Survey {4.1] and [4.2]. The CNI is
briefly described in the following paragraphs.
a. Target Population
The target population for the CNI study included all
soils of the conterminous United States. The frame was constructed of
every county of the conterminous United States except those strictly
metropolitan in character. Metropolitan areas include: Cities, villa-
ges, and other built-up areas of more than ten acres; industrial sites,
railroad yards, cemeteries, airports, golf courses, shooting ranges,
institutional and public administrative sites; and similar types of
areas. This separation did not necessarily include all land inside city
and village limits, and did include some land outside of such limits.
Where the acreage covered by roads and railroads is large enough to be
significant, it will be included under this item.
b. Sampling Rate and Size of Sample Area
The basic sampling rate was two percent. As standard
procedure, the statistical laboratories selected two sets of sample
areas without replacement for each county. Each set represented approxi-
mately two percent of the county. The majority of the counties in the
United States range from 250,000 to 500,000 acres. In these counties
only one set of samples was mapped. To maintain the same degree of
reliability in the data, the rate of sampling was reduced in larger
counties and increased in smaller ones. The size of sample areas ranges
from 40 to 640 acres; the sampling rate from less than one percent to
eight percent.
c. Selecting Sample Areas
The laboratories used the following procedure in select-
ing sample areas. The county was divided into blocks called "strata".
These were then subdivided into 48 (at Iowa State) or 49 (at Cornell)
equal-sized sample areas. One sample area was selected at random from
each stratum for the sample. One additional sample area was selected in
-------
-44-
each stratum to serve as a second 48-area sample. Farm boundaries were
ignored, but in sectionized areas each stratus consisted of twelve
sections (one-third of a township) and each sample area was a quarter-
section (160 acres). Sample areas in states of the northeast were 100
acres in size. Nationwide, the sample area size ranged from 40 acres to
640 acres.
Sample areas were located on county base maps. In mapping soils on
sample plots, sample areas in urban and built-up areas were classified
as to land use only and were not soil surveyed. Federal land identified
on the sample area map generally was not mapped. If a federal land
included cropland farmed under lease or permit and if a total acreage of
federal land could be obtained, that falling in sample areas was soils
mapped. Land use was mapped on all sample areas.
In measuring acreages on sample area maps, each individual soil
mapping unit was measured separately so that data could be combined by
land capability units or other desired interpretive groupings. The
soils data from the source areas were interpreted in terms of land
capability according to standards in use locally by the SCS in the
National Cooperative Soil Survey and in soil conservation district
operations. Land use and other information was reported at 18 to 25
points within the 100 acre units with approximately 38 to a 160 acre
unit. The points were randomly selected with a template prepared for
this purpose.
d. 1967 Selection of Sample Sites to Determine Pesticide
Residue Levels
The land use and treatment needs part of the CNI data
for each county was based on inventory collected by field inspection of
stratified random sample areas. The 1958 samples developed at Iowa
State University and Cornell University were used. For the 1967 inven-
tory, sample area patterns in all counties were reexamined by the Iowa
State University Statistical Laboratory to determine that coverage was
adequate. The laboratory drew new or additional samples for areas not
adequately covered, such as intensively developed irrigated areas, to
get an acceptable degree of precision.
-------
-45-
e. Selecting Soil Samples
To select a soil sampling site from the CNI sites, the
sample points in a sample site were divided into two land-use categories:
Cropland: includes land in corn, wheat, other grain, soybeans, hay,
vegetables and potatoes, orchards, sugar beets, sugar cane, tobacco,
cotton, and other crops. This area covers about 400 million acres.
Noncropland; includes woodland, pastures, grazing land and all other
lands not defined as cropland. This area covers about 1-1/2 billion
acres.
For each CNI site in a State, a record was made of the total number
of sample points and the number in each of the categories cropland and
noncropland to make this total. These numbers were used to provide the
ratios of cropland and noncropland to total in each sample site expressed
as a 5 place decimal. In those cases when the sampling rate deviates
from a 2 percent rate, a multiplier was applied to these ratios to make
the total equivalent to a 2 percent rate. A listing was made in some
preassigned order of the CNI sample location designations with the
associated numbers and ratios described above. The ratios were accumu-
lated in the order of the listing and added to the same listing. The
final values in each of the cropland and noncropland accumulations
should provide an estimate of the relative amounts of cropland and
noncropland in the state.
Cropland was sampled at 0.025% or one 10-acre block for every
40,000 acres of cropland. This provides for 9,468 cropland sample
sites. Noncropland was sampled at a rate of 0.0025% or one-tenth of
that for cropland. This provided for 3,832 noncropland sample sites.
A random method of selecting sample segments from the CNI was
devised. A method of selecting a 20-acre site within the CNI segment
was applied with the aid of data available for points within the seg-
ment. Two points are selected about which a 10-acre sampling site was
located for the soil monitoring program. The method of selecting the
samples with the 10-acre sampling sites is outlined by Wieroma, Sand
and Cox [4.3].
f. Generalization of the Results
Even though the above paragraphs describe the sampling
plan in general terms and in some specific detail, there is no reference
-------
-46-
to the probability framework. The selection probabilities of all the
elements in the population, along with the selection procedure, must be
known before the results can be generalized to any population other than
the specific sites. •
Before the results can be generalized, additional information must
be known. For example, as follows:
the complete frame from which the sample was selected,
the number of sample sites by State, by County, by Cropland
and Noncropland, and by all other stratification variables,
and
- the random method or methods of selecting the sample segments
from the CNI.
RTI has asked the statistical Laboratory at Iowa State University,
Ames, Iowa, for specific information about the sampling plan.
If detailed information is not available, it may be desirable to
develop a new area probability sample design following specification of
the population of interest; e.g., the entire U.S., certain urban and
rural areas, cropland, noncropland, areas subjected to specific pesti-
cide, etc.
4.1.2 Data Analysis
a. Introduction
The nationwide monitoring program, designed to determine
pesticide residue levels in agricultural soil and crops, became fully
operational in 1968. In accordance with the program sampling design,
a total of 1533 sites were scheduled for both soil and crop (mature)
sampling on an annual basis. Through this network of sampling sites,
pesticide residues were monitored during the years FY 1969 through FY
1974 (excluding FY 1971) at which time the program was discontinued.
Although every State was scheduled for sampling each year, some States
were excluded because of budgetary constraints. These were the larger
western States and some of the small grain producing mid-western States.
Soil and crop samples were collected and analyzed, however, from 33
States in each year the monitoring program was operating — these are
shown in Figure 4.1. Pesticide residue levels and related information
reported in this section apply only to the States shown in Figure 4.1.
Restricting the analysis to data from those States sampled in every test
-------
-47-
year enables one to make comparisons by time periods (year) that are
free of location effects, but the results of the comparisons are no
longer suitable for national-level inference.
This report covers an analysis of residue levels of PCB and seven
pesticides found in cropland soil during the period from 1968 (FY69) to
1973 (FY74) when the soil monitoring program was discontinued. These
pesticides were selected by EPA as being the most important pesticides
to examine and include: aldrin, dieldrin, heptachlor, heptachlor epoxide,
2JDDT, lindane, and parathion. A very large number of the soil sample
analyses indicated these compounds to be below the minimum detectable
level and the observations are shown as zeros in the data file. As a
result, each compound is examined for the percentage of occurrence
(i.e., number of positive detections expressed as a percentage) and how
this percentage changes with time. In addition, the geometric mean of
the positive values for each compound is examined to see if and how it
changed over the years from 1968 to 1973.
b. Percentage of Occurrence
Table 4.1 gives, by year and by compound, the number of
soil samples analyzed, the number of samples in which the residue level
was positive, and the percentage of positive detections. A plot of the
latter statistic is given in Figure 4.2. For three of the compounds —
lindane, parathion, and PCB — the number of positive detections were so
small that percentages were not calculated and plotted.
Aldrin, dieldrin, heptachlor and heptachlor epoxide show a similar
pattern with respect to percentage of occurrence over time. In all
cases, the estimated percentage of occurrence was slightly higher in
FY70 than in FY69. This was followed by a steady decline (see Figure
4.2) in the percentage of occurrence over the next three test years —
FY72, FY73, and FY74. The null hypothesis of a common percentage over
all years was tested using a chi-square (see Section 2 for a discussion
of this technique) statistic. As evidenced by the chi-square values
given in Table 4.1, this hypothesis was rejected for each compound.
This is not surprising since small changes in sample percentages can be
statistically significant with sample sizes of approximately 1400 as
employed in the soil monitoring program.
-------
-48-
In the case of JjDDT the estimated percentage of occurrence de-
creased from 27.8 percent in FY69 to approximately 23.6 percent in FY70
and FY72. There followed another decline to 21.7 percent for the years
FY73 and FY74. The later decline could be reflective of the restriction
placed on DDT in December 1972. A chi-square test rejected the null
hypothesis of a common percentage of occurrence over time. Apparently
this was not due to the higher percentage observed during FY69, although
this was not tested.
c. Geometric Means
The estimated geometric mean and associated standard
error for aldrin, dieldrin, heptachlor, heptachlor epoxide and £^DDT are
shown in Table 4.1 for each test year. Plots of the geometric means are
shown in Figure 4.2. An analysis of variance was used to test for
significant differences among yearly means. This statistical procedure
is discussed in Section 2. The resulting F-values were non-significant
for aldrin and heptachlor while the other three pesticides showed signifi-
cant differences over time (.see F-values in Table 4.1)- In soil samples
where measurable amounts of dieldrin and ^DDT were found, the levels of
residue increased from FY69 to FY72 then showed a decline in the last
two test years. This is in contrast to a general decline in the per-
centage of soil samples having a measurable amount of these pesticides.
In the case of heptachlor epoxide, the estimated geometric mean appears
to level off in FY74 after a steady increase from FY69.
Regulatory action by EPA on aldrin and dieldrin occurred after the
soil monitoring program was discontinued; hence, its effect cannot be
evaluated directly from data generated in this program. There is only
one year of test data on DDT following the regulatory action taken in
late 1972. A before/after comparison (FY 1973 vs. FY 1974) showed no
change in the percentage of positive detection but a significant de-
crease (from .221 to .107 ppm) in the geometric mean of positive values.
This is a continuation of a decline that apparently began two years
earlier. There is no evidence of an increase in the levels of parathion
(a recommended substitute for DDT) during this time period.
-------
-49-
Figure 4.1. States where agricultural soil and crops were sampled for
pesticides during 1968-73 — National Soils Monitoring
Program, U.S. Environmental Protection Agency.
-------
Table 4.1
1 Summary Statistics and Tests of Significance
Year Sample
(FY) Size
1969
1970
1972
1973
1974
1440
1388
1401
1403
1388
ALDRIN
Positive Detections
No.
184
196
139
129
51
%
12.8
14.1
9.9
9.2
3.7
2 /
GM—
0.057
0.059
0.062
0.059
0.067
Standard
I/
Error -
0.006
0.006
0.008
0.008
0.014
Test of Significance Among:
2
Percentages (x ) = 101.6**
GM (F values) = 0.18
Year Sample
(FY) Size
1969
1970
1972
1973
1974
1440
1388
1401
1403
1388
HEPTACHLOR EPOXIDE
Positive Detections
No.
128
141
102
95
82
%
8.9
10.2
7.3
6.8
5.9
GM
0.027
0.031
0.038
0.041
0.038
Standard
Error
0.002
0.002
0.003
0.004
0.004
DIELDRIN
Positive Detections
No.
436
441
395
397
354
% GM
30.3 0.055
31.8 0.074
28.2 0.086
28.3 0.078
25.5 0.055
Standard
Error
0.003
0.004
0.005
0.005
0.004
15.3**
12.04**
Positive Detections
No. % GM
400 27.8 0.189
327 23.6 0.267
332 23.7 0.305
305 21.7 0.221
302 21.8 0.107
Standard
Error
0.017
0.026
0.030
0.022
0.011
Test of Significance Among:
Percentages (x ) =
GM (F values)
22.6**
19.1**
HEPTACHLOR
Positive Detections
No.
66
95
73
57
25
4.6
6.8
5.2
4.1
1.8
GM
0.039
0.041
0.046
0.031
0.030
Standard
Error
0.006
0.005
0.007
0.005
0.008
43.5**
1.11
No. Positive Detections
3/
Lindane Parathion— PCB
14
6
0
0
0
6
0
3
7
0
0
0
0
2
24
i
Ul
o
3.55**
16.38**
** = Significant at .01 level of significance
I/ = Based on pooled within year variance
2/ = GM = Geometric Mean (ppm) of dry weight.
_3_/ Sample sizes for years 1969-1974 are respectively
59, 4, 1069, 1152 and 939. Analyses for organo-
phosphorous compounds were not made on all soil
samples.
-------
-il-
Figure 4.2
Percentage of Occurrence and Geometric Mean by Years
(a) Aldrin
a
0
ed
32.0'
30.0
28.0
26.0.
tl 24.0-
(b) Dieldrin
V
\
.09
.08
I
.07 g
(U
a
0)
.05
—I 1
1971 1972
Year (FY)
1969
1970
1973
1974
Percentage of Occurrence
Geometric Mean
-------
Figure 4.2 (cont'd)
-52-
7.0 ,
(c) Heptachlor
/ \
-------
-53-
Figure 4.2 (cont'd)
(e)
'
28.0 •
I 26.0-
c
OJ
g 24.0-
o
0 22.0-
0)
00
«
4J
S 20.0 •
u
& <
s /
.
\- — "*\
_—•*""'" x
x
N
^*^ • . \ ^
\
\
\
r
<
s
• .309
a
a
^^
c
efl
(U
S
0
OJ
o
0)
• .10
>
1 • I » 1 V
1969 1970 1971 . 1972 1973 1974
Year (JY)
— — — Geometric Mean
-------
-54-
REFERENCES
[4.1] Iverson, Leo G.K., 1968. Soil monitoring program - sampling
design. United States Department of Agriculture, Code ARS-PPCP-2.
[4.2] National Handbook for Updating the Conservation Needs Inven-
tory, August 1966. United States Department of Agriculture.
[4.3] Wiersma, G.B., P.P. Sand and E.L. Cox, 1971. A sampling
• design to determine pesticide residue levels in soils of the con-
terminous United States. Pesticides Monitoring Journal. Vol 5:1,
pp. 63-66.
-------
-55-
5. NATIONAL SURFACE WATER AND SEDIMENT RESIDUE NETWORK
5.1 Description of the Network
5.1.1 Sampling Procedures
Sampling points for the National Surface Water and Sediment
Residue Network were selected from the National Stream Quality Account-
ing Network (NASQAN). NASQAN is a series of stations at which systema-
tic and continuing water quality measurements are obtained. The United
States was subdivided into 27 major basins by the Water Resources
Council; these, in turn, were subdivided into 324 accounting units.
Stations were selected within each unit to provide a measure of dis-
charge and water quality for approximately 90 percent of the water
leaving the unit. For units along the periphery of the country where it
is impractical to meet the 90 percent outflow goal with measurements at
a reasonable number of stations (coastal regions, Great Lakes, inter-
national boundaries), a "representative" station or array of stations
was selected that could serve as an index for estimating the outflow
from the unit. The 324 hydrologic units are based on watershed con-
figurations and convenience of unit size. Stream-flow and water-quality
stations were selected to provide flow and quality information for each
unit independent of actual or suspected water-quality conditions.
In order to accomplish the objectives of the water and sediment
residue monitoring program, sampling of approximately one-half of the
stations of the National Network (NASQAN) was planned. Accordingly, one
station was selected from each of the odd-numbered accounting units
within each of the 27 major drainage basins. Exceptions to this selec-
tion process were made where there were no surface waters in a given
odd-numbered unit. In those instances, a station was selected from an
adjacent even-numbered unit. To complete the network, two to three
stations each in Alaska, Hawaii, and Puerto Rico were selected at random
from stations listed in the water quality section of the Office of Water
Data Coordination Catalog of Information on Water Data. There may be
additional restrictions and/or rules on the selection of stations for
the Water and Sediment Monitoring Network; background information is
still being sought by RTI.
-------
-56-
Given what is now known.the network can best be described as a
purposively selected sample of stations. There is no known probability
structure for the sample selection; therefore, even though there is a
broad geographical coverage of the United States, the only totally de-
fensible inferences that can be made are restricted to the stations in
the network, although selecting odd numbers may be reasonably random,
depending on the purpose and system of numbering.
For future water and sediment monitoring, a probability sample of
sites could be selected following specification of the population of
interest.
5.1.2 Data Available On Data File
The water and sediment residue file, after modification to
allow analysis, contained site and date information, and data on twenty-
four compounds thought to be of most interest. PCB's were originally
classified as 1242, 1248, 1254, or 1260, with the overwhelming majority
being 1254. For purposes of analysis, an occurrence of any of these
types was treated as simply "PCB". The six DDT residues on the file
were combined using the rough formula
= DDT (o.p1) + DDT (p,?1)
+ 1.11 [DDE (o.p1) + DDE (p.p1) + DDD (o.p1) + ODD (p.p1)].
