irtN&C,;


EPA/560/5-89/004
                            :-Office6f
                             Tbxfe'Su
                            : Washington, D.C. 20460
EPA 560/5-89-004
March 1990
                    oarison of Airborne
                     >tos Levels Determined by
                               ^^
                                              Direct
   -1  •*. •* ** ^ c
 *  i ^^-"  * ^
v , .^V'^t' *"


   -/* ' »-"?*„'


-------
                                           EPA 560/5-89-004
                                                March,  1990
                    FINAL REPORT
COMPARISON OF AIRBORNE ASBESTOS LEVELS DETERMINED BY
      TRANSMISSION  ELECTRON MICROSCOPY  (TEM)
   USING  DIRECT AND INDIRECT  TRANSFER TECHNIQUES
                    Prepared by:

               Chesson  Consulting,Inc.
            1717 Massachusetts  Avenue,  NW
                Washington, DC 20036

                         and

                     Battelle
                  Arlington Office
               2101 Wilson Boulevard
                Arlington, VA  22201

            EPA Contract No. 68-02-4294
                      for the:

            Exposure Evaluation Division
             Office of Toxic Substances
     Office  of Pesticides  and Toxic  Substances
        U.S.  Environmental Protection Agency
                 401 M Street, S.W.
              Washington,  D.C.   20460

-------
     This document has been reviewed and approved for publication
by the Office of Toxic Substances,  Office of Pesticides and Toxic
Substances, U.S. Environmental Protection Agency.  The use of
trade names or commercial products  does not constitute Agency
endorsement or recommendation for use.
                                11

-------
                    AUTHORS AND CONTRIBUTORS

     This report was prepared by Jean Chesson of Chesson
Consulting, Inc. under subcontract to Battelle.  Jeff Hatfield of
Battelle prepared an earlier draft of the Study 1 (EPA 1988)
results.  The R.J. Lee Group,  Inc., Monroeville, PA performed the
laboratory analysis of samples from Study 1 as part of the study
reported in EPA (1988).

     The EPA work assignment manager was Brad Schultz.
Substantial contributions were also made by Cindy Stroup, Betsy
Dutrow, and Joe Breen of the Exposure Evaluation Division in the
EPA Office of Toxic Substances.
                        ACKNOWLEDGEMENTS

Al Unger and Barbara Leczynski, the Battelle Task Managers and
Project Managers, and Edie Sterrett and Mary Frankenberry, the
EPA Project Officers, provided valuable managerial and
administrative support.  Janice Mesich of JAM Design designed the
report cover.  The peer reviewers, D. Wayne Berman, Michael
Beard, Gary Burdett, Eric Chatfield, Thomas Fishbach, Richard
Lee, James Millette, and Roger Wilmoth, provided many valuable
suggestions.
                               111

-------
                        TABLE OF CONTENTS
EXECUTIVE SUMMARY 	    ix
     Background	    ix
     Results and Conclusions  . 	   x
     Additional Research  	    xi

I. INTRODUCTION 	   1

II.  CONCLUSIONS AND RECOMMENDATIONS  	   3

III.  DESCRIPTION OF ANALYTICAL METHODS 	   5
     A.  Direct Transfer  	   5
     B.  Indirect Transfer  	   5
     C.  Comparative Advantages and Disadvantages  	   6

IV.  INDIVIDUAL STUDIES 	   7
     A.  Study 1 — EPA/GSA Study of Commercial and Public
          Buildings 	   7
     B.  Study 2 — Phase III Abatement Study	    14
     C.  Study 3 -- Lee (1987)  	    15
     D.  Study 4 — Burdett (1985a)	    18
     E.  Study 5 — Toronto Subway	    18
     F.  Study 6 — Lee (1987)  	    20
     G.  Study 7 — Cook and Marklund (1982)  	    22
     H.  Other Data	    23

V.  ANALYSIS OF COMBINED DATA	    27

VI.  DISCUSSION	    33
     A. Bias	    33
     B. Precision	    34

REFERENCES	    37

Appendix A: Definition of Asbestos Structure Types  ....    41

Appendix B: Data Listings	    45

Appendix C: Correlation Coefficients for Study 1  (USEPA
     1988)  	    53

-------
                         LIST OF FIGURES


Figure 1.  Airborne asbestos concentrations measured in Study 1
(EPA/GSA).   If direct and indirect transfer techniques were
equivalent, the points would fall on the dotted line. ...    10

Figure 2.  Airborne asbestos concentrations measured in Study 2
(Phase III).   If direct and indirect transfer techniques were
equivalent, the points would fall on the dotted line. ...    15

Figure 3.  Airborne asbestos concentrations measured in Study 3
(Lee 1987).  If direct and indirect transfer techniques were
equivalent, the points would fall on the dotted line. ...    17

Figure 4.  Airborne asbestos concentrations measured in Study 4
(Burdett 1985).  If direct and indirect transfer techniques were
equivalent, the points would fall on the dotted line. ...    19

Figure 5.  Airborne asbestos concentrations measured in Study 5
(Toronto Subway).   If direct and indirect transfer techniques
were equivalent, the points would fall on the dotted line.     21

Figure 6.  Airborne asbestos concentrations measured in Study 6
(Lee 1987).  If direct and indirect transfer techniques were
equivalent, the points would fall on the dotted line. ...    22

Figure 7.  Airborne asbestos concentrations measured in Study 7
(Cook 1982).   If direct and indirect transfer techniques were
equivalent, the points would fall on the dotted line. ...    24

Figure 8.  Estimated values of the proportionality parameter, (3,
for Studies 1 through 7.  Vertical lines indicate the 95 percent
confidence interval	    31
                                VI

-------
                         LIST OF TABLES

Table 1.  Summary Statistics for Length and Width of
     Chrysotile Asbestos Structures Measured with Direct and
     Indirect TEM	    12
Table 2. Size Distribution of Chrysotile Structures Measured
     in Study 1  (The Body of the Table Gives Number of
     Structures.) 	    13
Table 3. Comparison of Mean Airborne Asbestos Levels
     Obtained by Direct and Indirect TEM Analysis of Study 3
     Samples	    17
Table 4. Comparison of Mean Airborne Asbestos Levels
     Obtained by Direct and Indirect TEM Analysis of Study 4
     Samples (Burdett 1985a)  	    18
Table 5. Fiber Size Distributions of Two Samples from Study
     5 (Chatfield 1986) 	    20
Table 6.  Summary of Major Attributes of the Seven Studies     28
Table 7.  Parameter Estimates for Model Yj  = a  + 6YD, where
     Yj  and YD are Measurements Obtained Using the Indirect
     and Direct Transfer Methods Respectively (95%
     Confidence Intervals in Parentheses) 	    30
                               VII

-------
                        EXECUTIVE SUMMARY
Background

Transmission electron microscopy (TEM)  is the preferred
analytical method for measuring asbestos concentrations in
ambient atmospheres.  The absence of a standard protocol for TEM
analysis and the discovery and refinement of new techniques have
resulted in a variety of procedures which may not necessarily
provide comparable estimates of airborne asbestos concentration.
An important difference between protocols is the use of direct
and indirect transfer techniques.  The direct transfer method was
developed primarily to estimate structure concentration, whereas
the indirect transfer method was developed primarily to estimate
mass concentration.  In a direct transfer the original filter is
prepared for analysis with minimal disturbance of the particles
upon it.  In an indirect transfer, the particles are removed from
the original filter and resuspended on a second filter prior to
microscopic examination.  Although the original spatial
distribution of the particles is lost,  indirect transfer is
thought to provide greater control over analytical precision
through improved distribution of materials over the surface of
the filter.