After these conversions, the data were reduced to sixteen pesticides.
Sites from all water basins in the nation are on the file, although
some basins are represented sparsely, particularly those on the west
coast.
File entries were assigned by month code to season 1 or 2 for sedi-
ment, and season 1, 2, 3, or 4 for water, corresponding to the supposed
sampling intervals. For water, approximately 144 sites appear in each
season's data, from the last quarter of 1976 through the second quarter
of 1979; a small sample of 39 sites is present for the third quarter of
1976. Sediment data, varying from 39 to 112 entries, is present for
1976 through the first half of 1979.
Many sites contain "missing values" for all or some of the com-
pounds. It is understood that most of these represent data for which
tests are incomplete. Table 5.1 shows the location of these gaps.
-------
-57-
Table 5.1
Periods for Which Data is Missing
Compounds Water Dates Sediment
Aldrin, Chlordane, No large gaps Before July,
Dieldrin, Hepta- 1976
chlor, Heptachlor
Epoxide, Lindane,
Parathion, PCB,
DDT, Toxaphene
2,4-D, Silvex,
2,4,5,-T
Atrazine
Oct, 1978 -
present
April, 1978 -
present
early 1976
and July,
1978-present
July, 1978-
present
For the previous draft of this report, much of the 1978 data was
missing. It now has been keypunched and added to the file. However,
reanalysis is still in progress, and most of the new results will be
found in the addenda.
5.1.3 Assumptions Made in Analysis of the Network
The assumptions made in combining PCB's and DDT residues are
discussed above. Reported levels were assumed to be in parts per
billion, with a zero entry indicating tested for but below the detec-
table level.
In the initial analysis, PCB levels in water were left as they were
on the original file, although over 90 percent of the detections were
reported with values less than 15 parts per billion. Because of the
questionable validity of observations below .15 ppb, tests and statis-
tics for PCB's have been redone, changing these low numbers to half the
minimum detectable level (.075).
5.2 Data Analysis
5.2.1 Summary Statistics
For a given year or season, the typical distribution of a
compound's concentrations in water or sediment is a large majority of
zero values and a few, if any, positive detections. As discussed in the
section dealing with all the networks (Section 2), the proportion of
-------
-58-
positive values in the sample and the geometric mean of those values
were considered to be the best descriptive statistics for the file.
With some of the compounds, the number of observations above a
certain level may be a useful statistic. These results appear in the
addenda.
The nature of the distribution of the positive values was investi-
gated using normal probability paper. Atrazine in water for 1977 (the
frequencies are given in Section 2 of this report) and ^DDT in sediment
for 1976 were chosen because they had a reasonable number of positive
detections. In both cases, a plot of the original values yielded a poor
fit to the normal distribution; a plot of the logarithms of those
values produced a very good fit. This supports the assumption that the
positive values are distributed approximately lognormally, and therefore
that the geometric mean is an appropriate statistic.
In the original analysis, because the data for one year represented
different seasons than did the data for any other (due to the large gaps
in the file), it was thought best to report the summary statistics by
seasons within the years, rather than over entire years. These statis-
tics are found in Tables 5.2 and 5.3, but now include 1978 data. Sum-
mary statistics by year may be found in the addenda.
Only atrazine and PCB in water, and chlordane, dieldrin, PCB and
J^DDT in sediment had enough positive detections to justify giving the
full set of statistics. For the remaining compounds, if there were more
than two detections in the entire file, the percent positive detections
is listed. Included in these compounds are toxaphene and parathion,
with only a handful of positive detections. Other DDT substitutes were
undetected or not on the file. Percent detected and geometric means are
presented for each water basin in Tables 5.5 and 5.6. Because some
basins had very few observations, the data for the entire period of
monitoring were grouped together.
5.2.2 Tests for Trends Over Time
To prevent bias caused by possible seasonal variation, tests
for trends over time were done using only the seasons for which data
were present in each year being compared. For sediment, before we had
data from the first half of 1978, this meant that the only possible
tests would compare the second halves of years 1976, 1977, and 1978.
-------
-59-
For atrazine in water, both the third and fourth quarters could be used
in comparing 1976 and 1977 levels; PCB levels were compared two diffe-
rent ways, as shown in Table 5.4.
2
For each compound, a x test was performed on the proportions of
samples with positive detections, and a one-way analysis of variance was
used to test for differences in the geometric means over time. No
tests for trends were conducted for individual water basins because of
the sparseness of the data. Additional tests utilizing the 1978 data
now available are presented in the addenda.
5.2.3 Observed Trends
2
The results of the x and ANOVA tests are shown in Table
5.4. The only significant difference from year to year in the geometric
mean of detected values was that of PCB in sediment, from 1976 to 1978.
In general, it can be seen in the plot of geometric means over time
(Figures 5.1 and 5.2) that the differences are not in the nature of a
trend, but rather show an erratic, up and down behavior with the excep-
tion of PCB in water (the geometric mean of which varies little over
time).
2
The x test of percent of occurrence tells a different story. With
the exceptions of atrazine in water and PCB in sediment, the proportions
of detected values were significantly different from one year to the
next in the compounds showing up with any substantial frequency. From
graphs, it appears that in all these cases — PCB in water; chlordane,
dieldrin, and ^JDDT in sediment — the trend was down.
5.2.4 Geographic Differences
Because of the methodology used to select sites, any formal
statistical test comparing means or percent detected between water
basins would be of dubious validity. Nevertheless, a breakdown of the
data by basin may be of considerable interest, as can be seen from the
summary statistics (Tables 5.5 and 5.6). More extensive tables, incor-
porating 1978 data and modified PCB values, may be found in the addenda.
For the water file, the presence of PCB residues is almost constant
across the nation. However, atrazine appears only in the Mississippi,
Missouri, Ohio, and Great Lakes basins. In the sediment file, the
statistics are less dramatic, but may suggest to the informed eye prob-
lems with certain compounds in some regions.
-------
TABLE 5.2. Summary Statistics by Compound by Year and Season - Water - Over All Sampling Sites
ATRAZINE
PCB I/
Positive Detections
Year
1976
1977
1978
Season
3
4
1
2
3
4
1
2
Sample
Size
35
130
126
117
122
125
77
32
%
23
2
3
9
15
2
0
19
Geometric
Mean (ppb)
.82
.89
.30
1.28
.63
.48
1.01
Standard
Error
.15
.60
.08
.42
.09
.07
.37
Year
1976
1977
1978
1979
Season
3
4
1
2
3
4
1
2
3
4
1
2
Sample
Size
23
128
133
130
132
135
98
102
115
131
144
134
Positive Detections
%
43
58
23
78
5
9
7
1
1
2
2
4
Geometric
Mean (ppb)
.120
.083
.089
.086
.092
.090
.078
.086
.006
.081
.070
.069
Standard
Error
.038
.006
.014
.003
.018
.090
.015
.039
.006
.009
o
2/3/
OTHER COMPOUNDS - Percent Positive Detections——
Year
1976
1977
1978
1979
Season
3
4
1
2
3
4
1
2
3
4
1
2
Chlordane
0
0
1
2
0
1
0
0
0
0
0
1
2.4-D
3
2
1
2
0
1
0
3
1
Dieldrin
0
2
3
3
1
1
0
0
0
0
0
0
Heptachlor
Epoxide 2,4,5-T Toxaphene Diazinon Parathion
03000
10000
30030
40230
00203
00030
00010
00400
04101
0 000
0 000
0 001
I/ Values below .15 ppb left as originally reported. Reanalysis in addenda.
2J A blank indicates data missing or not yet analyzed.
3/ Based on sample sizes approximately the same as for Atrazine and PCB.
-------
TABLE 5.3. Summary Statistics by Compound by Year and Season - Sediment - Over All Sampling Sites
CHLORDANE
Positive Detections
Year
1976
1977
1978
1979
Year
1976
1977
1978
1979
Season
2
1
2
1
2
1
Season
2
1
2
1
2
1
Sample
Size
87
105
70
81
81
39
Sample
Size
84
104
69
81
81
38
Geometric Standard
% Mean (ppb) Error
22 4.18 1.92
20 6.95 3.29
3 3.11 1.42
11 4.30 1.84
5 8.91 4.04
8 21.7 26.6
PCS
Positive Detections
Geometric Standard
% Mean (ppb) Error
26 18.0 6.5
19 38.3 10.3
30 2.2 .47
23 9.8 5.7
26 22.9 6.9
8 48.2 35.7
OTHER COMPOUNDS - Percent Positive Detections-
Year
1976
1977
1978
1979
2/
3/
Season
2
1
2
1
2
1
DDT + 1.
Based on
Heptachlor
Epoxide
6
1
1
1
1
0
11 (DDF + ODD)
sample sizes
Toxaphene Lindane
2 0
2 1
0 1
1 0
0 0
0 0
approximately the same as f<
Sample
Size
87
105
69
80
80
39
DIELDRIN
Positive Detections
Positive Detections
Geometric
Mean (ppb)
32
21
13
11
8
8
48
77
1.71
05
41
4.98
Standard
Error
.98
2.58
.43
.70
1.87
5.29
-------
-62-
Table 5.4
Tests for Trends Over Time - Water and Sediment
Test Using
Proportion,
Detected=x'!
Test Using
Geometric
Mean = F
Compound
Water;
Atrazine
PCB
n 1
3/
Sediment:
Chlordane
Dieldrin
PCB
£DDT
Seasons Compared
Seasons 3 and 4
1976 vs. 1977
Seasons 1 and 4
1977 vs. 1978
Season 1 1977
vs. 1978 vs.
1979
Season 2 1976
vs. 1977 vs.
1978
"
"
"
Significance
.608
18.2
31.9
32.6
35.5
1.29
18.2
N.S.I/
2/
< .01-'
< .01
< .01
< .01
N.S.
< .01
Significance
1.8 N.S.
.19 N.S.
.18 N.S.
.47 N.S.
1.98 N.S.
18.30 .01
.80 N.S.
I/ N.S. = not significant at the .10 level
4
2J .01 = significant at the .01 level.
3/ Values below .15 ppb left as originally reported. Reanalysis
in addenda.
-------
-63-
Table 5.5
Summary Statistics by Compound by Water Basin - Water (1976 - 1979
Basin
North Atlantic
Slope (1)
South Atlantic
Slope & (2)
Eastern Gulf of
Mexico
Ohio River (3)
St. Lawrence
River (4)
Hudson Bay & Upper
Mississippi River
(5)
Missouri River (6)
Lower Mississippi
River (7)
Western Gulf of
Mexico (8)
Colorado River (9)
The Great Basin (10)
Pacific Slope -
California (11)
Pacific Slope -
Washington (12)
Snake River (13)
Pacific Slope -
Oregon & Lower
Columbia River (14)
Alaska (15)
Hawaii (16)
Sample
Size
39
87
77
51
62
94
93
81
41
33
15
22
10
23
4
11
Atrazine
GM*(.ppb)
Percent of
Detected Detected
0
0
12 .44
2 .97
19 .84
13 1.25
17 .85
0
0
6 .49
0
0
0
0
0
0
Sample
Size
67
144
120
82
93
158
125
127
67
52
30
37
17
35
11
16
PCB -
Percent
Detected
27
19
20
23
18
23
24
19
18
21
17
22
18
20
36
25
Combined)-/
GM (ppb)
of
Detected
.09
.08
.10
.08
.09
.09
.08
.09
.07
.06
.25
.11
.11
.06
.09
.09
GM = Geometric Mean
I/ Not including recently added 1978 data. Reanalysis in addenda.
21 Values below .15 ppb left as originally reported. Reanalysis in addenda.
-------
Table 5.6
Summary Statistics by Compound by Water Basin - Sediment (1976 - 1979 combined).!/
Basin
N. Atlantic (1)
S. Atlantic
& E Gulf of
Mexico (2)
Ohio River (3)
St. Lawrence
River (4)
Hudson Bay &
Upper Miss. (5)
Missouri
River (6)
Lower
Mississippi (7)
Western Gulf
of Mexico (8)
Colorado
River (9)
Great Basin (10)
Pacific Slope
California (11)
Snake River (13)
Pacific Slope
Oregon (14)
Alaska (15)
Hawaii (16)
CHLORDANE
Sample
Size
23
53
37
27
33
65
41
37
23
18
6
6
8
2
3
Positive
Detections
Geo-
metric
Mean
% (ppb)
35 4.1
15 3.4
27 9.8
7 1.2
18 8.7
0
5 2.5
11 4.6
4 2.4
11 4.0
50 2.3
0
25 4.0
0
100 48.1
DIELDRIN
Sample
Size
23
53
37
27
33
65
41
37
23
18
6
6
8
2
3
Positive
Detections
%
26
13
35
7 .
12
5
7
14
4
6
33
33
38
0
67
Geo-
metric
Mean
(ppb)
1.17
.84
.87
.32
2.08
.28
.19
.67
.65
.12
.27
.74
.54
41.8
PCB
Sample
Size
23
53
38
27
30
65
41
37
22
18
5
6
8
2
2
Positive
Detections
%
26
17
53
63
17
11
15
19
5
33
20
17
25
0
0
Geo-
metric
Mean
(ppb)
17.2
6.9
61.7
22.8
13.0
2.5
3.1
7.3
3.2
11.4
7.3
174.
20.8
ZODT
Positive
Detections
Sample
Size
23
53
37
27
33
65
40
37
23
18
5
6
8
2
3
%
35
21
16
19
9
2
15
35
22
6
60
33
25
0
67
Geo-
metric
Mean
(ppb)
4.1
3.7
3.8
4.1
i
12 o
\
2.8
10.0
1.2
2.7
13.3
.9
15.3
5.9
95.1
I/ Not including recently added 1978 data. Reanalysis in addenda.
-------
Figure 5.1.
o
c
OJ
1-1
!-i
3
a
o
o
M-l
o
0)
o
20.
-65-
Percentage of Samples with Positive Detections and Geometric Means
Over Time - Water - All Sampling Sites
(a) Atrazine
fc 1.2
1.0 "a,
1976
H 1
2 3
1977
Geometric
Mean
Percent of
*0 Occurrence
1
. .8
• .6
.4
• .2
cd
01
a
0)
o
01
o
21
0)
a
c
-------
-66-
Figure 5.2. Percentage of Samples with Positive Detections and Geometric
Over Time - Sediment - All Sampling Sites
o
c
o
o
o
14-4
o
QJ
60
03
4-1
0)
O
(a) PCB
10
2/
,0
ft
ft
e
OJ
0)
S
i-i
•u
01
o
01
o
(b)
iDT
cu
o
cu
M
^i
3
O
a
o
oo
cfl
•u
a
0)
o
30
25 J
20 "
15 •
10 -
5 •
2
1976
T
I/
H
1 2
1977^
Very small sample size (< 40 sites)
1 2
1978
•6
-5
•4
-3
•2
-1.0
c
(0
O)
S
o
•H
h
4-1
Ol
01
o
2/
1
1979
I/ 1 and 2 refer to the first and second half of the year.
-------
-67-
Figure 5.2. Percentage of Samples with Positive Detections and Geometric M
(cont'd) Over Time - Sediment - All Sampling Sites
/
8 2°-
G
01
S-l
3
o i 5 _
u J--> •
o
M-t
O
01
M 10 -
4-J
C
O)
CJ
01
h^ j
. (c) Chlordane *
I/ •
*' "*\ /
\ /
\ /
\ /
\ /
\ /
\ /
\ j*
\ "VV^ ^ *
^<<-^*^ \ ^v^ *^ ^^x*"^
^^» >^ \ c&d^ %_0ccurrene€x'^
X--"" 3/
1 1 I 1 1
21 1 l' 21 1 l' 2 1 1
s.
" 20 S
a.
G
cfl
-15 |J
*
a
•H
- 10 1
o
01
VJ?
- 5
11
1976 1977 1978 1979
(d) Dieldrin
/
30 '
25 -
0)
o
e
01
M 20 -
3
O
CJ
O
4-1
o 15 -
0)
£
S 10 -
n
01
PH
5 -
•
\
\
\
\
\
\ ^
\ <^^ N
\ / N\
V* A.
/* /^
/• N.V-
/ \ ^ V"
/ \ v°
/ \ ^%
/ \ N ^
/ \ ^ ^*
/ \ ^ ^^^^
/ \ \ S^^
/ •-^f-2««rence*
"•" • —
3/ *
• i - ' . 1 1 i 1
• 2.0
x^\
42
a.
a,
X-'
•1.5 a
BJ
s
a
-i.o -H
4-1
0)
o
01
o
- .5
c\ /
[ 1 i [ ' 1 2 ' 1 *
1976 1977 1978 1979
Very small sample size
1 and 2 refer to the first and second half of the year.
3/
-------
-68-
REFERENCES
[5.1] Ficke, J.F. and R.D. Hawkinson. The National Stream Quality
Accounting Network (NASQUAN) - Some Questions and Answers. Geo-
logical Survey Circular 719.
[5.2] Pelty, H.R., W.T. Sayers and H.P. Nickolson, 1971. National
monitoring program for assessment of pesticide residues in water.
Pesticides Monitoring Journal, Vol. 5, pp. 54-62.
-------
-69-
6. AIR NETWORK
6.1 Detailed Description of Network
6.1.1 Sampling Procedures
Many investigators have studied the role of air as a vehicle
of pesticide exposure for the general population. Selected work in this
area is reviewed by Kutz, Yobs, and Yang {6.2]. A pilot study was
conducted in 1967 and 1968 by Midwest Research Institute for the Divi-
sion of Pesticide Community Studies of the Environmental Protection
Agency. The purpose of this study was to determine levels of 19 pesti-
cides and metabolites in ambient air at 9 locations in the U.S. The
sampling procedures at these 9 locations and the analytical methods used
are discussed by Stanley, Barney, Helton, and Yobs [6.4].
During calendar years 1970, 1971, and 1972, an air monitoring
program was initiated by the federal government. Samples of ambient air
were collected at selected locations in the U.S. and analyzed for pesti-
cide residues. In 1970, ambient air was collected at selected locations
in 14'states with an ethylene glycol impinger sampler. In 1971 and 1972,
samples were collected at selected locations in 16 states. The locations
for this study, including both rural and urban areas, were selected
primarily for their potentially high concentrations of pesticides in
ambient air. Additionally, factors such as electrical power source
needed for operating the sampler and accessibility of sampler for
servicing and ethylene glycol collection were considered in location
choice, according to Kutz, Yobs, and Yang [6.2]. At each sample loca-
tion, two samplers were operated simultaneously for 24 hours, at a
height representing the human breathing zone. A composite of the 4 12-
hour samples was formed for chemical analysis. Accompanying each sample
were certain climatological data collected from the nearest station of
the U.S. Weather Bureau. Sampling sites during" the course of this study
were not always constant. Relocations prohibited yearly comparisons for
those sites involved. A summary of the data collected through this
program is given by Kutz, Yobs, and Yang £6.2].
During 1975 an air monitoring program was established to determine
pesticide residues in ambient air in three suburban locations Miami,
Florida; Jackson, Missippi; and Ft. Collins, Colorado. At each loca-
-------
-70-
tion, duplicate samples were collected by two ethylene glycol impinger
samplers operating simultaneously side by side under identical condi-
tions, twice at each location in May and June and once in April, 1975.
Weather information obtained from the nearest National Weather Service
Station was recorded at the beginning and ending of sampling. In the
event of inclement weather, the sample was taken on the next day that
weather permitted. The analytical procedures used for these samples is
described by Kutz, Yobs, and Yang [6.2] and 16.3]. Selection of these
three suburban locations was based on the particular interest in the
detection of chlordane, heptachlor, and associated chemicals. Suburban
locales appropriate for measuring major chlordane uses include turf,
ornamental, and residential insect management areas. Results of this
1975 study are summarized by Kutz, Yobs, and Yang [6.2] and [6.3]. Kutz
and Yang [6.1] examine the data with respect to evidence of polychlori-
nated biphenyls.
The number of sampling sites included in the air monitoring program
was subsequently expanded. Throughout the remainder of the monitoring
period', from 3 to 10 sites were included, and from 50 to 100 samples per
year were analyzed. These sites, primarily suburban areas, were pur-
posively selected. Considering this selection process, it is not possi-
ble to generalize the results to any population other than the particu-
lar sites.
For future air monitoring, a probability sample of sites could be
selected following specification of the population of interest (e.g. the
entire U.S., certain suburban areas, agricultural areas subject to use
of specified pesticides, areas subject to mosquito control activity,
etc.). The times and within-site locations for sampling could be ran-
domly selected while considering such factors as seasonality concerns,
pesticide spraying in the area, weather conditions, and other factors
suspected of influencing the presence of pesticides in ambient air.
Additionally, consistently monitoring the same sites throughout the time
of the study, as opposed to observing different sites during each time
period, would allow comparisons over time on a site-by-site basis.
6.1.2 Data Available on Data File
For this report, data from the network monitoring pesticide
residues in ambient air were obtained from four reports (6.3, 6.5, 6.6,
-------
-71-
6.7) using data from the periods 4/75 - 6/75, 8/75 - 3/76, 11/76 -
11/77, and 1/78 - 10/78, respectively.
All four documents provide, for several sites, the levels of vari-
ous compounds detected for each date on which sampling took place, as
well as meteorological observations for that date. Twenty-four com-
pounds appear somewhere in the data, but different reports — and even
different sites within reports — tabulate different subsets of these
residues. Compounds tested appear in Tables 6.2 - 6.5.
The test sites also vary from one report to another. The time
period for which RTI has data from a given site is listed in Table 6.1.
Observations were made once or twice a month at each site, sometimes at
erratic intervals.
Table 6.1
Periods of Air Sampling by Site
Fort Collins, Colo. 4/75-3/76 Wheaton, 111. 12/76-9/77
Miami, Florida 4/75-3/76 Springfield, 111. 1/78-9/78
Jackson, Miss. 4/75-3/76 Midvale, Utah 11/76-9/77
Greenville, Miss. 11/76-9/78 Florida I/ 2/78-10/78
Harrisburg, Penn. 9/75-3/76 South Carolina I/ 2/78-9/78
Lafayette, Ind. 8/75-3/76 Montana I/ 4/78-9/78
Pasadena, Calif. 11/76-9/78
I/ Specific location unknown.
6.1.3 Assumptions Made in Analysis of the Network
Analysis of this file, for reasons discussed below, was
limited, and few assumptions were required for what was done. Values
listed as "Trace" were set to .02 or .1 ng/m when computing geometric
means, the specific value depending on the apparent scale of the posi-
tive observations. Where this arbitrary procedure had any substantial
effect on the result, the mean was listed as "?". In the first two
reports, two values were reported for each test site and date. It was
assumed that the average of the two numbers could be used without seri-
ous consequences.
-------
-72-
6.2 Data Analysis
6.2.1 Summary Statistics
Tables 6.2 - 6.5 give summary statistics for each of the
four reports. The number of positive values detected and the geometric
mean of those values were computed for each site, grouping together all
the test dates. The percentage detected was not presented because of an
ambiguity in the reports: it is unclear whether a blank entry means the
compound was tested for but not detected, or that it was never tested
for in that sample. The reasons for using geometric means given in the
general discussion (Section 2) apply here. Because of the small sample
sizes, standard errors were not computed.
6.2.2 Tests
The fragmentary nature of these data makes it unsuitable for
formal testing. No two geometric means here are truly comparable. In
one of the few situations where detected occurrences of a compound are
present at one site during two time periods, namely malathion at Green-
ville in 1977 and 1978, it can be seen (Figure 6.1(c)) that the variance
of the data is so large that any formal tests would be nonsignificant.
6.2.3 Trends
Although formal tests were not justified, it may be of
interest to examine the levels of a compound plotted over time, when
that compound is detected at a particular site. Examples of this are
given in Figure 6.1 for chlordane, dieldrin, malathion, and DDT. Chlor-
dane concentrations in Miami and Jackson appeared to decrease over FY
1976, as did dieldrin in Jackson. Data needed to assess long-term
trends were unavailable.
6.2.4 Geographic Differences
Figures 6.1(a) and (d) compare concentrations of chlordane
and DDT/DDD, respectively, for two sites. It is difficult to conclude
from these which site has an overall higher level of these residues. As
with testing for trends, there is too little data and too much variability.
-------
Table 6.2
Summary Statistics - Pesticides in Air - April-June, 1975
(ng/m )
Pesticide
Alpha-BHC
Lindane
Heptachlor
Heptachlor
Epoxide
Oxychlordane
Chlordane
Dieldrin
Beta-BHC
Diazinon
Parathion
Malathion
End r in
Disyston
PCB
Fort Collins, Colorado
Number of Pos. Geom. Mean
Detections of Pos. Det.
5 2.2
4 .7
2 .6
1 3.8
1 .2
5 67.9
5 sampling dates
Miami, Florida
Number of Pos. Geom. Mean
Detections of Pos. Det.
5
5
5
0
3
4
4
4
3
4
1
5
.9
1.3
1.9
.8
18.2
.9
1.0
1.8
2.0
1.0
31.5
5 sampling dates
Jackson, Mississippi
Number of Pos.
Detections
5
5
5
1
1
5
5
1
4
2
1
1
4
5 sampling
Geom. Mean
of Pos. Det.
1.9
2.2
9.0
.8
.8
30.8
9.7
1.0
.9
.25
.5
.9
10.8
dates
LO
I
NOTE: Most of the sample points were doubled - i.e., there were two
detectors at each site. The average value was used here.
-------
Table 6.3
Summary Statistics - Pesticides in Air - August 1975 - March 1976
(ng/m )
Pesticide
Alpha-BHC
Lindane
Heptachlor
Heptachlor
Epoxide
Oxychlordane
Chlordane
Dieldrin
Beta-BHC
p p^DDE
Diazinon
Parathion,
Methyl
Malathion
Endrin
RGB
Thimet
PCB
Fort Collins,
Colorado
No . of Geom.
Pos. Mean
Detec- of Pos.
tions Det.
8 1.3
5 .2
7 .1
2 <.l
5 .8
2 .2
3 .1
5 .5
0
7 .1
8 3.8
8 sampling
dates
Miami ,
Florida
No. of Geom.
Pos. Mean
Detec- of Pos.
tions Det.
10 1.0
10 1.0
10 1.1
0
8 .6
10 9.2
10 .4
5 .2
10 .9
5 2.6
8 1.9
5 TR
10 4.9
10 sampling
dates
Jackson,
Mississippi
No. of Geom.
Pos . Mean
Detec- of Pos.
tions Det.
9 1.1
9 .9
9 3.0
4 .7
8 .4
9 4.2
9 2.4
1 .9
6 .1
5 .9
3 2.8
1 9.0
1 2.5
7 <.l
1 7.4
9 3.4
9 sampling
dates
Lafayette,
Indiana
No. of Geom.
Pos . Mean
Detec- of Pos.
tions Det .
9 1.1
8 .2
9 .9
4 .8
8 5.4
9 .7
2 .5
4 .1
6 <.l
9 4.1
9 sampling
dates
Harrisburg,
Pennsylvania
No. of Geom.
Pos . Mean
Detec- of Pos.
tions Det.
9 1.1
5 .3
8 .3
3 .3
6 3.5
6 .2
2 .8
6 .1
8 .5
5 .3
1 1.0
9 .1
9 3.3
9 sampling
dates
NOTE: Most of the sample points were double - i.e., there were two
detectors at each site. The average value was used here.
-------
Table 6.4
I/
Summary Statistics - Pesticides in Air - November 1976 - September 1977—
(ng/m )
Pesticide
Alpha-BHC
Lindane
Heptachlor
Heptachlor
Epoxide
Cis-Chlordane
Trans-Chordane
Dieldrin
Aid r in
DDT/DDD
DDE
Diazinon
Parathion,
Methyl
Malathion
HCB
Dursban
Toxaphene
Wheaton,
No. of Pos
Detections
8
8
0
1
2
1
1
2
7
0
0
5
1
0
Illinois
Geom.
Mean
of Pos.
Det.
.7
1.0
.3
2.0
.8
1.9
2.2
5.2
13.2
.6
10 sampling dates
Pasadena,
No. of Pos
Detections
3
10
6
0
3
11
0
0
10
3
0
0
1
0
California
Geom.
Mean
of Pos.
Det.
.6
2.5
1.3
1.8
3.3
6.0
1.9
.7
12 sampling dates
Greenville,
No. of Pos.
Detections
5
2
4
1
6
1
3
1
7
4
6
2
1
4
Miss.
Geom.
Mean
of Pos.
Det.
.8
17.5
1.4
.7
1.3
.3
6.1
3.4
4.8
9.8
20.5
17.5
.5
15.4
12 sampling dates
Midvale,
No. of Pos.
Detections
8
5
1
0
0
2
6
0
1
0
0
1
0
0
Utah
Geom.
Mean
of Pos..
Det.
1.0
.4
4.9
.4
2.5
1.9
.7
11 sampling dates
Ui
I
I/ The Pasadena data also has an observation for 11/23/77.
-------
Table 6.5
Summary Statistics - Pesticides in Air - 1978
(ng/m )
Pesticide
Alpha-BHC
Lindane
Heptachlor
Cis-Chlordane
Trans-Chlordane
Dieldrin
Aldrin
DDT/DDD/DDE
Diazinon
Parathion
Parathion,
Methyl
Malathion
HCB
HOC
Dursban
Toxaphene
Montana
No. Geom.
of Mean
Pos. of Pos.
Det. Det.
7 1.6
0
5 3.9
0
0
0
0
3 ?
2 1.4
0
2 .7
0
1 143.
0
8 sampling
dates
So . Carolina
No. Geom.
of Mean
Pos. of Pos.
Det. Det.
7 1.0
0
10 3.1
0
1 .4
0
0
11 1.2
7 1.0
0
7 8.2
0
0
0
12 sampling
dates
Greenville,
Mississippi
No. Geom.
of Mean
Pos. of Pos.
Det. Det.
7 1.2
2 7.2
7 11.3
1 1.6
4 .4
0
0
2 ?
5 1.3
7 7.1
7 25.3
0
1 72.7
0
11 sampling
dates
Florida
No. Geom.
of Mean
Pos. of Pos.
Det. Det.
9 TR
1 5.7
1 .3
0
0
0
0
0
7 1.0
4 .5
0
3 5.8
0
4 300.
5 .7
0
13 sampling
dates
Pasadena,
California
No . Geom .
of Mean
Pos. of Pos.
Det. Det.
7 1.3
0
8 11.1
0
5 .9
0
0
1 53.8
5 1.4
-
0
0
0
6 270.
0
11 sampling
dates
Springfield,
Illinois
No. Geom.
of Mean
Pos. of Pos.
Det. Det.
7 1.1
0
4 1.3
0
1 2.6
0
0
1 TR
6 4.0
0
0
0
1 64.8
1 17.8
0
11 sampling
dates
-------
(a) Concentration of Chlordane in Air 4/75-4/76
60-
401
0)
lu 30
T)
80-
60
13
> 40
M
01
CO
20
(c) Concentration of Malathion in Air
Greenville, Miss.-1977 & 1978
1978
(d) Concentration of DDT/DDD in Air - 1977
Wheaton, 111.
Month 2
9 10
Month 12
Year 1976
Pasadena, Ca.
1977
-------
-78-
REFERENCES
[6.1J Kutz, F.W. and H.S.C. Yang. (1975). A note on polychlorinated
biphenyls in air. Proceeding of the National Conference on Poly-
chlorinated Biphenyls, Sponsored by EPA, Office of Toxic Substances,
Washington, B.C., 182.
[6.2] Kutz, F.W., A.R. Yobs, and H.S.C. Yang. National Pesticide
Monitoring Programs. Air Pollution From Pesticides and Agricul-
tural Processes. Robert E. Lee, Jr. (ed), Cleveland, Ohio: CRC
Press, Inc.
[6.3] Report of Monitoring and Analyzing the Ambient Air for Pesti-
cide Residues in Three Locations: Miami, Florida; Jackson, Mississippi;
and Ft. Collins, Colorado during April, May, June, 1975.
[6.4] Stanley, C.W., J.E. Barney II, M.R. Helton, and A.R. Yobs
(1971) Measurement of atmospheric levels of pesticides. Environ.
Sci. Technol. _5_, 430.
[6.5] Report of Monitoring and Analyzing the Ambient Air for Pesti-
cide Residues in Five Locations: Fort Collins, Colorado; Miami,
Florida; Lafayette, Indiana; Jackson, Mississippi; and Harrisburg,
Pennsylvania. August 1975 to March 1976.
t
[6.6] Draft Report - Monitoring and Analyzing the Ambient Air for
Pesticide Residues in Four locations (Pasadena, California; Wheaton,
Illinois; Midvale, Utah; and Greenville, Mississippi) - FY 1977.
[6.7] Untitled data provided by EPA: Analysis of pesticide residues
in six locations (Montana; South Carolina; Florida; Greenville,
Mississippi; Pasadena, California; and Springfield, Illinois) -
January-October, 1978.
-------
-79-
7. NATIONAL SOIL APPLICATION NETWORK
7.1 Detailed Description of Network
7.1.1 Sampling Procedures
Cropping and pesticide use information was collected for
agricultural sampling sites as part of the National Soils Monitoring
Program (NSMP). This information was obtained through personal inter-
views with the landowners or operators during collection of composite
soil and crop samples. A summary of cropping and pesticide use data for
the year 1973 and references to reports of such data for the years 1968,
1969, 1971, and 1972 are given by Gowen and Carey [7.2]. The agricul-
tural sampling sites for the NSMP were selected from the sample segments
of the Conservation Needs Inventory (CNI) of the Soil Conservation
Service. The method of sampling within the CNI segments is outlined by
Wiersma, Sand, and Cox [7.3] and [7.1]. Reference is made to Section
4.1.1 of this report for a discussion of information currently available
to RTI concerning the sampling procedures of the NSMP and the inferences
that-may be drawn using data collected through this sampling plan.