Early TEM measurements of airborne asbestos used an indirect
transfer method and expressed the results in terms of mass
(ng/m3) .   Fiber concentrations were not reported because it was
thought that the indirect transfer technique might have broken up
larger asbestos structures and artificially inflated the fiber
count.  The U.S. Environmental Protection Agency (EPA) has used
the indirect transfer technique for many of its research
programs, in part to overcome the problem of non-asbestos debris
in some sampling situations, and in part because the type of
filter most suited for direct transfer (polycarbonate) was
thought to be more difficult to handle and transport in the
field.  However, improvements in the direct transfer technique
applied to mixed cellulose filters have made direct transfer a
feasible option.

Prior to carrying out a recent study of airborne asbestos levels
in public buildings (USEPA 1988), EPA convened a meeting of
microscopists and other asbestos measurement experts to determine
the most appropriate analytical protocol.  A direct transfer
method using mixed cellulose ester filters was selected.  A
similar TEM protocol was later specified under the Asbestos
Hazard Emergency Response Act (AHERA) to determine when an
asbestos work site is sufficiently clean for the containment
barriers to be removed.
                                IX

-------
To investigate the relationship between airborne asbestos levels
measured by the two transfer techniques and possibly provide a
basis for comparison with earlier studies based on indirect
transfer, a subset of the samples collected in the 1988 EPA study
were reanalyzed using an indirect transfer method.  This document
reports the results of the EPA analysis and extends the
discussion to include data from six other studies.

Results and Conclusions

The investigation confirmed the generally held opinion that the
direct and indirect transfer methods provide different estimates
of airborne asbestos concentration.  There is insufficient
information, however, to determine the mechanisms responsible for
the difference and thereby recommend one method over the other.
The specific conclusions are listed below followed by
recommendations for further research.

•    TEH analysis of air samples using indirect transfer methods
     tends to provide estimates of total airborne asbestos
     structure concentration that are higher than those obtained
     using direct transfer methods.  This conclusion is
     consistent with general opinion and implies that airborne
     asbestos levels estimated by one method are not directly
     comparable to those estimated by the other.

     Evidence. A review of available data (seven studies)
     revealed this relationship in every study despite variations
     in sampling, analytical, and counting protocols.

•    There is no single factor that can be applied to convert
     measurements made using an indirect transfer method to a
     value that is comparable with measurements made using a
     direct transfer method.  The quantitative relationship
     between estimates obtained by the two transfer methods is
     expected to depend on sampling and analytical protocols as
     well as the nature of the asbestos structures in the air.

     Evidence. In the studies considered here, measurements made
     by the indirect transfer method were 3.8 times to 1,700
     times higher than measurements made by the direct transfer
     method.  The highest value of 1,700 was estimated from a set
     of 45 samples collected in a school district.  The lowest
     value of 3.8 was obtained in an interlaboratory study of 12
     samples of amphibole.

-------
     Provided a single method is applied consistently,  the choice
     of method is not as critical when measurements are to be
     used only for comparative purposes (for example,  comparison
     of airborne asbestos levels inside and outside an abatement
     site).   When measurements are to be interpreted in relation
     to a fixed standard, the choice of method is more important.

     Evidence. Both methods appear to detect changes in airborne
     asbestos concentrations.  Although the relationship between
     the two methods is not strong, higher concentrations
     determined from one method tend to correspond to higher
     levels  obtained by the other.  A statistically significant
     relationship of this type was found between measurements
     made by the two transfer methods in all seven studies.  In a
     study designed to compare indoor and outdoor airborne
     asbestos levels, the same trend was revealed by both
     methods.
     Based on data from the studies considered in this report, it
     seems unlikely that the larger airborne asbestos
     concentrations estimated by the indirect transfer method can
     be explained solely by breakdown of large asbestos
     structures into smaller components.  Alternative hypotheses
     involving interference by debris and association of
     unattached structures may also be important.

     Evidence. In the two studies for which data are readily
     available, the indirect transfer method counted more
     structures than the direct transfer method in all size
     categories.  One would expect to count fewer large
     structures with the indirect transfer method if larger
     asbestos structures were being broken down into smaller
     ones.
Additional Research

The information needed to select the appropriate protocol for a
given situation could be obtained with a relatively modest
research program.  A series of studies is suggested to:

•    Further investigate structure size distributions for direct
     and indirect TEM preparations in order to distinguish among
     alternative hypotheses and thereby determine which method
     more accurately reflects biologically meaningful airborne
     asbestos concentrations; and

•    Compare the spatial distribution of asbestos structures on
     samples prepared by direct and indirect transfer methods in
     order to characterize the precision of each method.


                                xi

-------
                         I. INTRODUCTION
Transmission electron microscopy (TEM) is the preferred
analytical method for measuring asbestos concentrations in
ambient atmospheres.  A measured volume of air is drawn through a
filter.  Particles trapped on the filter are coated with a thin
carbon film and the filter is dissolved by solvent leaving the
carbon film supported on a fine metal mesh grid.  The grid is
examined with the transmission electron microscope.  The absence
of a standard protocol for TEM analysis and the discovery and
refinement of new techniques have resulted in a variety of
procedures which may not necessarily provide comparable estimates
of airborne asbestos concentration.  An important difference
between protocols is the use of direct and indirect transfer
techniques.  The direct transfer method was developed primarily
to estimate structure concentration whereas the indirect transfer
method was developed primarily to estimate mass concentration.
In a direct transfer the original filter is prepared for analysis
with minimal disturbance of the particles that have been
collected on its surface.  In an indirect transfer, the particles
are removed from the original filter and resuspended on a second
filter prior to coating with the carbon film.  Although the
original spatial distribution is lost, indirect transfer is
thought to provide greater control over analytical precision
through improved distribution of materials over the surface of
the filter (Cook and Marklund 1982, Chatfield 1985).