7.1.2 Data Analysis
The set of 33 states that were sampled in each test year and
used in the soil residue analysis of Section 4. was also used to obtain
information on crops being raised and pesticides used.
Over the five year period for which data were available, the six
most frequently occurring crop-code categories included five major
crops — wheat, soybeans, field corn, alfalfa/burr clover, and mixed
hay — and idle cropland (fallow). The number of sites by crop-code and
year is shown below:
Crop-Code
Wheat
Soybean
Corn
Alf ./Burr
Clover
Mixed Hay
Fallow
FY 69
216
234
372
105
49
164
FY 70
90
248
343
105
110
91
FY 72
109
248
423
112
112
85
FY 73
111
252
346
96
102
126
FY 74
78
222
333
77
82
100
For each of the six crop-code categories shown above, an examination
was made of the pesticides used each year. As one would expect, the
total number of pesticides used on a given crop in a single year is
-------
-80-
quite large. For example, some 20 or more pesticides were used on wheat
and soybeans in a given year and about twice that number used on field
corn. Because many pesticides were used on a small number of sample
sites (in many instances only on a single site), the analysis was re-
stricted to those pesticides used on ten or more sites in the case of
field corn and soybeans and five or more sites for the remaining crop-
codes.
Under the above restrictions, only one pesticide was used on five
or more sites in a given year for the last three crop-codes — alfalfa/
burr clover, mixed hay, and fallow. This was 2,4-D used on seven idle
cropland sites in FY69. No further analyses of pesticide usage for
these crop-codes were made.
Cotton was the seventh most frequently occurring crop. However,
the number of sites reporting cotton was small relative to the six crop-
codes shown above and for this reason was excluded.
In the case of wheat, soybeans and field corn, the pesticides of
interest used each year are shown, respectively, in Tables 7.1, 7.2, and
7.3. The cell entries are the number of sites for which the pesticide
was used. In Table 7.1 the frequency of use for all compounds is less
during the latter years of the test program. The one exception is hexa-
chlorobenzene, which shows a slightly higher frequency during FY72-FY74;
however, the frequency is less than 10 percent. With regard to soy-
beans, th.e frequency of use of alachlor, linuron and trifluralin appear
to he increasing with time (.Table 7.2). From Table 7.3, alachlor, buty-
late, carbofuran and cyanazine appear to be increasing over time with
regard to their frequency of use. The remaining pesticides of Table
7.3 show either a stable or decreasing use pattern over time. It should
be noted that all comments relative to usage patterns over time are
based only on a visual examination of Tables 7.1 - 7.3. Statistical
tests were not conducted because of time constraints.
-------
-81-
Table 7.1
List of Pesticides Applied to at Least
Five Sites in One or More Years - Wheat
No. of Sites
Pesticide
Captan
EMTS
Ethylmercury
Chlor.
Hexachlorobenzene
Methylmercury
Acet.
Methylmercury
Dicyandiamide
Parathian, Methyl
2, 4-D
11
56
9
70
4
1
9
0
9
4
28
0
1
7
27
FY 74
78
1
16
14
1
0
5
3
0
0
4
3
7
5
5
4
10
1
2
0
7
0
2
0
27
Table 7.2
List of Pesticides Applied to at Least
Ten Sites in One or More Years - Soybeans
No. of Sites
FY 70
248
FY 74
222
Pesticide
Alachlor
Captan
Chloramben
Linuron
Parathion, Methyl
Trifluralin
0
10
32
7
2
31
9
3
45
8
10
21
25
4
39
16
3
37
41
4
46
31
5
58
38
5
45
24
3
48
-------
-82-
Table 7.3
List of Pesticides Applied to at Least
Ten Sites in One or More Years - Field Corn
FY 69 FY 70 FY 72 FY 73
No. of Sites 372 343 423 346
Pesticide
Alachlor 0 2 38 36 40
Aldrin 59 51 39 29 30
Atrazine 109 137 186 178 152
Butylate 0 5 17 17 29
Bux-Ten 0 16 16 23 13
Captan 137 84 114 76 64
Carbofuran 0 1 17 14 15
Cyanazine 0 0 0 1 10
Diazinon 23 9 11 4 4
Dicamba 2 1 8 10 7
Heptachlor 29 19 6 4 5
Malathion 102 70 95 69 44
Methoxychlor 35 19 20 9 5
Porate 4 16 20 19 16
Propachlor 20* 40 37 34 43
2, 4-D 120 71 72 70 66
-------
-83-
REFERENCES
[7.1] A sampling design to determine pesticide residues levels in
soils of the conterminous United States (1968). USDA Agricultural
Research Service, Plant Pest Control Division. Federal Center
Building, Hyattsville, Maryland 20782.
[7.2] Gowen, J.A. and A.E. Carey (1976). Pesticide application and
cropping data from 37 states in 1973 - National Soils Monitoring
Program. Ecological Monitoring Branch. U.S. Environmental Pro-
tection Agency, Washington, D.C. 20460.
[7.3J Wiersma, G.B., P.F. Sand, and E.L. Cox (1971). A sampling de-
sign to determine pesticide residue levels in soils of the conter-
minous United States. Pestic. Monit. J. 5(1), 63-66.
-------
-84-
8. NATIONAL URBAN SOIL NETWORK
8.1 Detailed Description of Network
a. General
The National Soil Residue Network for Urban Soils for the
years 1972 to 1979 is a two-stage, stratified sample design. The first
stage sampling units were Standard Metropolitan Statistical Areas (SMSAs)
and the second stage sampling units were 231 square meter sites (usually
15.2-by-15.2 meter plots of ground). One dimension of stratification,
the size of the SMSAs, was used at the first stage. Also, one dimension
of stratification, urbanity, was used at the second stage. A sample of
five first stage units was selected. A sample of second stage units was
selected at the following rates:
- Urban stratum—one site per square mile.
- Suburban stratum—one site per twenty square miles.
b. Definition of the Target Population
The target population includes all soil in the Standard Metro-
politan Statistical Areas (SMSAs) in the continental United States.
c. Stratification of the First Stage Frame
The first stage frame is the list of all SMSAs in the conti-
nental United States stratified as follows:
Stratum Description (Based on 1970 Census)
1 All SMSAs with population greater than 1,000,000.
2 All SMSAs with population greater than 100,000
but less than or equal to 1,000,000.
3 All SMSAs with population less than or equal to
100,000.
d. Sample Selection of the First Stage Units CSMSAs)
The sample of five SMSAs was selected from the three strata as
follows: one from stratum 1, three from stratum 2, and one from stratum
3. These SMSAs were selected with equal probability and without replace-
ment with the proviso that an SMSA would be rejected if its land area
was too large; i.e., if the sampling rate of one site per square mile in
the urban stratum and/or one site per twenty miles in the suburban
stratum would yield more samples than the laboratory could analyze in
one year.
-------
-85-
e. Stratification of the Second Stage Frame
The selected first stage units were stratified by urbanity
into two strata an urban stratum and a suburban stratum. The urban
stratum was the city(s) of the SMSA and the suburban stratum was sur-
rounding counties (usually contiguous).
f. Sample Selection of the Second Stage Units (Sites)
Urban—A map was obtained of the city(s) part of the SMSA. On
this map was drawn a one square mile grid. Each block in the grid was
subdivided into an 8x8 grid. A two digit random number from 00 to 88
was selected. Starting with the lower right corner, the first digit was
the number of spaces to the left and the second digit was the number of
space up. This was the selected point on the map. If the point fell in
an inaccessible spot, the random digits were reversed.
Suburban—A map was obtained of the suburban part of the SMSA.
On this map was drawn a one square mile grid. The blocks in this grid
were numbered. An equal probability sample without replacement was
selected from these one square mile blocks at a rate of 1 in 20; i.e.,
if there were 400 square miles in the suburban area 20 one square mile
blocks were chosen. The location of a point in each of the chosen
square miles was the same as for the urban stratum.
Each selected point on the map represented a 15.2-by-15.2 meter
square (50-by-50 foot); i.e., the sampler was permitted to select his
sample anywhere in this 15.2-by-15.2 plot.
g. Generalization of the Results
The target population includes all soil in the Standard Metro-
politan Statistical Areas (SMSAs) in the continental United States;
however, the way the sample was selected the sampled population is a
subset of the target population.
Excluded from the targeted population are all SMSAs whose land area
is too large; i.e., if the sampling rate of one site per square mile in
the urban stratum and/or one site per twenty square miles in the subur-
ban stratum would yield more samples than the laboratory could analyze
in one year. Also, because of the size of the second stage sampling
unit the 15.2-by-15.2 meter plot relative to the sampling grid, the area
frame is incomplete. As a result, conclusions drawn from the sample
apply to the sampled population not the targeted population. If the
-------
-86-
conclusions are extended to the targeted population, serious nonmeasur-
able bias may be introduced.
8.2 Data Analysis
The Urban Soil Data File provided to RTI by Viar & Company gives
pesticide residue concentration for the following:
Year (FY)
1971
1972
1973
1974
1975
1976
1977
Number of
Cities Sampled
10
4
4
5
5
8
10
Total No.
Samples
252
223
381
441
443
377
240
Different cities were sampled during the period 1971-1976. How-
ever, the cities and sampling sites used in 1977 were the same as those
used in 1971. This is consistent with the sample design which calls for
resampling cities every six years to determine changes in residue levels.
It should be noted that fourteen cities were actually sampled in 1971
and again in 1977. Also, five cities were sampled in 1972 and 1973.
Reasons why the present data file does not contain complete test data
will be examined by RTI under another task assignment.
The frequencies of pesticides found annually in urban and suburban
soil samples are shown in Table 8.1. Over the seven year test period,
fifteen pesticides were detected in at least two soil samples in one or
more years. Those pesticides which were detected only once in one or
more years are not listed. In terms of the number of detections the
major pesticides found in the soil include: ^DDT, chlordane, dieldrin
and heptachlor epoxide. Because of their relatively high frequency of
occurrence, additional information is provided on these four pesticides.
Specifically, Table 8.2 gives, by year, city and pesticide, the number
of sampling sites and the number and percentage of sites with detectable
levels of the various compounds. It is apparent from these results that
within a given year the frequency of occurrence of a particular pesti-
cide varies widely by city.
-------
-87-
TABLE 8.1. NUMBER OF POSITIVE DETECTIONS FOUND IN URBAN SOIL BY YEAR
BY PESTICIDE
No. Cities
Sampled
No . Samples
Analyzed
Pesticide
Chlordane
Dieldrin
Heptachlor
Epoxide
Toxaphene
PCB
Heptachlor
Gamma Chlordane
Endrin
Aldrin
Trifluralin
Hexachlorobenzene
Diazinon
Mecoprop
Methoxychlor
Year CFY)
1971
10
252
120
47
7
16
4
0
2
0
0
0
0
0
0
0
0
1972
4
223
168
47
32
20
11
2
5
0
1
4
0
0
0
0
0
1973
4
381
95
36
37
2
0
4
0
0
0
0
0
0
0
0
0
1974
5
441
202
84
53
13
1
11
2
4
0
1
0
0
0
2
2
1975
5
443
187
80
115
54
21
13
6
0
11
7
4
2
0
0
0
1976
8
377
277
155
109
90
7
13
5
28
1
5
0
5
0
0
0
1977
10
240
149
62
55
29
9
17
1
4
3
0
0
1
2
0
1
-------
TABLE 8.2. NUMBER AND PERCENT OF POSITIVE DETECTIONS OF MAJOR PESTICIDES FOUND IN URBAN SOIL DURING 1971-1977,
BY CITY BY YEAR
Chlordane
Year (FY)
1971
1972
1973
1974
1975
City
Mobile
Wilmington
Honolulu
Charleston
Grand Rapids
Greenville
Sikeston
Portland
Philadelphia
Cheyenne
Gadsden
Hartford
Macon
Newport News
Des Moines
Lake Charles
Fitchburg
Pittsburg
Washington
Evansville
Pittsfield
Greenville
Tacoma
Pine Bluff
San Francisco
Springfield
Gary
Durham
State
Ala.
Del.
Hawaii
S.C.
Mich.
Miss.
Mo.
Ore.
Pa.
Wyo.
Ala.
Conn.
Ga.
Va.
Iowa
La.
Mass.
Pa.
D.C.
Ind.
Mass.
S.C.
Wash.
Ark.
Calif.
111.
Ind.
N.C.
Number
Sampling
Sites
29
27
21
27
23
28
27
25
26
19
55
47
43
78
88
69
36
188
132
84
45
85
95
59
162
71
85
66
Positive
Detections
No.
7
2
6
3
10
2
2
3
11
1
4
22
11
10
17
3
2
14
40
17
5
8
14
8
31
26
15
0
%
24.1
7.4
28.6
11.1
43.5
7.1
7.4
12.0
42.3
5.3
7.3
46.8
25.6
12.8
19.3
4.3
5.6
7.4
30.3
20.2
11.1
9.4
14.7
13.6
19.1
36.6
17.6
0.0
Dieldrin
Positive
Detections
No.
3
0
0
0
0
1
1
2
0
0
7
5
13
7
14
15
3
5
19
10
0
8
16
14
23
48
25
5
%
10.3
0.0
0.0
0.0
0.0
3.6
3.7
8.0
0.0
0.0
12.7
10.6
30.2
9.0
15.9
21.7
8.3
2.7
14.4
11.9
0.0
9.4
16.8
23.7
14.2
67.6
29.4
7.6
Heptachlor
Epoxide X/DDT
Positive
Detections
No.
6
1
1
0
5
0
0
0
2
1
1
7
9
3
2
0
0
0
9
0
0
2
2
3
8
30
13
0
%
20.7
3.7
4.8
0.0
21.7
0.0
0.0
0.0
7.7
5.3
1.8
14.9
20.9
3.8
2.3
0.0
0.0
0.0
6.8
0.0
0.0
2.4
2.1
5.1
4.9
42.3
15.3
0.0
Positive
Detections
No.
11
11
4
2
20
25
7
17
20
3
39
40
42
47
44
1
25
25
78
13
25
53
33
49
79
17
35
7
%
37.9
40.7
19.0
7.4
87.0
89.3
25.9
68.0
76.9
15.8
70.9 ,
85.1 g
97.7 '
60.3
50.0
1.4
69.4
13.3
59.0
15.5
55.6
62.4
34.7
83.0
48.8
23.9
41.2
10.6
-------
Chlordane
Year (FY)
1976
1977
City
Bakersfield
Waterbury
Miami
Manhatten
Camden
Houston
Salt Lake City
Milwaukee
Mobile
Wilmington
Honolulu
Charleston
Grand Rapids
Greenville
Sikeston
Portland
Philadelphia
Cheyenne
State
Calif.
Conn.
Fla.
Kan.
N.J.
Tex.
Utah
Wise.
Ala.
Del.
Hawaii
S.'C.
Mich.
Miss.
Mo.
Ore.
Pa.
Wyo.
Number
Sampling
Sites
42
44
50
50
50
47
47
47
24
25
21
25
22
28
27
25
24
19
Positive
Detections
No.
2
32
24
21
15
28
20
13
8
6
14
0
6
0
7
8
9
4
%
4.8
72.7
48.0
42.0
30.0
59.6
42.6
27.7
33.3
24.0
66.7
0.0
27.3
0.0
25.9
32.0
37.5
21.1
Dieldrin
Positive
Detections
No.
10
18
27
14
11
12
13
4
6
3
14
0
1
8
2
7
11
3
%
23.8
40.9
54.0
28.0
22.0
25.5
27.7
8.5
25.0
12.0
66.7
0.0
4.5
28.6
7.4
28.0
45.8
15.8
Heptachlor _
Epoxide Z-DDT
Positive
Detections
No.
4
15
6
14
5
17
12
17
5
1
8
0
4
1
1
3
6
0
%
9.5
34.1
12.0
28.0
10.0
36.2
25.5
36.2
20. a
4.0
38.1
0.0
18.2
3.6
3.7
12.0
25.0
0.0
Positive
Detections
No.
35
37
49
22
50
18
26
40
8
21
7
11
14
26
12
22
21
7
%
83.3
84.1
98.0
44.0
100.0
38.3
55.3
85.1
33.3
84.0
33.3
44.0
63.6
92.9
44.4
88.0
87.5
36.8
I
oo
-------
-90-
Time Trends
The ten cities sampled in FY1971 (see Table 8.2) were resampled six
years later for the purpose of determining changes in residue levels
over time.
Table 8.3 gives the number and percentage of positive detections in
1971 and 1977 for the five most frequently occurring pesticides. The
difference in the two percentages for a given pesticide is an estimate
of the effect of time. The results given in Table 8.3 clearly show that
the percentage of soil samples with a detectable level of a given pesti-
cide was higher in 1977 than in 1971. For four of the five pesticides,
the increase was significant at the .10 (or lower) level of significance.
The remaining pesticide - toxaphene - showed an increase in the percent-
age of positive detections in 1977 over 1971; however, this increase is
not considered statistically significant at the .10 level.
In addition to examining how the percentage of occurrence changes
with time, data for 1971 and 1977 were also analyzed to determine the
impact of time on pesticide residue levels actually found in urban soil.