Early TEM measurements of airborne asbestos  (e.g., USEPA 1975,
1979, 1980) used an indirect transfer method and the results were
expressed as mass (ng/m3) .   Fiber concentrations were  not
reported because it was thought that the indirect transfer
technique might have broken up larger asbestos structures and
artificially inflated the fiber count (Chatfield 1978).  The U.S.
Environmental Protection Agency  (EPA) has used the indirect
transfer technique for many of its research programs (e.g., USEPA
1983, 1985, 1986a, Tuckfield et al. 1988), in part to overcome
the problem of non-asbestos debris in some sampling situations,
and in part because the type of filter most suited for direct
transfer (polycarbonate) was thought to be more difficult to
handle and transport in the field.  Comparability with earlier
studies was also an issue.

Ortiz and Isom (1974) developed a direct transfer method for use
with mixed cellulose ester filters, but data collected by
Chatfield  (1986)  indicates that fiber loss is high.  Burdett and
Rood (1982) proposed a direct transfer method incorporating an
etching step.  Their method appears to give results comparable to
direct transfer using a polycarbonate filter (Chatfield 1986),
while benefitting from the easier handling and transport
associated with a mixed cellulose ester filter.  The NIOSH 7402
protocol (NIOSH 1985) also involves a direct transfer method with
an etching step.

-------
Prior to carrying out a study of airborne asbestos levels in
public buildings (USEPA 1988),  EPA convened a meeting of
microscopists and other asbestos measurement experts to determine
the most appropriate analytical protocol.  A direct transfer
method based on the Burdett and Rood protocol was selected.   A
similar TEM protocol (allowing either polycarbonate or mixed
cellulose ester filters)  was later specified in the Asbestos
Hazard Emergency Response Act (AHERA, 40 CFR Part 763) to
determine when an asbestos work site is sufficiently clean for
the containment barriers to be removed.  (The AHERA protocol
differs from the study protocol by restricting fiber counting to
asbestos fibers longer than 0.5 ^m and with an aspect ratio of
5:1 or greater.)

To investigate the relationship between airborne asbestos levels
measured by the two transfer techniques and possibly provide a
basis for comparison with earlier studies based on indirect
transfer, a subset of the samples collected in the 1988 EPA study
was reanalyzed using an indirect transfer method.  The purpose of
this document is to report the results of the EPA investigation
and extend the discussion to include data from other sources.
Conclusions are presented in Section II.  Section III describes
the two transfer methods and discusses their advantages and
disadvantages.  Individual studies are described in Section IV
and results analyzed in Section V.  The discussion in Section VI
suggests areas for future research.

-------
              II.  CONCLUSIONS AND RECOMMENDATIONS
The results from the recent EPA study (USEPA 1988),  together with
a review of six other studies in the literature (Tuckfield et al
1988,  Lee 1987 (two data sets), Burdett 1985a,  Chatfield 1986,
and Cook and Marklund 1982)  lead to the following conclusions:

•    TEM analysis of air samples using indirect transfer methods
     tends to provide estimates of total airborne asbestos
     structure concentration that are higher than those obtained
     using direct transfer methods.  This conclusion is
     consistent with general opinion and implies that airborne
     asbestos levels estimated by one method are not directly
     comparable to those estimated by the other.

     Evidence. A review of available data (seven studies)
     revealed this relationship in every study despite variations
     in sampling, analytical, and counting protocols.

•    There is no single factor that can be applied to convert
     measurements made using an indirect transfer method to a
     value that is comparable with measurements made using a
     direct transfer method.  The quantitative relationship
     between estimates obtained by the two transfer methods is
     expected to depend on sampling and analytical protocols as
     well as the nature of the asbestos structures in the air.

     Evidence. In the studies considered here,  measurements made
     by the indirect transfer method were 3.8 times to 1,700
     times higher than measurements made by the direct transfer
     method.  The highest value of 1,700 was estimated from a set
     of 45 samples collected in a school district.   The lowest
     value of 3.8 was obtained in an interlaboratory study of 12
     samples of amphibole.

•    Provided a single method is applied consistently/ the choice
     of method is not as critical when measurements are to be
     used only for comparative purposes (for example, comparison
     of airborne asbestos levels inside and outside an abatement
     site).  When measurements are to be interpreted in relation
     to a fixed standard, the choice of method is more important.

     Evidence. Both methods appear to detect changes in airborne
     asbestos concentrations.  Although the relationship between
     the two methods is not strong, higher concentrations
     determined from one method tend to correspond to higher
     levels obtained by the other.  A statistically significant
     relationship was found between measurements made by the two
     transfer methods in all seven studies.  In a study designed

-------
     to compare indoor and outdoor airborne asbestos levels,  the
     same trend was revealed by both methods.


•    Based on data from the studies considered in this report,  it
     seems unlikely that the larger airborne asbestos
     concentrations estimated by the indirect transfer method can
     be explained solely by breakdown of large asbestos
     structures into smaller components.  Alternative hypotheses
     involving interference by debris and association of
     unattached structures may also be important.

     Evidence. In the two studies for which data are readily
     available, the indirect transfer method counted more
     structures than the direct transfer method in all size
     categories.  One would expect to count fewer large
     structures with the indirect transfer method if larger
     asbestos structures were being broken down into smaller
     ones.

Selection of an appropriate protocol in a given situation
involves consideration of bias (systematic error) and precision
 (random error).  The conclusions above, combined with opinions
expressed by microscopists, indicate that the indirect and direct
transfer methods differ with respect to bias and precision, but
there is insufficient information to recommend one method over
the other.  The necessary information could be obtained with a
relatively modest research program involving the analysis of
existing data and experiments designed specifically for this
purpose.  It is recommended that studies be performed to:

•    Further investigate structure size distributions for direct
     and indirect TEN preparations in order to distinguish among
     alternative hypotheses and thereby determine which method
     more accurately reflects biologically meaningful airborne
     asbestos concentrations; and

•    Compare the spatial distribution of asbestos structures on
     samples prepared by direct and indirect transfer methods in
     order to characterize this component of precision.

These recommendations are discussed in more detail in Section VI.

-------
             III.  DESCRIPTION OF ANALYTICAL METHODS
A variety of preparation techniques have been used in the
analysis of airborne asbestos samples collected on membrane
filters.  Early work is described in USEPA 1978a and 1978b.  The
descriptions below summarize direct and indirect transfer methods
as they are presently applied.

A.  Direct Transfer

In a direct transfer the original filter is prepared for
microscopic examination.  The direct transfer technique retains
the spatial distribution of particles on the filter and minimizes
disturbance which might change their number and size.

     1.  Polycarbonate Filter
Polycarbonate filters are strong and smooth-surfaced with a
sieve-like construction which makes them particularly suitable
for direct transfer.  After sample collection,  a portion of the
filter is carbon coated in a vacuum evaporator.  Portions of the
coated filter are placed on mesh metal grids and the filter
material is dissolved in chloroform.  The remaining carbon film
is a replica of the original filter surface.  Particles that were
deposited on the surface of the filter are embedded in the carbon
film.