This analysis was restricted to those pesticides exhibiting a substan-
tial number of non-zero values of residue — i.e., JjDDT, chlordane,
dieldrin and heptachlor epoxide. For each pesticide an analysis of
variance was used to test for differences between yearly means (time
effect) and for differences among the means for the ten cities sampled
in 1971 and 1977. A general discussion of this statistical technique is
given in Section 2. The analyses were actually conducted on log (X +
.01), where X is the residue level observed in a particular soil sample.
It was necessary to add a constant to each observation in order to make
the transformation since many of the X values were zero. The results of
these analyses are shown in Table 8.4. The mean (geometric) level of
dieldrin increased from .001 (ppm) in 1971 to .004 in 1977. Statisti-
cally, this increase was significant; however, from a practical stand-
point, this increase may not be very important. In interpreting this
result one should keep in mind that dieldrin was detected in seven soil
samples in 1971 and fifty-five samples in 1977 (see Table 8.3). The
other three pesticides — ^DDT, chlordane and heptachlor epoxide — did
not show a significant change in the residue level from 1971 to 1977.
Although each of these pesticides showed a significant increase
-------
-91-
TABLE 8.3. NUMBER AND PERCENTAGE OF POSITIVE DETECTIONS FOUND IN URBAN
SOIL SAMPLES FROM TEN CITIES SAMPLED IN 1971 AND AGAIN IN 1977
Year (FY) Test of
1971 1977 Significance
No. Samples Analyzed 252 240
Pesticide
NO. 120 149
% 47.6 62.1 sign. @ .01 level
Chlordane No. 47 62
% 18.7 25.8 sign. @ .1 level
Dieldrin No. 7 55
% 2.8 22.9 sign. @ .01 level
Heptachlor No. 16 29
Epoxide % 6.3 12.1 sign. @ .05 level
Toxaphene No. 4 9 not
% 1.6 3.8 sign. @ .1 level
PCB No*. 0 17
% 0.0 7.1 sign. @ .01 level
-------
-92-
TABLE 8.4. GEOMETRIC MEANS BASED ON SAMPLES SELECTED FROM SAME
LOCATIONS IN 1971 AND 1977
Pesticide
Year (FY)
1971
1977
City
Mobile
Wilmington
Honolulu
Charleston
Grand Rapids
Greenville
Sikeston
Portland
Philadelphia
Cheyenne
Test of Sig.
Between
Yearly Means
State
Ala.
Del.
Hawaii
S.C.
Mich.
Miss.
Mo.
Ore.
Pa.
Wyo.
Test of Sig.
Among City
\t^
2] DDT
.046
.036
Not Sig.
@ .1
.022
.060
.017
.007
.079
.143
.012
.095
.175
.005
Sig.
@ .01
Chlordane
.013
.012
Not Sig.
@ .1
.012
.003
.117
.002
.028
.002
.006
.008
.045
.005
Sig.
@ .01
Dieldrin
.001
.004
Sig.
@ .01
.005
.001
.008
0.0
< .001
.003
.001
.004
.003
.001
Sig.
@ .01
Heptachlor
Epoxide
.001
.002
Not Sig.
@ .1
.003
< .001
.006
0.0
.005
< .001
< .001
.001
.003
.001
Sig.
@ .01
-------
-93-
in the percentage of positive detections from 1971 to 1977 (see Table
8.3), the observed levels were apparently too small to have an effect on
a measure of central tendency such as the geometric mean.
With regard to cities sampled in 1971 and again in 1977, Table 8.4
shows that the residue level of every pesticide examined differs signifi-
cantly from city-to-city.
Effects of Land Use and Location of Sites
Sampling sites within a given city are categorized according to two
factors each with two levels. These are:
Factor Level
Location (L) Urban - located within the political
boundaries of the city.
Suburban - located in the adjacent
counties within the Standard
Metropolitan Statistical Area.
Land Use (U) Lawn I/
Waste
This section discusses analyses conducted for the purpose of deter-
mining: (1) if, and to what extent, these factors affect the residue
level of a given pesticide, (2) if there is an interaction effect of
these two factors, and (3) if effects of these factors are the same from
one city to another (i.e., city by factor (L or U) interaction). The
analyses are restricted to J^DDT and chlordane as these are the only
pesticides with a sufficient number of positive values in cities tested
in a given year to warrant an evaluation.
In the case of JjDDT, an analysis of variance model involving the
factors L, U, and C (.city) and all possible interactions of these fac-
tors was fitted separately to three data sets:
1. Year: 1972. Cities: Gadsden, Ala.; Hartford, Conn.; and
Macon, Georgia.
2. Year: 1974. Cities: Evansville, Ind.; Greenville, S.C.; and
Washington, D.C.
3. Year: 1975. Cities: Gary, Ind.; Pine Bluff, Ark.; and San
Francisco, Calif.
I/ Wiersma, G.B., H. Tai, and P.P. Sand, 1972. Pesticide residues in
~~ soil from eight cities — 1969. Pestic. Monit. J. 6(2):126-129.
-------
-94-
In each analysis the independent variable was log (J^DDT + .01). The
analysis of variance results (shown in Table 8.5) indicate highly sig-
nificant main effects in two of the three years examined. This means
that with respect to the mean level of J^DDT cities differ, urban areas
differ from suburban areas and sites classified as lawn differ from
those classified as waste. Geometric means for the various levels of
the three factors are given in Table 8.6. In 1972 for example, the
geometric mean of ^DDT in Macon, Georgia, was more than six times the
level observed in Gadsden, Alabama. Urban areas show a higher level of
/JDDT than suburban areas and lawn sites exhibit a higher level than
sites classified as waste. Even in the years where differences were not
statistically significant, the estimates showed these patterns.
Table 8.5
Tests of Significance - ^DDT
Effect 1972 1974 1975
C (city)
L (location)
U (land use)
C x L
C x U
L x U
C x L x U
** = Effect significant at the 1% level
NS = Effect not significant at the 5% level
The analysis of interaction effects involving C, L, and U showed
mixed results (see Table 8.5). In all three years the C x U interaction
was not significant indicating that the effect of U (land use) was inde-
pendent of city. The C x L interaction effect was highly significant in
two of the three years thus providing strong evidence that the effect of
location (L) changes from one city to another. This interaction is
brought about by the fact that the mean (geometric) level of 2^DDT in
urban sites is much higher than in suburban sites in some cities and is
about the same in other cities. There was no case where the mean
AA
NS
AA
NS
NS
AA
NS
AA
A*
NS
A*
NS
NS
NS
NS
AA
AA
AA
NS
NS
AA
-------
-95-
TABLE 8.6. SUMMARY STATISTICS FOR
Factor
City
Location
Land Use
NS = Not
Factor
Level
1 (See
2 list
3 below)
Urban
Suburban
Lawn
Waste
significant
1972 1974 1975
G.M. G.M. G.M.
N (ppm) N (ppm) N (ppm)
MC
54 .032** 78 .005** 85 .035
47 .128 85 .025 59 .089
43 .190 130 .082 160 .021
MO
58 .093 114 .085** 101 .069**
86 .086 179 .016 203 .022
MO
59 .138** 144 .053 118 .095**
85 .065 149 .020 186 .014
at the 5% level.
** = Factor significant at the 1% level.
1972:
1974:
1975:
1 = Gadsden, Ala.
2 = Hartford, Conn.
3 = Macon, Ga.
1 = Evansville, Ind.
2 = Greenville, S.C.
3 = Washington, D.C.
1 = Gary, Ind.
2 = Pine Bluff, Ark.
3 = San Francisco, Calif.
-------
-96-
level was significantly lower in urban sites. The L x U and C x L x U
interaction effects were not significant in two of the three years.
In the case of chlordane, an analysis of variance model involving
the factors U and C (city) and their interaction was fitted to two data
sets:
1. Year: 1973. Cities: Des Koines, Iowa and Pittsburg, Pa.
2. Year: 1976. Cities: Camden, N.J.; Houston, Texas; Salt Lake
City, Utah; and Waterbury, Conn.
The independent variable was log (chlordane + .01). The factor L was
not examined because of the small number of positive values. The ana-
lysis of variance results (shown in Table 8.7) indicated land use (U) to
be a very significant factor and the effect of this factor did not
change from one city to another (i.e., no C x U interaction effect).
Sites classified as lawn exhibited much higher levels of chlordane than
waste sites. Geometric means are shown in table 8.8.
Table 8.7
Tests of Significance - 'Chlordane
Effect 1973 1976
C (city) NS *
U (land use) ** **
C x U NS NS
NS = effect not significant at the
5% level.
* = Effect significant at the 5% level.
** = Effect significant at the 1% level.
-------
-97-
TABLE 8.8. SUMMARY STATISTICS FOR CHLORDANE
1973
1976
Factor
Factor Level
City
Land
NS =
* =
** =
1 (see
2 list
3 below)
4
Use Lawn
Waste
not significat at
factor significant
factor significant
1973: 1 =
2 =
1976: 1 =
2 =
3 =
4 =
G.M.
N (ppm) N
MO
57 .024 50
56 .015 47
47
44
59 .049** 109
54 .003 79
the 5% level.
at the 5% level.
at the 1% level.
Des Moines, Iowa
Pittsburg, Pa.
Camden, N.J.
Houston, Tex.
Salt Lake City, Utah
Waterbury, Conn.
G.M.
(ppm)
.014*
.092
.022
.117
.092**
.013
-------
-98-
9. GENERAL ACCURACY OF PESTICIDE CHEMICAL ANALYSIS
To measure valid analytical results, the laboratories participating
in the pesticide analysis program instituted a variety of quality
control measures. One of the most reliable techniques for assessing
overall analytical proficiency involved repeated analysis of a quality
control (QC) sample containing known pesticide levels. Including QC
sample(s) in daily analytical runs provided a means for monitoring
analyst and instrumental performance.
The degree to which an analytical determination can be reproduced
is termed precision and is conveniently expressed as standard deviation
units (SDU) from a mean or true value. The expected fluctuation in a
calculated concentration value is also expressed as relative standard
deviation, (SDU/mean or true value) x 100. It has become customary to
allow a variation corresponding to 2 SDU in the calculated concentration
value for a single sample on repeated analysis. Analysis performed
during a period when the concentration values for a QC sample differ by
more than 2 SDU, are termed unacceptable and are repeated under proper
analytical conditions. Thus, variations covering the range + 2 SDU are
permitted for QC samples and are likely to occur in the sample results.
The uncertainty in analytical accuracy corresponding to this
"acceptable range" in sample results depends on the percent relative
standard deviation (%RSD) appropriate for the particular measurement.
The main factors in estimating %RSD are observed concentration level and
pesticide type. A %RSD range of 7.5 - 10.0 % is reasonable for single
component organochlorine and organophosphorous pesticides found at the
ppm level. Thus, accuracy uncertainty is + 15 - 20% of the reported
-------
-99-
concentration value. At the detection limit (10-30 ppb) for this group
of pesticides, the %RSD equals 50% (by definition). This situation
results in a +_ 100% uncertainty in the accuracy of the reported data.
Intermediate concentration values yield accuracy confidence limits
between these extremes.
Reported
Concentration
Levels Accuracy %RSD
High: ppm +15-20% 7.5-10%
Interm: 100-500ppb + 20-30% 10-15%
Low: 10-30 ppb + 100% 50%
PCB concentration values are realistically qualified by a general
+_ 50% uncertainty in accuracy. The PCB analysis is an extremely complex
case since this material is a mixture of isomers present in varying
proportions. Toxaphene and Chlordane are also multi-component prepa-
rations and consequently are associated with greater uncertainty in
analytical accuracy. At best, the three formulations mentioned above
posses a detection limit of 100 ppb.
-------
-100-
10. RECOMMENDATIONS AND SUMMARY
In this report, RTI has described the preliminary results of
(i) taking EPA computer data files, unprocessed data, and hard copy
documentation from six pesticide networks and then preparing from these
materials machine readable analysis files at RTI and (ii) subsequently
analyzing these data files. In the analyses, RTI has attempted to
answer several specific questions about each data network including
(i) the exact sampling procedures used and the population it represents,
(ii) selected summary statistics and tests of significance which seem
appropriate for summarizing the data, and (iii) an examination of trends
over time and differences in geographic areas for particular pesticides.
The six pesticide networks examined were (i) the Human Adipose Tissue
Network (1970-1977), (ii) the Soil Residue Network (1969-1974),
(iii) the Surface Water and Sediment Residue Network (1976-1979), (iv)
the Air Network (1975-1978), (v) the Soil Application Network (1969-
1974), and (vi) the Urban Soil Network (1971-1977).
In its analyses, RTI has found that (i) with the available documen-
tation only the Human Adipose Tissue Network may be used to make infer-
ence about the entire U.S. (i.e., it is the only network where infor-
mation presently at hand will allow the computation of sampling weights
for observed pesticide levels); (ii) the Human Tissue, Soil Application
Urban Soil, Soil Residue, and Water and Sediment Residue Networks do
allow trends in various pesticide levels to be examined over time and by
geographic area; therefore, these networks certainly have the potential
to monitor trends in pesticide levels in the various media; (iii) a vast
majority of the data for the various pesticides examined have distribu-
tions which have a relatively large number (quite often more than 50%)
of zero or non-detected values; (iv) the data from the various pesticide
networks cannot be easily matched for simultaneous analysis (except in
some ad hoc manner); and (v) unweighted analysis of the Human Tissue
Network indicates several statistically significant differences by race,
age, sex, census division and year for several of the pesticides examined.
(It is important to state as discussed in Section 2 that statistical
significance is not the same as practical significance; therefore,
-------
-101-
statistically significant results given in this report should be exam-
ined with this in mind. Also, as stated in Section 3, RTI is presently
examining weighted analysis of the Human Tissue Network since unweighted
analysis may overstate significance levels).
The results of the preliminary analyses provide a basis for several
recommendations regarding the existing data networks. These include:
(1) An investigation should certainly be undertaken to determine
the feasibility of designing sampling plans so that the data
from the various pesticide networks can be simultaneously
analyzed. This might involve changing the sampling plans for
some of the networks. If it could be accomplished in a
reasonable manner, issues such as the source of pesticides in
human tissue and the lag time from pesticide application to
elevated levels in humans could be addressed. The trade-off
is that the overall coverage of particular networks would be
reduced to accommodate the simultaneous coverage.
(2) All of the pesticide networks should employ computer generated
monitoring techniques (e.g., control charts, trend analyses,
plots, etc.) which automatically flag unusual or unexpected
trends in the various pesticide levels. When such events are
identified, particularly pesticides which have been non-
detected in the past, then each network should examine where
these flagged values are occurring and determine if the levels
are sufficiently high to warrant concern. If, for example,
one geographic area has a majority of the flagged values (and
these values are sufficiently high), then sampling procedures
could be implemented to oversample the suspected area to
better determine the extent of the apparent problem. This
oversampling might be similar to the 1976 special Mirex study
in southern states for the Human Tissue Network.
(3) Periodic EPA Reports (e.g., yearly) should be generated for
all of the networks which not only indicate current pesticide
levels but also time trends, geographic differences, etc.
This has the advantage of combining all of the pertinent
information from the network in one report. In fact, if
-------
-102-
simultaneous analysis of the various networks becomes possible,
it might be reasonable to publish a single report containing
pertinent results from all of the networks, such as compari-
sons of Human Tissue levels in 1978 versus corresponding Soil
levels in 1975. The periodic network report is not meant to
discourage publishing interesting results from the analysis of
network data in several journals but only to have in one
document all the analysis results for the network(s). An
example of a portion of this type of document is the "Pesticide
Residue Levels in Soils and Crops" published annually by EPA
through 1974 for the Soil Residue Network. This document
contained results only for one year but could be expanded to
indicate trends, geographic differences, etc. for several
years.
(4) A document should be prepared for each network which provides
detailed documentation of the network's data files over time.
This would provide in one document all the information neces-
sary to analyze the data from the network. For example, site,
State and census division codes; pesticide codes; how minimum
detectables are to be handled; the exact definition of O's on
the data filejsampling design; etc. This document should be
updated periodically.
(5) The following recommendations apply to the data file systems
for each pesticide network:
A. Have a single residue format for all surveys:
a. use the format that is now part of the Human Tissue
System with some modification. Combine the units
used into one byte similar to what can exist under
SWEMS.
b. use a consistent set of codes for identifying resi-
due. Note the differences that now exist between
SWEMS and the Human Tissue System.
B. All survey data should be handled by a single software
system, including the Air Monitoring System should it
-------
-103-
ever be automated. This system would have three basic
subsystems.
a. data entry - reduction of hard copy data to machine
readable form.
b. editing - develop a table driven edit system that
will handle future modifications of the data.
c. data management system with analysis capability such
as the Statistical Analysis System [2.4].
C. Specific codes should be introduced into the existing
files and future files that indicate whether a specific
residue was not looked for as opposed to not having been
found under analysis.
D. Include Age on the Human Tissue System (at the present
time it was necessary to compute it as the difference
between birth date and date of collection).
E. Sampling weights and sampling design modifiers should be
given for each data record.