     2.  Mixed Cellulose Ester Filter
Mixed cellulose ester filters are thicker, sponge-like, and have
a more irregular surface than polycarbonate filters.  For this
reason, they are thought to retain fibers better during handling
and transport.  Due to their construction, however, they are
likely to trap fibers below, as well as on, the filter surface.
The irregular filter must be collapsed to form a continuous
surface film suitable for carbon coating.  Current protocols also
include a plasma etching step to remove a thin layer of filter
and further expose trapped particles before applying the carbon
film.  After carbon coating, the procedure is similar to that for
polycarbonate filters with the exception that different chemicals
are used to dissolve the  filter material.

B.  Indirect Transfer

The procedure for indirect transfer is the same irrespective of
the filter type.  A measured fraction of the filter is ashed in a
low temperature plasma asher, the ash sonicated in liquid to
redisperse the particles, diluted as needed, and filtered on to a
second filter (usually a polycarbonate filter).  The second
filter is prepared for microscopic examination according to the
direct transfer method described above.

-------
A second type of indirect transfer, which does not involve
ashing, can be used with polycarbonate filters.  Particles are
washed from the surface of the filter and redeposited on a second
filter.  Ultrasonic agitation may also be used to remove
particulate matter from the filter surface.

C.  Comparative Advantages and Disadvantages

The main advantage of the direct transfer method is that it is
thought to most closely represent the number and size
distribution of asbestos structures present in the sampled air.
This claim is based on the lack of disruption of the sample,
rather than conclusive data.  Since health effects, although
poorly understood, are thought to be related to fiber dimensions,
a method that provides relevant size information is desirable.
The reduced number of preparation steps reduces sample
preparation time and hence cost.  The direct transfer method is
at a disadvantage in dusty atmospheres or with larger volume
samples where debris may obscure asbestos structures.  Also, any
inhomogeneity in the spatial distribution of particles on the
filter is retained and reduces the precision of estimated
airborne asbestos concentrations.

Compared to direct transfer, the indirect transfer method is more
likely to disrupt the number and size distribution of asbestos
structures.  Nevertheless, it may be necessary when the sample is
obscured by a large amount of debris.  Rinsing the cassette
allows particles that may be adhering to the walls to be included
in the sample.  The redeposition phase also provides control of
filter loading  (particles may be concentrated to increase
analytic precision or diluted to permit counting of otherwise
overloaded samples) and the effect of uneven deposition of
particles on the original filter is eliminated.  Improper
filtration can, however, produce an uneven distribution on the
second filter.

Contamination of the sample by asbestos structures unrelated to
the air being sampled is of concern with both transfer methods.
(Serious problems have occurred with polycarbonate filters, USEPA
1986b.)  With direct transfer, contamination of the filter
surface is the main source of contamination.  With indirect
transfer, the additional preparation steps increase the
opportunity for contamination from the filter  (both surface and
internal contamination) and other sources  (Anderson et al, 1989).

-------
                     IV.  INDIVIDUAL STUDIES
A.  Study l — EPA/6SA Study of Commercial and Public Buildings

     1.  Study Design
A primary objective of the original study (USEPA 1988) was to
determine if airborne asbestos levels are elevated in buildings
that have asbestos-containing material (ACM).   A total of 406
filters (387 air samples and 19 field blanks)  were analyzed by
TEM using a direct transfer technique.  The samples were
collected in 49 buildings (339 samples)  and at 48 sites outside
the buildings (48 samples).

As discussed above, previous EPA studies had used an indirect
transfer technique.  To investigate the relationship between
measurements obtained using direct transfer with those obtained
by indirect transfer, 30 of the original 406 filters were chosen
for reanalysis using an indirect transfer technique.  A better
understanding of the relationship between the two transfer
methods might allow comparisons between the results of different
studies.  The reanalysis also provided additional quality
assurance.  In this study, airborne asbestos levels measured
using the direct transfer technique tended to be at the low end
of the range commonly measured in indoor atmospheres.  Reanalysis
using indirect transfer confirmed that asbestos structures were
present on the filters.

The thirty samples were chosen to represent a range of sample
types and asbestos structure concentrations.   Two filters were
chosen at random from the 19 field blanks.  Four filters were
chosen at random from the 48 outdoor samples.   The remaining 24
filters were chosen from the 339 indoor samples.  Their selection
was based on the direct TEM results.  Of the 24, eight filters
were chosen at random from the 282 samples which, with the direct
transfer method, had no structures in ten grid openings.  Eleven
samples were chosen at random from the 52 samples which had
either one or two structures in ten grid openings.  All five of
the filters with three or more structures in ten grid openings
were chosen.  (One structure in ten grid openings corresponds to
16.1 s/mm2 of filter surface.   Two structures  in ten grid
openings correspond to 32.3 s/mm2 of filter surface;  three
structures in ten grid openings correspond to 48.4 s/mm2 of
filter surface.)

     2.  Sampling Protocol
Air samples were collected on 37-mm cellulose ester  (Millipore®)
filters with a pore size of 0.45 /zm.  Approximately 5,000 L of
air were drawn through each filter at a rate of approximately 5
L/min for 16 hours.  Each sample was collected over two
consecutive weekdays during periods of normal building activity

-------
     3.  Analytical Protocols
The direct transfer protocol used in the original study is
described in Appendix B of USEPA (1988).  Sample preparation
involved collapsing a portion of the filter, plasma etching, and
directly coating the filter with a thin layer of carbon by
evaporative deposition under vacuum.  The samples were cleared
with acetone, leaving the particles attached to the carbon film.
The results reported here are based on the examination of ten
grid openings with a total area of 0.062 mm2 at a magnification
of approximately 20,000 X.  (Additional grid openings were
examined on selected filters to investigate the spatial
distribution of asbestos structures.  The results of those
analyses are reported in USEPA (1988)).

The indirect transfer protocol used to analyze the 30 samples
selected for this study was similar to that described in Appendix
B-5 of Tuckfield et al. (1988).  A known portion  (approximately
1/4) of the original 37 mm filter was ashed, suspended in 100 ml
of filtered water, sonicated, and a known aliquot (either 70 ml
or 100 ml) was deposited on a 25 mm cellulose ester filter.  Ten
grid openings with a total area of 0.067 mm2 were examined on
each filter.

Counting procedures for grid examination are given in Appendix B
of USEPA  (1988) and were taken from Yamate et al. (1984).  The
type of asbestos  (chrysotile or amphibole) and type of structure
(fiber, bundle, cluster, or matrix) were recorded along with
length and width measurements.  (Non-asbestos fibers were also
identified.)  Total structure concentrations, as well as separate
estimates for fibers and for bundles, clusters, and matrices
(BCM), are available.  Fibers were defined as asbestos structures
with an aspect ratio of 3:1 or greater.   No minimum length was
designated.