(6) The available documentation for the Soil Application, Soil
Residue, and Water and Sediment Networks was inadequate for
determining how to generalize the results to any population
other than the specific sampling sites. Thus, either (i) the
precise sampling procedures for these networks should be
documented so that it may be determined if generalizations to
larger populations may be made or (ii) a newly designed area
probability sample should be developed for each network,
perhaps in conjunction with the Human Tissue Network and the
Urban Soils Network, so that inferences to the entire U.S. (or
specific subsets of the U.S., e.g., states where particular
pesticides are being applied) may be made for each network.
(7) Consideration should be given to the reinstatement of the
National Soil and Crop Pesticide Monitoring Program, perhaps
on a scaled-down level. The basis for this recommendation is
exactly the same objective for which the program was founded
more than a decade ago—that is, to monitor changes in pesti-
cide levels through time. Whilte the program may have provided
-------
-104-
reliable information on pesticide levels during 1968-1973, its
value is very limited in assessing current levels. Certainly,
extrapolating to 1979 and 1980 from data generated in 1968-
1973 should not be attempted. A monitoring system can be
useful without analyzing a large number of samples each year.
Gross changes from baseline levels can be detected with
relatively small sample sizes tested at intervals exceeding
one year. When baseline levels are low, which appears to be
the case for agricultural soil residues, then large changes
are of primary concern. A national estimate may not be the
most appropriate or meaningful statistic to monitor. An
alternative approach could include monitoring purposively
selected areas representing worst case conditions if these
potential hot-spots are known. As mentioned above, in con-
sidering a sample design for the Soil and Crop Residue Network,
the Urban Soil and Human Tissue Network sample designs should
be taken into consideration.
(8) Summary statistics for each network should include (i) percent
of positive values detected, (ii) percent of values greater
than some "meaningful" level, and (iii) the geometric mean of
the positive values (or the geometric mean of all value if
almost all values are positive). Of course, other summary
statistics may also be used in characterizing pesticide levels
(e.g., medians, percentiles, ...) but these three statistics
seem to be reasonable in view of the distributions of the
various pesticide levels. In addition, these summary statistics
2
may easily be tested by a x anci F-test (see Section 2). The
above implies that, if possible, EPA personnel should define
"meaningful" levels for each pesticide that is being monitored.
(9) All laboratories performing pesticide analysis for the various
monitoring networks should follow the quality control procedures
discussed in the EPA Manual for Analytical Quality Control for
Pesticides and Related Compounds in Human and Environmental
Samples, January, 1979, (EPA-600/1-70-008). Every effort
should be made to establish inter-laboratory analytical con-
-------
-105-
sistency, particularly with respect to gas chromatographic
(GC) operating conditions and interpretation of the raw GC
data. All the monitoring networks deserve and demand the
same quality control effort. The need for establishing
data validity exists whether a particular program involves
the analysis of 10 samples or 1,000 samples. An explicit
statement (qualification) should accompany submitted analysis
results to indicate the degree of control under which the raw
data were generated.
(10) An effort should be directed at reducing lag-time between
pesticide network data collection and the results of
statistical analysis of this data.
-------
APPENDIX 1
Data Collection Forms
-------
Table 1.1
ENV
DIVISIC
TISSUE P
DO NOT WRITE IN SHADED AREAS
[„! , „ „,, , ;i
IRONMENTA.. PROTECTION AGENCY
)N OF PESTICIDF COMMUNITY STUDIES
Chamblee, Georgia 30341
HUMAN MONITORING SURVEY
ESTICIDE RESIDUE ANALYSIS REPORT
PATIENT'S HOSPITAL 1 O."
HOS=>! T AL
G.C.MAKE MODEL
OPERATING
PARAMETERS COLUMN i
WET WEIGHT
MILLILITTRS OR
MILLIGRAMS
METHOD OF ANALYSIS
LA6OR ATORY
SPECIMEN ADIPOSE DLOOD SCRUM
a a
DE TEC TOR
COLUMN 2
"• LIPIO EXTRACTADLE MATERIAL (XXX X!
DATE OF ANALYSIS (MONTH. DAY, YEAR)
CHEMIST
Circle code and rocord results of Pesticide Residue Analysis in appropriate boxes.
Place '"X" in proper units column. Place alphabetic confidence level in column CL. Place " "C '' in
column so marked if detectable but unrneas urab le amount of residue is present and enter smallest mcasur
rr HM A
OMB NO
E
3 4 i 8 7 E
. 158-FK
2
1 1
9 10
,,.4, 22 23 24 25 2!J
BLANK
27 20 .2
4 |
3
30 ill 32 33
34 ')*, 96 37 3ft 3(1 40 41
_ . J _
43 44 45 46 47 40
-
1. 1 E.;
n
'•••-'
P R F. 1- '
A N A L •
able amount in AMOUNT colu
J
Code
2.1
37
38
39
40
41
44
'•45
'46
.47
49
'52
'"85
Pesticide Ros iduo
pp' DDT
op DDT
pp' DDE
op DDE
pp' ODD
a -BHC
/J -BHC
Y -BHC
,-, -BHC
Die Idrin
Hept. Epoxide
PCB's
29
Amount
30
{ i i _,., , ,
1 1 , >
•1,1
L_L , ^J • I , I
1 -L , J
1 I , , -1,1
1 , , ,
1 1 , ,
1 , , ,
1 , , I
1,1,
• 1 i 1
•L^J
•1,1
•1 , 1
L . ,_J-L^
1 , , ,
•1 , 1
ppm
T r.
B
2
ppb
17
B A M E A '.i O 1
CL
1 M
Codo
86
87
Pesticide Res idue
Oxy ch lor done
Mirex
,f
Amount PTfTi!
/> n 1*1
i_^, . -1,1
_i_L ^ j
.
1 1
1
1 1
• 1 ! 1
M , 1
i > i
.| . ,
, , i -1,1
, , ,
. , ,
,
i
i
i
i
I
1 1
•1,1
,
•1,1
1-L^J
Remarks: ~ ^ed for FY 71-FY 76.
-------
Table 1.2
MKnifAI. KI'.C'OK'I): Thif; form c niilams m«-ilira! nformnl ion the disc- losiirr ,,r reh-asr of which Hi
rcslni-tfd I'V U.S.C. SSJ, MO lf>); -IS CFK I'iiil 5.
U.S. ENVIRONMENTAL PROTECTION AGENCY
OFFICE OF PESTICIDES PROGRAMS
TECHNICAL SERVICES DIVISION (»n-ssy)
ECOLOGICAL MONITORING QRANCH
WASHINGTON. D.C. I 0 4 r. f
!•' rirrn A f/'ror
ow/f A,',), y.s.s
•ll
•KnUii
NATIONAL HUMAN MONITORING PROGRAM FOR PCSTICIDF3
TISSUE PESTICIDE RESIDUE ANALYSIS REPORT
Krrord results of pesticide residue analysis in appropriate )oxes. Plai.-p "\(> in projHT mills
column. Place alphabetic confidence level in column CL. Place " < " in c-olumn so marked if
cltftcctahle hut unmcasurnblu amount of n's ic. ue is present and enter smallest measurable amount
in AMOUNT c-olumn. ENTER 7F.RO IF NOT DICTECTAHLE,
HOSPITAL OR LOCATION (Name, City unil Scocc;
PATIENT'S IDENTIFICATION NO. PATIENT'S INITIALS
DATE COLLECTED (Month, Dny, 1'rnr) TISSUE TV
1 "1 AQIP
[~] ULOC
[" ' 01 HF
OPERATING
PARAMETERS
WET WEIGHT
PE
OSE
D b (I R U M
H (Snt-CI/l-)
G.C. MAKE MODEL DETECTOR
COLUMN 1 COLUMN 2
% LIPID EXTRACTABLE M A T E Rl A L( 'XXX.X)
METHOD OF ANALYSIS DATE OF ANALYSIS (Month, Dny, Yunr)
LABORATORY
COD
52
»
3fl
39
4"
4 t
42
44
45
46
47
4H
49
CHEMIST
PESTICIDE
RFSIDUfT
pp' -DDT
op' -DDT
pp' -DDC
op' -UDIL
pp • o r> D
(ip DOO
a -HMC
[1 -t.MC
)' -BMC ' 1. ,
f> - n M c
A L D n 1 N
dirt ON IN
REMARKS
- Used
54
AMOUNT
1 , , i .lil
1 , , .lil
I . i i .lil
i , , , .lil
1 i i i >lil
1 i i i *lil
i , , , . l i 1
1 , , , •!,!
1 , , , .1,1
1 i , i . 1 i 1
1 , , , • 1 , 1
I,,, .lil
ppm
r,\
•
ppb
6?
CL
t
t
CODF
SO
SI
" 52
,— "*'
' 8 5
^
87
A P
A E
PESTICIDE
RE5IDUF
ti N C H 1 N
HElf'TACHLOR
HEPT. EPOXIDE
VCil '5
O X, Y CHLOHDAMC
M,HEX
T 1 •- i1. t J 5 • N O N A C H 1- O T-
HEXACHLOHO-
UHN ZI1NE
DO NOT WRITE IN SHADED AREAS
SUMMARY ACTION
8789
"517 in
-
1 0
1 P
31 3-1 Ji
4? 43 44
zs
<29
36
f, r--
c. 4
for FY 77.
1 1
—
B
£7
10
37
f 0
a
i ?.
20
,,.
1
4r;
I]
4 5
l :i i 4
:: l
Q
? ' 32
ip ^n
J 7
—
a
-n
12
J3
jr-
11 49
D
AMOUNT
,1.'
1 i , 1 1 « 1 , 1
1,1,
1 1 | 1
|
|
. L
.L
_J_
1_
J
J
, , , 1 .1 1 1
, I 1 1 .1 , |
1 , , ,
| ,
L
_1_
J
L . t , J.L^J
I,,,
i ' '
l«l , 1
l-l , 1
1 , , , I • 1 , 1
PI
r,
;n
I , , , L, , 1
SEX
AGE
ORIGIN
Prl,
CL
7 1
a
72 73
a
EPA Hq Form 8510-9 (Rev. 4-76)
RE FJ LACES EPA FORM S510-9 lf HI C H IS OOSOLtTE.
-------
Table 1.3
SECTION 1. SAMPLE IDENTIFICATION DATA
HAS"
NO.
SAMPLE ACCES-
SION NO.
l|2 ' 3 4 5 6
"ctT
NO.
10
DATE RECEIVED AT LAB
PESTICIDE ANALYSIS WORKSHEET
SECTION 2. SAMPLING DATA (To be completed at sampling tune)
SAMPLED BY (Agency and last name):
DATE SAMPLED
nll2J13|l4|l5|l6
I ! I i
SITE
STATION/SITE NUMBER
STATE
17
18
COUNTY OR REGION
19
20
21
22
23
24
25
26
27
23
29
30
31
SYSTEM
32
33 NS " NATIONAL SOILS MONITORING
ME if NATIONAL ESTUARINE MONITORING
NW £ NATIONAL WATER MONITORING
MATERIAL
CROP NUMBER (If applicable)
34
35
36
PESTICIDES USED (Check or specify)
2, 4—D
ALDRIN
CHLORDANE
DDT
DIELDRIN
ENDRIN
MALATHION
PARATHION
TRIFLURALIN
ATRAZINE
OIAZINON
HEPTACHLOR
TOXAPHENE
SAMPLING REMARKS
SECTION 3. SPECIFIC SAMPLE CHARACTERISTICS
ESTUARINE
SPECIES
(Code)
38 39
40
41
42
43
pH
SOIL
33
39
401;
% SAN D
41
%SILT
DATE ANALYSIS COMPLETED:
53
4~2J43l ' T
L:, 1
"3 441
b4J55"[5s|57 sal
45
461
% CLAY
% ORGANIC MATTER
47 48
49
50
51
52'
SECTION 4. RESIDUES DETECTED
10
PESTICIDE
CODE
11
21
31
41
51
12
22
32
42
52
13
23
AMOUNT
14
24
15 16
17
25 26(27
18
19
20
28 29 30
43
53
-t-
33134 35 36 37r381i9l40
44
J—i-t-4—
45|4G|47|48
54
49
55
56157
58
59
50
60
10
PESTICIDE
J
CODE
21 22
31'32
I
41 42
51
52
13
23
33
AMOUNT
14
24
34
43 44
53
54,
15
25
35
45
55
16
26
17
27
36.3 7
46 47
56
57
18
28
19
29
38 39
48 49
58
59
20
30
40
50
60
62
63
64
65
66
67
68
69 70
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
71
72
73
74
75
76
77
78
79
80
* M - P.P.M. (default); B - P.P.B. wtf/fctf whole boily. wet weight; T = P.P.T.
REMARKS
DATE
ANALYST'S
MTIALS
SPA Form 8550-2 'Rev. 2-75)
PREVIOUS EDITION IS OBSOLET
-------
—
D
"*«-.
1
-1 B
1
Sac
2-8
Ian!
Y
7fi
?
iple
Whole 1
i
Bottom
lf),-1>lff~~lttd 17 - 80 Blank
M D
• /
Li3t
Time
Lab No. _
Table 1-^
Date Rece.io
/ed at Lab ' { • • 7 .
/) ^
bequcnce '
1U No
Cft?.{\ R - 11 Klnn., q^A^I FM PI 1 1
7
Type
Vater
Material
List
ID
A
B
1^ 17 18 Iff " Zu
Chlor-
Aldrin dane DDD*
u u IA
DDE'- DDT* Dieldrin Endrin
U (A U IA
llepta- Heptachlor Lin- '^tlioxy
chlor epoxide dane chlor
U LA- U^ U_
roxa
plliif. :
^i
Sample
. i
Type
( ! '.s'hole Water
; Hoc
i ,
ton
Sample
'..•hole
Bottom
Material
Type
Wa t e r
Material
List
ID
C
D
List
ID
L
F
2,4-D 2,4,5-T Silvex
Diazinon Ethion
U IA
Atrazine Simazine Polychlorinatcd Sample Type
biphonvl's fPCBl
•Hiole Water
Sottom Material
Methyl
Malathion Parathion
Ul A
Methyl
Trithion Paratliion
(~A^__ [_^(
Sanple
'. ,'h o 1 e
Type
Water
Bottom Material
List
ID
G
U
.is
1U
t
G
H
PCU
U^
Tri thion
C-^^-_
-
1242 1248 1254 1260 Other Sample Type Lisc
ID
Whole Water ^
Bottom Material
^ »
Sample
'.-.'hole
Type
Water
List
ID
K.
P.P'DDD
(IDE) O.P'DDE
P.P'DUC O.P'DDT P.P'DDT
V
Othel Sample Type
Ifliole Water
-------
Table 1.5
Positions
Start
1
2
3
5
11
13
16
26
32
35
39
43
43
44
47
50
53
56
59
43
49
43
49
55
61
67
68
69
71
97
98
98
101
107
* Units
End
1
2
4
10
12
15
25
31
34
38
42
60
43
46
49
.52
55
58
60
48
60
48
54
60
66
67
68
70
96
97
99
100
106
107
: M - PPM
B - PPB
T - PPT
SWEMS Format for Soil and Water Data
Field Name Comment
Reserved
Sample Category
Lab Number
Accession Number
State Code
County Code
Site Code
Date Sampled
Sample Material Code
Rain Fall
Temperature
Category Data
Crop Number
Ph
% sand
% silt
% clay
% organic
region code
species code
blank
Flow rate
suspended sediment
blank
Analysis date
Cropping Year
Land Use Indicator
Fiscal Year
blank
Residues
Group
Code
Amount
Units*
Usually zero
1 = Estuarine
2 = Water
3 = Soil
ZIPS
(month, day, year)
Soil Data
Estuarine data
filler
Water Data
filler
(month, day, year)
filler
Repeat for each residue
up to 40 sets
-------
Positions
Start
1
4
13
14
20
23
25
26
28
30
32
34
36
45
51
53
54
55
58
61
64
94
98
103
107
109
111
113
114
116
118
124
125
127
128
129
135
137
138
End
3
12
13
19
22
24
25
27
29
31
33
35
44
50
52
53
54
57
60
63
' 93
97
102
106
108
110
112
113
115
117
123
124
126
127
128
134
136
137
138
Table 1.6
Human Tissue Format
Field Name
Comments
Hospital
Patient
ID - suffix
Date Collected
County
Tissue Code
Design
Region
EPA State Code
EPA Census Code
FIPS State Abbrev.
FTPS State Code
Patient - Access No.
Patient Date of Birth
Initials
Sex
Race
Occupation Code
Height
Weight
Diagnosis
Indicators
Wet Weight
% LIPID
Lab Code
Mtnd. Code
Inst. Code
Detet. Code
Col 1 Code
Col 2 Code
Analysis Date
Pref - Analysis
Residue
Code
Suffix
Trace
Amount
Units
Confidence
Override
year, month, day
month, day, year
6-5 byte fields
P-S, HOSP, EMBLM, OUTSIDE
month, day, year
Repeats for 20 residues
PPM, PPB
-------
Table 1.7
Pesticides Included in Analysis Files
Soil File
Aldrin
DDTOP
DDTPP
DDEOP
DDEPP
TDEOP
TDEPP
Dieldrin
Heptachlor
Heptachlor Epoxide
Lindane
PCB
Parathion
Human Tissue
Total DDT Equivalent
ct-BHC
B-BHC
Lindane
6-BHC
Aldrin
Dieldrin
Endrin
Heptachlor
Heptachlor Epoxide
Polychlorinated Biphenyls (PCB)
Oxychlordane
Mirex
Trans Nonachlor
Hexachlorobenzene
Water File
Aldrin
Atrazine
Chlordane
2, 4-D
DDEOP
DDEPP
DDTOP
DDTPP
Dieldrin
Heptachlor
Heptachlor Epoxide
Lindane
Parathion, Methyl
Parathion
PGP
PCB's (4)
Sileex
/ 2, 4, 5-T
TDEPP
TDEOP
Toxaphene
-------
APPENDIX 2
(See Addendum)
-------
APPENDIX 3.1
CENSUS BREAKDOWN OF THE UNITED STATES
-------
CENSUS BREAKDOWNS OF THE UNITED STATES
Region
North East
Division
New England
North Central
Middle Atlantic
East North
Central
West North
Central
South
South Atlantic
East South
Central
West South
Central
States
Connecticut
Maine
Massachusetts
New Hampshire
Rhode Island
Vermont
New Jersey
New York
Pennsylvania
Illinois
Indiana
Michigan
Ohio
Wisconsin
Iowa
Kansas
Minnesota
Missouri
Nebraska
North Dakota
South Dakota
Delaware
District of Columbia
Florida
Georgia
Maryland
North Carolina
South Carolina
Virginia
West Virginia
Alabama
Kentucky
Mississippi
Tennessee
Arkansas
Louisiana
Oklahoma
Texas
-------
CENSUS BREAKDOWNS OF THE UNITED STATES (continued)
Region Division States
West Mountain Arizona
Colorado
Idaho
Montana
Nevada
New Mexico
Utah
Wyoming
Pacific Alaska
California
Hawaii
Oregon
Washington
-------
APPENDIX 3.2
MATERIALS FOR SELECTING SAMPLE SITES FOR
MIDDLE ATLANTIC CENSUS DIVISION FOR FY 1973
-------
PCN7tSTLVA5tA
cuiAMca *MT .^T
FUr Uwn hnmxli '.i^». j^rfr.
::::::::55
cl*ir :
jXu«l*r CM
!?&»•!• ~IITM||JJ| -f^"
?>••*!a »•*»
Pmui
?*rtn A«i»r
?l«ia/l«ld
"crtl 5*1
i»«.t Once* <^ .£?~| 43,2U I
i3,li-*
IC0.9M
dt,M»
74.44C
U1, -t 14
47,09*
1O,]94
W,334
43,iJl
43,339
59,938
233,457
37,252
41,Qi9
I5,JL3
It,108
4.3, iS*
374,3TS I
41,909 '
. .Ji.Ui
33,339
J8,103
53,731
142,919
la,3«4
46,3-*4
29,034
2T(3aT
32,370
102,211
58,632
46,7^1
33,£06
40,0<1
suer>io <
Zl*lr» cJ
?T««port
Cu-daa CJ
Gl.n Covi
Ju»»ta»n "-JT
Kims ton CI:T . .
L^ClUBVUtl —L»j-_
Llaainfturjt ~r++
LoGHpar* «*^u.
.<<«* Hoeavil* •***
..-.«*.yr
\'S&
:^t.
.^r
^AV:.
..#*-
. f^
'•:&
:s<£
113,837
U,;22
4i7,aoa
39.J73
ftocB*it»r i**^ U.51
AacKTtLl* C«tttr» u-UL*«jy«rf^M
.=!
Srrmeui* -tar
Tror ciey*.'...
• »»it« pi.tn. c»*r .-•jr-
, Yonk*r> atar ftf:
13,730
U,44a
23,14 i
39, .12 i
23,19 I
23,39 I
:«,26 i
23,2^0 |
32,307
7,791,737 i
a4,7S2 !
33.-I13 '
293.J93
J7.274 '
47,924 |
77,134 |
192.32*
43,007
90, »2
40,332
3n,i2i j
49,373
204,78» 1
- ~?7s/
-------
<£ SAjufiMbt 71, U'
''di, ti.y,
; ¥3?
-*/
ii\
.//
. 7I.Q.
77. t.
Jtf 3 So
21,310
7], I
LL
131
/^fy.oll
I, /00j f& /
7,^/4//A
f^u, //y
-------
y/7
20
2b
So-
1*
6
K\
% \ Uiu-nLru-TV; <3a,/
1'k_QalK['f*«,t.&'..
Ok.
it /
I
1,311.*.
4,M,'-
y, fa-, a J/
5,4*3,177
J, 111,
-------
77
S7
•, n. y. i i
,*.'/- n
,1.y.
7.o\ . s/wni, 7J , Ur,
/7.
>/
3?
•M>-
tf.u
/4*,f'9
ft, ,(.1,1,
tot, ///
3*
bo
/l-utT' Zr
•'-!<••'- , '
' &
^,
3
-------
/7
01
J1
of
#.#,
71. 1.
^
uy
/
/
-------
APPENDIX 3.3
ORIGINALLY SELECTED CITIES FY70, FY73, FY77
AND SELECTED CITIES INCLUDING ALTERNATES FY73-FY78
-------
ORIGINALLY SELECTED CITIES
FY70
NORTHEAST
NORTH CENTRAL
Bridgeport, Conn.
Fitchburg, Mass.
Newburg, N.Y.
New York City, N.Y.(3)'
Ridgewood, N.J.
Philadelphia, Pa.
Bangor, Maine
Pittsburgh, Pa.
New Rochelle, N.Y.
Witchita, Kansas
Detroit, Mich.
St. Joseph, Mo.
Des Moines, Iowa
Oak Lawn, 111.
Chicago, 111. (2)2
Salina, Kans.
St. Louis, Mo.
Belleville, 111.
Minneapolis, Minn.
Lorain, Ohio
South
West
Houston, Texas
New Orleans, La.
Miami Beach, Fla.
Atlanta, Ga.
Oklahoma City,. Okla.
Rock Hill, S. C.
Owensboro, Ky.
Macon, Ga.
Washington, D.C.
San Francisco, Calif.
Tucson, Ariz.
Portland, Oreg.
Gardena, Calif.
San Bernardino, Calif.
Los Angeles, Calif.
San Diego, Calif.
1 Cities replaced by alternates during the first year of operation of
the statistical design.
2 Number of pathologists selected in city; where no number is given
assume number to be one.
-------
NEW ENGLAND (4)
SOUTH ATLANTIC (11)
* Boston, Mass. •
Worcester, Mass.
Pittsfield,- Mass.
Westfield, Mass.
MIDDLE ATLANTIC (14)
* Erie, Pa.
Hoboken, N. J.
* Philadelphia, Pa.
Hazleton, Pa.
* Ridgewood, NY J.
Troy, N. Y.
* New York City - 7
Bayonne, N. J.
NC
EAST NORTH CENTRAL (15)
Allen Park, Mich.
* Toledo, Ohio
Wyandotte, Mich.
* Indianapolis, Ind.
* Columbus, Ohio
Mansfield, Ohio
* Evansville, Ind.
* Detroit, Mich.
Parma, Ohio
* Cleveland, Ohio
Inkster, Mich.
*.Chicago, 111. - 2
Beloit, Wise.
Bay City, Mich.
WEST NORTH CENTRAL (6)
Iowa City
* St. Louis, Mo.
St. Louis Park, Minn
Raytown, Mo.
'* Wichita, Kansas
Omaha, Nebr.
C.
Norfolk, Va.
Greenville, S. C.
Panama City, Fla.
Baltimore, Md.
College Park, Md.
Hagerstown, Md. •
Winston-Salem,' N.
Huntington, W. Va,
* Washington, D. C.
Wilson, N. C.
Charlotte, N. C.
EAST SOUTH CENTRAL (5)
* Louisville, Ky.
* Tuscaloosa, Ala.
Mobile, Ala.
* Memphis, Tenn.
Kingsport, Tenn.
WEST SOUTH CENTRAL (7)
San Antonio, Tex.
* Dallas, Tex.
* Enid, Okla.
El Paso, Tex.
* Houston, Tex.
* Oklahoma City
* New Orleans, La.
W
MOUNTAIN (3)
* Salt Lake City, Utah
Denver, Colo.
Phoenix, Ariz.
PACIFIC (10)
San Bruno, Calif.
* Los Angeles City, Calif.
Long Beach, Calif.
Bakersfield, Calif.
Lynwood, Calif.
Chula Vista, Calif.
Tacoma, Wash.
* San Francisco, Calif.
Glendale, Calif.
- 2
-------
NATIONAL HUMAN MONITORING PROGRAM
COLLECTION SITES BY STATE
FY 1974
Alabama (6)
Mobile
Tuscaloosa
Arizona (8)
Phoenix
California (9)
Bakersfield
Glendale
Lakewood
Long Beach
Los Angeles - 2
National City
San Francisco
Colorado (8)
Denver
District of Columbia (5)
Florida (5)
Hialeah
Panama City
Illinois (3)
Chicago - 3
Oak Park
Indiana (3)
Evansville
Indianapolis
Mishawaka
Iowa (4)
Iowa City
Kansas (4)
Salina
Wichita
Kentucky (6)
Louisville
Louisiana (7)
New Orleans
Maryland (5)
Baltimore
Massachusetts (1)
Boston
Pittsfield
Westfield
Worcester
Michigan (3)
Detroit
Wyandotte
Minnesota (4)
St. Louis Park
Missouri (4)
St. Louis
Nebraska (4)
Omaha
New Jersey (2)
Hoboken
New York (2)
Buffalo
Jamestown
New York City
Troy
- 7
North Carolina (5)
Charlotte
Winston-Salem
Ohio (3)
Cleveland
Columbus
Mansfield
Parma
Toledo
Oklahoma (7)
Enid
Oklahoma City
Oregon (9)
Eugene
Pennsylvania (2)
Erie
Hazelton
Philadelphia
South Carolina (5)
Anderson
Greenville
Tennessee (6)
Kingsport
Memphis
-------
NATIONAL HUMAN MONITORING PROGRAM
COLLECTION SITES BY STATE
BY 1974 (continued)
Texas (7)
Dallas
El Paso
Houston
San Antonio
Utah (8)
Salt Lake City
Virginia (5)
Norfolk
Petersburg
Washington (9)
Tacoma
West Virginia
Huntington
Wisconsin (3)
Beloit
-------
NATIONAL HUMAN MONITORING PROGRAM
COLLECTION SITES BY STATE
BY 1975 (continued)
Texas (7)
Dallas
El Paso
Houston
San Antonio
Utah (8)
Salt Lake City
Virginia (5)
Norfolk
Petersburg
Washington (9)
Tacoma
Wisconsin (3)
Beloit
-------
NATIONAL HUMAN MONITORING PROGRAM
COLLECTION SITES BY STATE
FY 1975
Alabama (6)
Mobile
Tuscaloosa
Arizona (8)
Phoenix
California (9)
Bakersfield
Glendale
Lakewood
Long Beach
Los Angeles - 2
National City
San Francisco
•
Colorado (8)
Denver
District of Columbis (5)
Florida (5)
Hialeah
Panama City
Tampa
Illinois (3)
Chicago - 3
Oak Park
Indiana (3)
Evansville
Indianapolis
Iowa (4)
Iowa City
Kansas (4)
Salina
Wichita
Kentucky (6)
Louisville
Louisiana (7)
New Orleans
Maryland (5)
Baltimore
Massachusetts (1)
Boston
Pittsfield
Westfield
Worcester
Michigan (3)
Bay City
Detroit
Wyandotte
Minnesota (4)
St. Louis Park
Missouri (4)
St. Louis
Nebraska (4)
Omaha
New Jersey (2)
Hoboken
New York (2)
Buffalo
Jamestown
New York City
Troy
- 7
North Carolina (5)
Charlotte
Winston-Salem
Ohio (3)
Cleveland
Columbus
Mansfield
Parma
Toledo
Oklahoma (7)
Enid
Oklahoma City
Oregon (9)
Eugene
Pennsylvania (2)
Erie
Hazelton
Philadelphia
South Carolina (5)
Anderson
Greenville
Tennessee (6)
Kingsport
Memphis
-------
Census Division
New England
NATIONAL HUMAN MONITORING
PROGRAM FOR PESTICIDES
Sampling Sites FY'76
No. Sites
Middle Atlantic
18
East North CentraV
17
West North Central
South Atlantic
13
Sites
Boston, MA
Pittsfield, MA
Westfield, MA
Worcester, MA
Buffalo, NY
Jamestown, NY
New York, NY (11)
Troy, NY
Erie, PA (2)
Hazel ton, PA
Philadelphia, PA
Chicago, IL (4)
Oak Park, IL (2)
Evansville, IN
Indianapolis, IN
Bay City, MI
Detroit, MI
Wyandotte, MI
Cleveland, OH
Columbus, OH
Mansfield, OH
Parma, OH
Toledo, OH
Beloit, WS
Iowa City, IA
Salina, KS
Witchita, KS
Omaha, NB
St. Louis Park, MN
St. Louis, MO
Washington, D.C.
Hialeah, FL
Panama City, FL
Tampa, FL
Baltimore, MO (2)
Charlotte, N.C.
Winston-Sal em, N.C
Anderson, S.C.
Greenville, S.C.
Norfolk, VA (2)
Petersburg, VA
-------
Census Division No. Sites Sites
East South Atlantic 5 Mobile, AL
Tuscaloosa, AL
Louisville, KY
Kingsport, TN
Memphis, TN
West South Central 8 New Orleans, LA (2)
Enid, OK
Oklahoma City, OK
Dallas, TX
El Paso, TX
Houston, TX
San Antonio, TX
Mountain 3 Phoenix, AZ
Denver, CO
Salt Lake City, UT
Pacific 9 Bakersfiled, CA
Glendale, CA
Lakewood, CA
Long Beach, CA
Los Angeles, CA
National City, CA
San Francisco, CA
Eugene, OR
Tacoma, WA
-------
NATIONAL HUMAN MONITORING PROGRAM FOR PESTICIDES
SAMPLING SITES FY77
Census Division
New England
Middle Atlantic
East North Central
West North Central
South Atlantic
East South Central
West South Central
Mountain
Pacific
No. Sites
8
Sites
Hartford, CT
Springfield-Chicopee-Holyoke, MA-CT
Lancaster, PA
New York, N.Y. (2)
Northeast, PA (Wilkes-Barre-Scranton)
Philadelphia, PA-NJ
Pittsburgh, PA
Reading, PA
Chicago, IL
Cleveland, OH (2)
Detroit, MI (2)
Madison, WI
Dayton, OH
Akron, OH
Minneapolis-St. Paul
Omaha, NE-IA
St. Louis, MO-IL
MN-WI
Charlotte-Gastonia, NC
District of Columbia, DC-MD-VA
Fort Lauderdale-Hollywood, FL
Greenville-Spartanburg, SC
Macon, GA
Tampa-St. Petersburg, FL (No Collection)
Birmingham, AL
Nashville-Davidson, TN
Lexington, KY
Tuscaloosa,AL (Extra Collection Point)
Dallas-Fort Worth, TX
El Paso, TX
Lubbock, TX
Shreveport, LA
Denver-Boulder, CO
Salt Lake City-Ogden, UT
Los Angeles-Long Beach, CA (2)
Portland, OR-WA
San Diego, CA
Seattle-Everett, WA
-------
NATIONAL HUMAN MONITORING PROGRAM FOR PESTICIDES
Census Division
New England
Middle Atlantic
East North Central
West North Central
South Atlantic
East South Central
West South Central
Mountain
Pacific
13
SAMPLING SITES FY1978
No. Sites
Sites.