     4.  Data Analysis
The asbestos structure concentration per cubic centimeter of air
(s/cc) is calculated by estimating the number of  structures
deposited on the original filter from the number  counted in ten
grid openings and dividing by the volume of air sampled.  The
computation uses the effective area of the filter (385 mm2 for a
filter 25 mm in diameter and 855 mm2 for a filter 37  mm in
diameter).  The estimated total number of structures on the
original 37 mm filter  (TS) is given by:

               TS = n x (855 mm2/0.062 mm2)

for the direct transfer method and by:

TS = n x  (385 mm2/0.067 mm2) x  (100 ml/aliquot)  x  (855 mm2/area ashed)

for the indirect transfer method.  The symbol n is the number of
asbestos structures counted in 10 grid openings,  aliquot is the

-------
amount of material suspended on the 25 mm filter (either 70 ml or
100 ml) and area ashed is the area of the 37 mm filter ashed
during the indirect preparation (expressed in mm2) .   The
estimated airborne asbestos concentration (s/cc) is obtained by
dividing TS by the volume of air sampled (cc).   The estimated
concentration of asbestos structures on the original filter
(s/mm2)  is  obtained by dividing TS  by 855.

The Wilcoxon Signed-Rank test was used to test whether the two
transfer methods differed in their estimates of asbestos
structures per square millimeter of filter.  The test examines
the median difference between two sets of measurements.  The
degree of association between the two transfer methods was
measured by the Pearson and Spearman correlation coefficients.
Both coefficients vary between +1 and -1 depending on the
strength of the relationship and whether the relationship is
positive or negative.  The Pearson correlation coefficient is
calculated from the actual airborne asbestos concentrations and
detects linear relationships.  The Spearman coefficient is a
nonparametric measure of association based on ranks.

Summary statistics for the lengths and widths of asbestos
structures were produced separately for the direct and indirect
TEM measurements; the Wilcoxon Rank-Sum test was used to test for
statistically significant differences.

     5.  Results
Asbestos structure concentrations for the 28 air samples are
reported in Appendix B.  The two field blanks,  chosen at random
from the 19 blanks analyzed with direct TEM in the original
study, were found to have zero and three asbestos structures,
respectively, in the ten grids counted for the indirect
preparation.  The three structures counted were all fibers.
These counts correspond to structure concentrations of 0 and 96.2
s/mm2 on the filter and 0  and 0.016 s/cc  in the  air  when  sampling
5000 liters of air.  The average background contamination is 48.1
s/mm2 or 0.008  s/cc.   This is an order of  magnitude less than
all but three of the nonzero indirect TEM asbestos structure
concentrations measured in the study.

The relationship between structures/cc measured by indirect TEM
and fibers/cc measured by direct TEM is illustrated in Figure 1.
(To maintain consistency with the other studies discussed below,
structures per cubic centimeter are plotted against fibers per
cubic centimeter because these are the units reported in the
remaining studies.) The two field blanks are not included in this
or any subsequent analyses.

The data show that concentrations measured by the indirect
transfer method are greater than those measured by the direct
transfer method.  The result holds for total structure
concentrations, fiber concentrations, and BCM (bundle, cluster,

-------
09 r-
tr
o
H—
o5
w
| 0.1
H ,
& <
T3
c
~ 0.05 <
)
I
'
0|
(



*


* • •
l_
j * 	 * 	 * 	
D 0.01 0.02 0.03 0.
Direct Transfer (s/cc)







04

  Figure 1.   Airborne asbestos  concentrations  measured in Study 1
  (EPA/GSA).   If direct  and  indirect  transfer  techniques  were
  equivalent,  the points would  fall on the  dotted line.
and matrices) concentrations.  The Wilcoxon Signed-Rank test
yields p-values less than 0.0001 for all three comparisons.

The correlation coefficients are reported in Appendix C.  None of
the Pearson correlation coefficients are significantly different
from zero at the 5 percent level.  However a Spearman correlation
coefficient of 0.37 suggests a weak correlation between direct
structure concentration and indirect fiber concentration
(p < 0.05).  Spearman correlation coefficients for direct
structure concentration versus indirect structure concentration
(0.35), direct fiber concentration versus indirect fiber
concentration (0.31), and direct BCM concentration versus
indirect fiber concentration (0.34) are also significant at the 5
percent level.  Recall that the Spearman correlation coefficient
is sensitive to a broader range of relationships than the Pearson
correlation coefficient.

                                10

-------
The lengths and widths of chrysotile structures detected by the
two transfer methods are summarized in Table 1.  (Amphibole
structures, which tend to be larger than chrysotile structures,
were relatively rare in this study.  On the 28 samples analyzed
by both direct and indirect TEM, 11 amphibole structures were
counted by indirect TEM and 8 amphibole structures by direct
TEM.)   Chrysotile fibers detected using the indirect transfer
method tend to be shorter and thinner than those detected by the
direct transfer method.  The difference in fiber width is
statistically significant (p = 0.0003, t-test), whereas the
difference in fiber length is not (p = 0.79).   No statistically
significant size differences were found between chrysotile BCM
detected by each method.

Chrysotile structure size distributions are shown in Table 2.
Although the largest differences are seen with the smaller
fibers, more fibers of all sizes were counted with the indirect
transfer method (Table 1).  Also, more BCM were counted with the
indirect transfer method than with the direct transfer method and
the average size of BCM counted with the indirect transfer method
was no smaller than the average size of BCM counted with the
direct transfer method.  Therefore, it seems unlikely that the
larger airborne asbestos concentrations estimated by indirect TEM
can be explained solely by breakdown of large structures into
smaller components.
                                11

-------
Table l.  Summary Statistics for Length and Width of Chrysotile
Asbestos Structures Measured with Direct and Indirect TEM
Fibers
                      Direct

                  Length      Width
                      Indirect

               Length    Width
 Mean               0.94
 Median             0.50
 Standard Deviation 1.25
 Standard Error     0.27
 Sample Size        22
0.08
0.05
0.05
0.01
22
0.88
0.60
0.97
0.06
 249
0.06
0.05
0.02
0.001
249
BCM
                      Direct

                  Length      Width
                     Indirect

              Length     Width
 Mean               1.26
 Median             0.80
 Standard Deviation 0.81
 Standard Error     0.27
 Sample Size        9
0.11
0.05
0.11
0.04
9
1.40
1.00
2.01
0.15
 173
0.12
0.10
0.13
0.01
173
                                12

-------
Table 2. Size Distribution of Chrysotile Structures Measured in
Study 1  (The Body of the Table Gives Number of Structures.)
a)  Direct Transfer Method

Width Category              Length Category

1
2
3
4
5
Total
1
0
0
0
0
0
0
2
0
5
2
0
0
7
3
0
6
7
0
0
13
4
0
2
1
0
0
3
5
0
3
3
1
0
7
6
0
0
0
0
0
0
7
0
1
0
0
0
1
Total
0
17
13
1
0
31
b)  indirect Transfer Method

Width Category              Length Category
1
2
3
4
5
Total

1
1
0
0
0
0
1
Key:
Structure Length
Category
1 1
2
3
4
5
6
7
0.25
0.50
0.75
1.00
3.00
5.0
< 1
< 1
< 1
< 1
< 1
< 1
2 3
8 9
53 86
6 56
0 1
0 0
67 152
(Mm)
< 0.25
< 0.5
< 0.75
< 1.0
< 3.0
< 5.0

4 567 Total
2 11 0 0 31
26 63 4 6 238
18 52 3 4 139
1 6 3 2 13
0 001 1
47 132 10 13 422
Structure Width (jLtm)
Category
1 w < 0.05
2 0.05 < w < 0.1
3 0.10 < w < 0.25
4 0.25 < W < 0.5
5 0.50 < w


                                13

-------
B.  study 2 — Phase III Abatement Study

     1.  Study Design
The main objective of the original study (Tuckfield et al. 1988)
was to compare airborne asbestos levels in six schools before,
during, and after removal of the asbestos-containing material.  A
secondary objective was to compare estimates of airborne asbestos
concentrations made by TEM using direct and indirect transfer
techniques.

Two side-by-side samples were collected at each sampling site.
One sample was collected on a mixed cellulose ester filter (47 mm
diameter, 0.45 /im pore size), and the other on a polycarbonate
filter (37 mm diameter, 0.4 jm pore size).  A total of 103 mixed
cellulose ester filters were analyzed by TEM using an indirect
transfer method.  Based on the indirect TEM results, 25
polycarbonate filters were selected for analysis using a direct
preparation technique.  The filters were chosen to represent a
range of airborne asbestos concentrations.  Three of the 25
filters could not be successfully prepared for direct analysis
due to heavy filter loadings.  Thus, TEM analyses of 22 mixed
cellulose ester/polycarbonate filter pairs were available for
comparison.

     2.  Sampling Protocol
Approximately 10,000 L of air were sampled at a flow rate of
5 L/min.  Samples were collected over 5 days during periods of
normal activity.  Each sampling pump was equipped with two
orifices.  The mixed cellulose ester filter was attached to one
orifice, the polycarbonate filter to the other.

     3.  Analytical Protocols
The direct and indirect transfer protocols are given in Appendix
B of the study report  (Tuckfield et al. 1988).  The mixed
cellulose ester filters were ashed and sonicated prior to
deposition on a polycarbonate filter.  Subsequent steps for both
protocols were similar.  The polycarbonate filters were carbon
coated and the filter material dissolved.  Bundles, clusters, and
matrices are recorded, but not included in the estimates of fiber
concentration.

     4.  Results
The results of the 22 pairs of analyses are listed in Appendix B.
Figure 2 shows that with the exception of one sample (Sample 85-
324) estimates based on the direct transfer method are smaller
than those based on the indirect transfer method.  In several
instances, fibers were detected on the mixed cellulose ester
filter using the indirect transfer method, but not on the
corresponding polycarbonate filter using the direct transfer
method.  Since different filter media were used, the observed
differences could be due to the filter medium as well as the
preparation method.  It has been suggested that asbestos

                                14

-------
           40
           30
         O
         o
         03
         >4—
         CO

         c6 20
         o
         Q)
           10
                        1           2
                          Direct Transfer (s/cc)
  Figure  2.   Airborne asbestos concentrations measured in Study 2
  (Phase  III).   If direct and indirect transfer techniques were
  equivalent,  the points would fall on the dotted line.


structures may have been lost from the filters during  the on-off
cycling of the pumps.

C.  Study 3 — Lee (1987)

     1.  Study Design
At the request of attorneys representing National Gypsum Company,
62 filters collected in EPA's 1983 study of airborne asbestos
levels in schools  (USEPA 1983) were analyzed by TEM using a
direct transfer method  (Lee 1987).  The 62 filters were chosen
because sufficient filter material remained for further analysis.
In the original study samples were analyzed by TEM using an
indirect transfer method.  The objective of the study  was to
document exposure to airborne asbestos in schools.  Forty eight
asbestos-containing sites in 25 schools were sampled.  An outdoor
ambient sample, and an  indoor control sample from an area without
                                15

-------
ACM were also collected at each school.  Of the 62 filters later
analyzed by direct TEM, 46 were analyzed in the original study.
(Most of the remaining filters are blanks that were collected but
not analyzed.)

     2.  Sampling Protocol
Samples were collected on 47 mm diameter, mixed cellulose ester
filters with a pore size of 0.45 /urn.  Approximately 10,000 L of
air were sampled over 5 days at a rate of 5 L/min.  Sampling took
place during the hours of normal school activity.


     3.  Analytical Protocols
The indirect transfer protocol is given in Appendix E of USEPA
(1983).  A quarter of the filter was ashed, sonicated, and
filtered onto a polycarbonate filter prior to carbon coating and
microscopic examination.  The protocol calls for recording of
bundles, clusters, and matrices, but does not include them in the
reported fiber concentration.

The direct transfer analysis was performed according to the
protocol specified in the AHERA proposed rule (52 FR 15820, April
30, 1987).  The filter was collapsed, etched, carbon coated and
dissolved.  Fibers are defined as structures longer than 0.5 jum
with an aspect ratio of 5:1 or greater.  Bundles, clusters, and
matrices are included in the total structure count.

     4.  Results
Appendix B lists results for the 46 pairs of analyses.  The
relationship between the estimates obtained by direct and
indirect transfer methods is illustrated in Figure 3.  (Note that
one extreme value is not included in Figure 3.  Based on the size
and type of particles on this filter, Lee  (1987) suggested that
this sample had been deliberately "spiked.")  As expected,
estimates based on the direct transfer method are lower than
those based on the indirect transfer method.  Although the 46
samples were selected according to availability and are not
necessarily a representative sample of the experimental
categories, the direct transfer results show a similar trend to
that observed in the original study  (Table 3).  Measured airborne
asbestos levels are lowest for outdoor samples and highest for
indoor samples at sites with ACM.
                                16

-------
            140
            120
          O 100
          Q)
         H—
          CO
          c
          03
          2
         T3
80
            60
            40
            20
                    0.02    0.04    0.06    0.08    0.1

                            Direct Transfer (s/cc)
                                       0.12
0.14
  Figure 3.   Airborne  asbestos concentrations  measured in Study  3
  (Lee 1987).  If direct and indirect transfer techniques were
  equivalent, the points would fall on the  dotted line.
Table 3. Comparison of Mean Airborne  Asbestos Levels Obtained by
Direct and  Indirect TEM Analysis of Study 3 Samples
Type of Site
Asbestos sites
Indoor Control
Indoor Control
excluding "spike"
Outside Ambient
No. of
Samples
27
9

8
10
Direct
Transfer
(s/cc)
0.021
2560.400

.007
.001
Indirect
Transfer
(f/cc)
29.49
4.80

4.09
0.28
                                 17

-------
D.  Study 4 — Burdett (1985a)

     1.  Study Design
Fifteen samples collected in EPA's 1983 study in schools (USEPA
1983)  were reanalyzed by Burdett (1985a)  as part of an inter-
laboratory exchange.  The basis for selection of the 15 samples
is not stated, but they consisted of 6 samples from sites with
ACM, 4 indoor controls, 4 outdoor samples,  and 1 blank.