Hartford, CT
Sprinqfield-Chiconee-Holyoke, MA-CT
Lancaster, PA
New York, NY (2)
Northeast, PA (Wilkes-Barre-Scranton)
Philadelphia, PA-NJ
Pittsburgh, PA
Reading, PA
Chicago, IL (2)
Cleveland, OH
. Detroit, MI (2)
Madison, WI
Dayton, OH
Akron, OH
Minnea'polis-St. Paul
Omaha, NE-IA
St.. Louis, MO-IL
MN-WI
Charlotte-Gastonia, NC
District :of Columbia, DC-MD-VA
Fort Lauderdale-Hollywood, FL
Greenville-Spartanburg, SC
Charleston, W.VA
Orlando, FL
Birmingham, AL
Nashville-Davidson, TN
Lexington, KY
Dallas-Fort Worth, TX
El Paso, TX
Lubbock, TX
San Antonio, TX
Denver-Boulder, CO
Salt Lake City, UT
Los Angeles-Long Beach, CA
Portland, OR-WA
Sacreniento, CA
Seattle-Everett, WA
(2)
10/31/77
-------
APPENDIX 3.4
SURVEY AND SITE QUOTAS FOR FY73-FY76
-------
National Human Monitoring Program
Collected by Census Division
Sex
Age Groups
M
Total
New England (1) -
4 Collection Sites - 1%
•
Middle Atlantic (2) -
14 Collection Sites - 19%
East North Central (3) -
14 Collection Sites - 19%
West North Central (4) -
5 Collection Sites - 7%
0-14
15-44
45+
Total:
0-14
15-44
45+
Total:
0-14
15-44
45+
Total:
0-14
15-44
45+
Total:
16
20
16
52
# Negroes
56
70
56
182
# Negroes
56
70
56
182
# Negroes
20
25
20
65
# Negroes
16
20
20
56
= 4
42
84
70
196
= 42
56
84
56
196
= 42
20
25
25
70
= 5
32
40
36
108
98
154
126
378
112
154
112
378
40
50
45
135
-------
National Human Monitoring Program
Collected by Census Division
Sex
South Atlantic (5) -
11 Collection Sites - 15%
East South Central (6) -
7 Collection Sites - 9%
West South Central (7) -
7 Collection Sites - 9%
Age Groups
0-14
15-44
45+
Total:
0-14
15-44
45+
Total:
0-14
15-44
45+
Total:
M
44
55
44
143
# Negroes
28
35
28
91
# Negroes
28
42
21
91
# Negroes
•
F
44
66
44
154
= 66
28
42.
28
98
= 35
28
42
28
98
= 28
Total
88
121
88
297
56
77
56
189
56
84
49
189
Mountain (8) -
3 Collection Sites - 4%
0-14
15-44
45+
Total:
12
18
12
42
# Negroes
12
15
if-
39
= 3
24
33
24
81
Pacific (9) -
10 Collection Sites - 13%
0-14
15-44
45+
Total:
40
60
40
140
# Negroes
30
60
40
130
= 20
70
120
80
270
-------
National Human Monitoring Program
Collected by Census Division
Sex
Age Groups
M
Total
Percent
Summary:
Total Males = 49%
Total Females = 51%
Total Negroes = 12%
0-14
15-44
45+
Total:
300
395
293
988
276
438
323
576
833
616
1037 2025
# Negroes = 245
Percent = 12%
28.4
41.1
30.4
99.9
-------
Census Divisions
National Human Monitoring Program
Age, Race, Sex Distributions
New England (1)
*
Middle Atlantic (2)
East North Central (3)
West North Central (4)
South Atlantic (5)
Age Groups
0-14 •
15-44
45+
Total:
0-14
15-44
45+
Total :
0-14
15-44
45+
Total:
0-14
15-44
45+
Total:
0-14
15-44
45+
Total:
Sex
M
4
5
_4
13
# Negroes =
4
5
_4
13
# Negroes =
4
5
_4
13
# Negroes =
4
5
4
13
# Negroes =
4
5
_4
13
# Negroes =
•
F
4
5
_5
14
1
3
6
_5
14
3
4
6
_4
14
3
4
5
_5_
14 .
1
4
6
4
14
6
Total
8
10
_9
27
7
11
_9.
27
'
8
11
J5
27
8
10
_9
27
8
11
_8
27
-------
Census Divisions
National Human Monitoring Program
Age, Race, Sex Distributions
Age Groups
East South Central (6)
0-14
15-44
45+
Total:
West South Central (7)
0-14
15-44
45+
Total:
Mountain (8)
0-14
15-44
45+
Total:
Pacific (9)
0-14
15-44
45+
Total:
Sex
M
4
5
_4
13
# Negroes = 5
4
6
J3
13
# Negroes = 4
4
6
_4
14
# Negroes = 1
4
6
_4
14
# Negroes = 2
•
,F
4
6
_4
14
4
6
_4
14
4'
5
_4
13
3
6
_4
13 •
Total
8
11
_8
27
8
12
_8
27
8
11
_8
27
7
12
_8
27
-------
APPENDIX 3.5
Survey and Site Quotas for FY77 +
-------
NATIONAL HUMAN MONITORING PROGRAM FOR PESTICIDES
SAMPLING DESIGN
CENSUS DIVISION DEMOGRAPHIC QUOTA
New England - 2 Collection Sites - 5.83%
Sex
Age Groups
0-14
15-44
45+
TOTAL
#Non-Caucasians
Middle Atlantic - 7 Col
Age Groups
0-14
15-44
45+
TOTAL
#Non-Caucasians
i_dSt North Central - 8
Age Groups
0-14
15-44
45+
.M
12
16
11
40
= 2
lection
M
42
56
' 42
140
= 28
F
10
16
li
40
Sites - 18.28%
Sex
F
35
56
49
140
Total
22
32
26
80
Total
77
112
91
280
Collection Sites - 19.81%
M
48
64
48
Sex
F
48
64
48
Total
96
128
96
TOTAL 160
rNon-Caucasians = 32
160
320
-------
West North Central - 3 Collection Sites - 8.03% .
Sex
Age Groups M F Total
0-14 18 15 33
15-44 24 24 48
45+ . 18 11 39
TOTAL 60 60 120
irNon-Caucasians = 6
South Atlantic - 6 Collection Sites - 15.1.0%
Sex
Aqe Groups
0-14
15-44
45+
M
36
48
30
F
36
54
36
Total
72
102
66
TOTAL 114 126 240
#Non-Caucasians = 48
East South Central - 3 Collection Sites - 6.30%
Sex
Age Groups M_ £ Total
0-14 18 18 36
15-44 24 24 48
45+ 18 18 36_
TOTAL 60 60 120
#Non-Cau.casians = 24
-------
West South Central - 4 Collection Sites -.9.50% '
Sex
Age Groups M £ Total
0-14 24 24 48
15-44 32 36 68
45+ 20_ 24- 44_
TOTAL 76 84 160
#Non-Caucasians = 24
Mountain - 2 Collection Sites - 4.08%
Age Groups
0-14
15-44
45+
*
M
12
16
10
Sex
F
12
18
12
Total
24
34
22
TOTAL 38 42 80
,#Non-Caucasians = 2
Pacific - 5 Collection Sites - 12.53%
Sex
Age Groups M. £ Total
0-14 30 25 55
15-44 45 45 90
45+ 25 30 55_
TOTAL 100 100 200
#Non-Caucasians = 10
-------
CEKSVS DIVISION'S
NATIONAL HUMAN MONITORING PROGRAM
AGE, PACE, SEX QUOTAS
NEW ENGLAND (1)
AGE GRC-'JTS
0-14
15-44
45 +
TOT;
MIIOLE ATLANTIC (2)
0-14
15-44
45 +
EAST NJFTH CE:
0-14
15-44
TOTAL
WEST NORTH CENTRAL (4;
0-14
15-44
45+
TOTAL
SEX
M
6
8
6
20
NEGROES
6
8
6
20
NEC-ROES
6
8
6
20
NEGROES
6
8
6
F
5
8
7
20
= 1
5
8
7
20
= 4
6
8
20
= 4
5
8
7
TOTAL
11
16
13
40
C (VVA • Cc, *ft 1 I'* - 5
11
16
13
40
C fl/CM C&*et.S t'en*s
12
1C
12
40
(/Von - Ct, vets *
11
16
13
20 20
# NEGROES = 2
40
)
-------
-2-
SOUTH ATLANTIC (5)
AGE GROUPS
0-14
15-44
45 +
TOTAL
SEX
M F_
6 6
8 9
5 6
19 21
# NEGROES = 8
TOTAL
12
17
11
40
EAST SOUTH CENTRAL (6)
0-14
15-44
45+
TOTAL
6 6
8 8
6 6
20 20
# NEGROES = 8
12
16
12
40
WEST SOUTH CENTRAL (7).
0-14
15-44
45+
TOTAL
6 6
8 9
5 6
19 21
# NEGROES =-6
12
17
11
40
MOUNTAIN (8)
0-14
15-44
45+
TOTAL
6 6
8 9
5 6
19 21
# NEGROES = 1
12
17
11
40
-------
PACIFIC (9)
AGE GROUPS
TOTAL
-3-
SEX
M
20 20
# NEGROES = 2
TOTAL
0-14
15-44
45+
6
9
5
5
9
6
11
18
11
40
"LfJ
C so.
L //• o
-------
APPENDIX 3.6
Copy of "Guidelines and General Information
About collecting Adipose Tissue for the
National Human Monitoring Program for Pesticides"
-------
UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
WASHINGTON, D.C. 20460
Guidelines and General Information
About Collecting Adipose Tissue
For the National Human Monitoring Program for Pesticides
The National Human Monitoring Program for Pesticides is responsible
for determining, on a national basis, the incidences, levels and other
evidences of exposure to pesticides in the general population of the
United States. At present, the program collects and analyzes adipose
tissues for selected pesticides and their metabolites known to be stored
in'the lipid portion of these tissues. The results from the program are
used in evaluating various factors and conditions pertaining to human
health and effective pesticide regulation.
The adipose tissue for this program is secured through the coop-
eration of participating pathologists and medical examiners located
throughout the continental United States. The tissue is obtained from
surgical specimens previously excised for pathological examination and
from postmortem examinations. The specimens are sent to the program
office in Washington, D. C., from which they are subsequently forwarded
to contract laboratories for chemical analysis. Periodic reports of the
laboratory results are sent to each participating pathologist for the
tissues which were submitted under his auspices. Summaries comparing
results with other regions of the country are also provided as they
become available.
-------
2
In order to develop valid information on a national basis, col-
lections must be made according to an experimental design which dictates
the number of samples required according to the demographic distribution
of the population in the appropriate census division. You should have a
copy of the annual quota of samples expected to be collected from your
location on a fiscal year basis. All collections should be made
according to this age/sex/race distribution. You should be able to
collect the number of samples required in each category. Since our
total sample is relatively small and the validity of the results depends
on a high response rate, your participation is particularly important.
If you feel that you will be unable to collect the number of samples
required, please let us know.
Criteria for Selection p_f_ Patients tp_ be_ Sampled
Since the program objective is to reflect pesticide incidences and
levels in the-general (man-on-the-street) population, a few suggestions
are listed here for your guidance:
• The highest priority should be given to satis-
fying the number and demographic distribution
of your annual quota. This quota should be
completed as soon after the start of the fiscal
year as possible.
f Patients having known or suspected pesticide
poisoning should not be sampled. If you are
involved with a potential pesticide poisoning,
we would like to know about it. However,
samples should not be taken for the National
Human Monitoring Program for Pesticides.
-------
3
t Patients exhibiting cachexia or who have been
institutionalized for long periods should not
be sampled for the national program.
Legal Considerations
The National Human Monitoring Program for Pesticides is both
interested and deeply concerned about the legal ramification of this
human research project. Since the program operates in about 40 states,
it is not feasible for us to handle the variety of local or state
interpretations from our location in Washington. Therefore, as a
matter of policy, the legal requirements, i.e., informed consent,
confidentiality, are matters for your consideration and resolution.
Collections for this program must be made in conformance with the appli-
cable HEW guidelines on the protection of human subjects of biomedical
and behavioral research. We will, however, be pleased to* assist you in
any way possible.
We have completed several studies on these matters and do not
believe that they present major obstacles to your participation. In
most documents authorizing postmortem examinations, there is a clause
granting the examining physician permission to remove tissues for
research purposes. We consider this project to be included in that
category. In the case of specimens recovered from your surgical practice,
the use of a small amount of tissue from a previously excised specimen
certainly does not place the patient at risk in any way whatsoever.
As you will notice in our discussion of data needed for each patient
sampled, we do have several mechanisms to assure confidentiality. In
fact, the disclosure or release of certain data is protected by federal
statute. The fees paid to you by our program are solely intended to
remunerate you or your designee for professional services rendered.
-------
4
Collection of Surgical Adipose Tissue
Collect samples of adipose tissue from unfixed specimens which have
been surgically excised for therapeutic reasons. Take special care to
keep samples from different patients separate, correctly and securely
labeled, and avoid their contact with other chemicals, such as paraffin,
disinfectants, preservatives, or plastics.
At least five grams of good quality (subcutaneous, perirenal, or
mesenteric) adipose1 tissue should be collected; avoid fibrous or connective
tissue, i.e., omentum. Place the fat, without any fixatives or preservatives
into the provided chemically-cleaned container; legibly complete and
attach the self-adhesive label in ball-point pen or pencil. The bottle
labels should be affixed before freezing. Store the specimens up-right
in a freezer at -4°F (-20°C) until shipment.
Collection of_ Postmortem Adipose Tissue
Adipose tissue samples must be obtained only from une'mbalmed
cadavers. The interval between death and the collection of tissue
should be as short as possible and must not exceed 24 hours, assuming
refrigeration during the interval. Samples of adipose tissue must weigh
at least five grams and should be placed in the supplied, chemically-
clean container with a completed label affixed. Specimens should be
stored at -4°F (-20°C) without any fixative or preservative until
shipment. Submit only good quality fat; do not submit omentum as it
contains too much connective tissue for satisfactory analysis.
-------
5
Adipose should be taken dry, and should not be rinsed before
placing in the provided containers. Many water supplies contain materials
which would interfere with chemical analysis.
Instruments should be well-rinsed with distilled water and dried
before taking the adipose sample.
Completion £f the Patient Summary Report
A Patient Summary Report should be completed for each patient from
whom a sample was taken. Special attention should be given to the
completeness of the data. All medical information submitted is protected
from disclosure or release by U.S.C. 552, (b) (6); 45 CFR Part 5. First
and last initials, in that order, should be used instead of the complete
name to insure that confidentiality is maintained. The initials, along
with the data of birth, sex, and race, are used in this office to
compose the AMA identification number. The patient's identification
number and/or the pathology department's accession number are for your
information in referring back to the individual patient when you receive
the results of the pesticide analysis.
Confirmed diagnosis should be detailed in the spaces provided.
Only the major ones should be supplied.
Other information required should.be completed as accurately as
possible. The complete forms should be held and sent under the lid of
the insulated container when shipment is made.
-------
Packing and Shipping
Tighten all lids on the specimen bottles carefully. This is
important since we are required to use special aluminum foil cap liners
which make tightening a little difficult. Be certain that a completed
bottle label is firmly attached to each specimen bottle. Wrap each
bottle in gauze or paper to prevent breakage during shipment and to keep
the label on the container. Place the specimen bottles in the insulated
mailer and fill it with dry ice. If you have difficulty obtaining dry
ice, please call us and we can arrange alternative methods of refriger-
ation for you.
A franked addressed label is on the reverse side of the address
card. This card is marked AIR MAIL - SPECIAL DELIVERY. (Do not send
Air Express, please). There is no cost to the sender because of the
franked label. All insulated mailers should have a PERISHABLE-PACKED IN
DRY ICE label visible from all sides on the outside.
Specimens should be mailed on a Monday or Tuesday of a week with n£
federal holidays. This assures that they will arrive before the end of
the work week on Friday.
Patient Summary Reports should be sent in the carton with the
specimens when possible. They can be folded and placed on the top of
the polyfoam lids.
-------
7
Only samples which meet our criteria and are handled according to
the guidelines can be accepted. No substitute containers will be
accepted.
For Further Information
If you have any questions or comments, please contact us. Telephone
(collect): 202/755-8060.
Sandra C. Strassman
Frederick W. Kutz, Ph.D.
National Human Monitoring Program
for Pesticides (WH-569)
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
REPORT NO.
EPA-560/13-79-014
2.
3. RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
Analysis of EPA Pesticide Monitoring Networks
5. REPORT DATE
December, 1979
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
T. Hartwell, P. Piserchia, S. White, R. Lucas
N. Gustafson, A. Sherdon, D. Lucas, D. Myers, J. Batts,
R. Handy, S.
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Research Triangle Institute
P. 0. Box 12194
Research Triangle Park, North Carolina 27709
10. PROGRAM ELEMENT NO.
11. CONTRACT/GRANT NO.
68-01-5848
12. SPONSORING AGENCY NAME AND ADDRESS
Environmental Protection Agency
Office of Toxic Substances
Washington, D. C. 20460
13. TYPE OF REPORT AND PERIOD COVERED
October, 1979-December, 1979
14. SPONSORING AGENCY CODE
15. SUPPLEMENTARY NOTES
16. ABSTRACT
This report describes a brief investigation of six pesticide networks run by
EPA. In this investigation an attempt was made to answer several specific questions
about each network including (i) the sampling procedures used, (ii) selected summary
statistics and tests of significance appropriate for summarizing the data, and
(iii) an examination of trends over time and differences in geographic areas for
particular pesticides.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b. IDENTIFIERS/OPEN ENDED TERMS
c. COSATI Hold/Group
Pesticide Networks:
Human Adipose Tissue,
Soil Residue,
Surface Water and Sediment Residue,
Air,
Soil Application,
Urban Soil,
Statistical Analysis
18. DISTRIBUTION STATEMENT
19. SECURITY CLASS (This Report!
Unclassified
21. NO. OF PAGES
165
20. SECURITY CLASS (This page)
Unclassified
22. PRICE
EPA Form 2220-1 (Rev. 4-77) PREVIOUS EDITION 15 OBSOLETE
------- |