     2.  Sampling Protocol
The sampling protocol is described under Study 3 above.

     3.  Analytical Protocols
Burdett used the direct transfer method described in Burdett and
Rood (1983).  This protocol is similar to the AHERA protocol.  A
total structure count was obtained by summing the individual
counts for fibers and bundles,  clusters,  and matrices.  The
indirect transfer method is described under Study 3 above.

     4.  Results
The fifteen pairs of results are listed in Appendix B and plotted
in Figure 4. The relationship between the two sets of
measurements is similar to that seen in Study 3.  Table 4 shows
that the increasing trend in measured airborne asbestos levels
from outdoor ambient to indoor sites with ACM is apparent despite
the small number of samples.


Table 4. Comparison of Mean Airborne Asbestos Levels Obtained by
Direct and Indirect TEM Analysis of Study 4 Samples (Burdett
1985a)
Type of Site No. of
Samples
Asbestos sites
Indoor Control
Indoor Control
excluding "spike"
Outside Ambient
6
4

3
4
Direct
Transfer
(s/cc)
0.059
0.013

0.007
.001
Indirect
Transfer
(f/cc)
27
25

7
0
.48
.50

.44
.76
E.  study 5 — Toronto Subway

     1.  Study Design
Chatfield (1986) reports analyses of 8 samples collected in the
Toronto Subway System.  The samples were initially analyzed by
TEM using a direct transfer method.  Later they were reanalyzed

                                18

-------
80
^~*. 60
O
.C)
CD
CO
C§ 40
CD
c
"~ 20
1
(
01
c



0

•
.
) 0.05 0.1 0.15 0.2 0.25






Direct Transfer (s/cc)
  Figure 4.   Airborne asbestos concentrations  measured in Study 4
  (Burdett  1985).   If direct and indirect transfer techniques
  were equivalent,  the points would fall on the dotted line.
using an indirect transfer method.  The objective of the original
study is not stated.  Sampling and analysis details not given in
Chatfield (1986) were obtained through personal communication
with Dr. Chatfield.

     2.   Sampling Protocol
Samples were collected on 47mm, 0.4/im polycarbonate filters.

     3.   Analytical Protocols
The direct transfer analysis involved carbon coating and
dissolution of the filter in chloroform following the Yamate
protocol (Yamate et al 1984).

For the indirect transfer analysis, filters were washed in
double-distilled water and the detached particulate ashed to
remove organic material.  After redispersion in water, the

                                19

-------
residual ash was prepared for analysis according to the direct
transfer method.  There was no ultrasonic treatment.  Fiber
concentrations are reported for both preparation methods.
Bundles with a 3:1 aspect ratio or greater were counted as one
fiber.  To the extent possible, the individual components of a
cluster were counted.

     4.  Results
Results for the eight pairs of analyses are listed in Appendix B
and illustrated in Figure 5.  Reported concentrations are higher
for the indirect preparation.  Fiber length distributions are
reported for two of the samples (Table 5).  The increased number
of fibers for the indirect method is due mainly to an increase in
the number of short fibers.  However, there is also a small
increase in the number of longer fibers.

Table 5. Fiber Size Distributions of Two Samples from Study 5
(Chatfield 1986)
Fiber Length
    (/urn)
                    Sample SH43

                 Direct   Indirect
                Transfer  Transfer
                                           Sample LA49
                                               Direct
                                              Transfer
                                                 Indirect
                                                 Transfer
0.50
0.73
1.08
1.58
2.32
3.41
5.00
7.34
-  0.73
   I.
   1,
   2.
-  5
   08
   58
   32
-  3.41
   00
 7.34
10.77
4
1
1
0
2
0
0
0
70
56
14
 9
 3
 2
 2
 0
2
1
0
0
0
0
0
0
4
1
1
0
0
0
0
0
F.  study 6 — Lee  (1987)

     l.  Study Design
As part of the investigation described under Study 3, 12 of the
filters analyzed by direct TEM were reanalyzed by the same
laboratory using an indirect transfer preparation technique.  The
filters were selected based on the results of the direct
analysis.  Unlike the larger set of 46 pairs of results described
in Study 3, the set of 12 pairs does not include the effects of
differences between laboratories.

     2.  Sampling Protocol
The sampling protocol is described under Study 3 above.
                                20

-------
3
O
2.5
'o"
O „
H^ 2
to
c < c
ro 1-5
r^
o
-^ 1
TJ
_c
0.5

0
C



-





*



-
V* 	 < 	 * 	 , 	 ; 	
) 0.05 0.1 0.15
Direct Transfer (s/cc)
















  Figure 5.   Airborne  asbestos  concentrations  measured  in  Study 5
  (Toronto  Subway).  If direct  and  indirect  transfer techniques
  were  equivalent, the points would fall  on  the  dotted  line.


     3.  Analytical  Protocols
The direct transfer  protocol is described under Study 3 above.
Few details are given about the indirect protocol other than
ashing lasted approximately 75  minutes and the objective was to
duplicate the protocol used in  Study 3 (Lee 1987).

     4.  Results
The 12 pairs of results are listed in Appendix B and plotted in
Figure 6.  The relationship between the two sets of measurements
is similar to that seen in Studies 3 and 4.    (All three studies
use samples from the same EPA study, although not necessarily the
same samples.)
                                21

-------
           10
         o
         o
         CD
         ^H-
         w
         c
         03
         O 4
         CD
             i
                     0.02
0.04
0.06
0.08
                                                        0.1
                           Direct Transfer (s/cc)
  Figure 6.   Airborne asbestos concentrations measured in Study  6
  (Lee 1987).   If direct and indirect transfer techniques were
  equivalent,  the points would fall on the dotted line.
G.  Study 7 — Cook and Marklund  (1982)

     1.  Study Design
Pieces of filter from twelve air  samples  collected by the
Minnesota Department of Health in 1975 were  sent  to six
laboratories as part of an interlaboratory study.   Each
laboratory analyzed all twelve samples using their standard
analytical protocol.  Unlike studies  1 through 6  in which
chrysotile is the only, or predominant, type of asbestos, these
samples contained amphibole.

     2.  Sampling Protocol
The samples were collected on 1.2 fj.ru  pore size mixed cellulose
ester  filters (Millipore®) over a period  of  approximately 55
hours  so that visibly heavy sample loadings  were  achieved
                                22

-------
(approximately 5 cubic meters of air per square centimeter of
filter).

     3.   Analytical Protocols
Each laboratory followed a different protocol.   Three
laboratories used some type of direct preparation and three used
some type of indirect preparation.   The protocols labelled
"LTA/C-coat, Nuc-Jaffe" and "C-Coat, Direct/Jaffe" have been
selected as being most similar to the direct and indirect
transfer techniques currently in use.  LTA refers to low
temperature ashing, C-coat to carbon coating, and Jaffe to use of
a Jaffe wick.

     4.   Results
The twelve pairs of results are listed in Appendix B and plotted
in Figure 7.  Cook and Marklund disagree with earlier claims by
Peters and Doerfler (1978) that an increase in the number of
fibers counted using an indirect transfer technique is caused by
fracturing of larger fibers.  Although the percentage of large
fibers measured in with the indirect transfer technique was less
than the percentage measured with the direct transfer technique,
the number of fibers in every size category was greater using the
indirect transfer technique.

H.  Other Data

Additional data comparing measurements made by TEM using direct
and indirect transfer methods are reported by USEPA (1978a),
Steen et al. (1983), Sebastien et al. (1984), Burdett (1985b) and
Chatfield (1986).  These data are not included in the combined
analysis in Section V below because they involve only a small
number of samples, or differ in substantial ways from the six
studies described previously.

In USEPA 1978a, Samudra et al. describe the results of the
analysis of a single sample by direct and indirect transfer by
each of five analysts.  The sample, which was part of a round-
robin test,  was collected on a 0.4 /um polycarbonate filter at 560
L/min for one hour at the Johns-Manville Plant in Waukegan,
Illinois.  Four of the five analysts obtained a higher estimate
of airborne chrysotile fiber concentration with the indirect
transfer method.  The ratio of indirect to direct measurements
ranged from 0.5 to 8.7 with an average of 4.9.   The relative
frequency of short fibers was greater with the indirect transfer
method,  but larger numbers of fibers of all lengths were counted
with indirect transfer.

Steen et al. report airborne asbestos concentrations only for
fibers longer than 5 /im.  Their indirect measurements tend to be
higher than their direct measurements, but compared to studies
where all fiber sizes are reported, the magnitude of the
difference is small.

                                23

-------
         o
         o
           0.5
           0.4
           0.3
         c
         cd
         o
         £ 0.2
           0.1
                   0.05     0.1      0.15     0.2
                          Direct Transfer (s/cc)
0.25
0.3
  Figure 7.   Airborne asbestos concentrations measured in Study  7
  (Cook 1982).   If direct and indirect transfer techniques were
  equivalent,  the points would fall on the dotted line.


A study of ambient airborne asbestos concentrations  in Quebec
mining towns was preceded by a methodological  study  in order to
determine how the air samples should be analyzed  (Sebastien  et
al. 1984).  Eighteen air samples were analyzed using an  indirect
transfer technique.  Only four of the 18  samples  had loadings
sufficiently low to permit analysis by the direct transfer
technique.  The direct measurements were  0.006, 0.032, 0.002, and
0.007 s/cc.  The corresponding indirect measurements were  0.084,
0.207, 0.016, and 0.244 s/cc.  (Note that in Table VIII  in
Sebastien et al. the direct and indirect  labels appear to have
been reversed.)  The authors conclude that more smaller  fibers
are counted by the indirect method and that fiber breakage during
the ultrasound treatment is not sufficient to  explain the
increase.
                                24

-------
As part of a study of fiber release from amosite insulation,
Burdett (1985b) analyzed three samples by both direct and
indirect transfer techniques.  The concentrations obtained with
direct transfer were 0.0005, 0.33, and 0.06 f/cc.  The
corresponding indirect measurements were 0.003, 0.43, and 0.25
f/cc.  Slightly smaller fiber widths and lengths were measured
with the indirect transfer method.

Chatfield (1986) compares fiber counts obtained using direct and
indirect transfer methods on laboratory generated samples.  The
two transfer methods give comparable results for a sample
generated with a single fibril aerosol.  The indirect transfer
method gives considerably higher counts for a sample generated
with an aggregated chrysotile aerosol.
                                25

-------
                  V.  ANALYSIS OF COMBINED DATA
Table 6 summarizes the main features of the seven studies
described in the previous section.  The studies differ in details
of the sampling and analytical protocols and in the type of
asbestos structures included in the estimates of airborne
asbestos concentration.  Inter-laboratory differences such as
quality of TEM specimen preparation, identification criteria, and
analyst skill are also expected to contribute to differences in
estimated concentrations.  Therefore, it is not surprising that
the relationship between direct and indirect measurements varies
from study to study.  In addition, variation from study to study
is expected because of differences in the size distribution of
asbestos structures in the sampled air.

A model of the form

                         Y,  =  a +  BYD,

where Y,,  and YD are the  indirect and direct measurements
respectively, and a, and & are unknown parameters to be
estimated, was fitted separately to each data set.  Since Yj  and
YD are both  subject to measurement error,  and  no distinction  is
made between explanatory and response variables, standard linear
regression techniques for estimating a and B are inappropriate.
Instead, a and 6 are estimated by a nonlinear constrained maximum
likelihood estimation technique (Britt and Lueke 1973).  Y;  and
YD are assumed to  be normally  distributed  with variance a2.

An example of the use of this technique and additional references
appear in Bishop et al (1981).  (The model was also applied to
ln(Yj)  and ln(YD) giving  a model with measurement errors which
increase with the mean.  Since the pattern of results is
essentially the same for both models, and interpretation of the
log model is more difficult, only the results for the original
model are presented.)  If there were perfect agreement between
the two TEM methods, a would be 0 and 6 would be 1.  The strength
of the relationship is indicated by the Pearson correlation
coefficient.  A correlation coefficient of 0 indicates no
correlation.  A correlation coefficient of 1 indicates maximum
positive correlation.  The Pearson correlation coefficient may be
compared with the Spearman correlation coefficient which is based
on ranks, and therefore  is sensitive to other types of
relationships in addition to linear relationships.
                                27

-------
Table 6.   Summary of Major Attributes of the Seven Studies
Study  Collection Medium
       Type  Diam.  Fore
             Lab
Protocol
Counting
           Dir  Ind   Direct  Indirect  Direct Indirect
1
2a

3
4
5
6
7
MCE
PC
MCE
MCE
MCE
PC
MCE
MCE
37mm
37mm
47mm
47mm
47mm
47mm
47mm
•p
0. 45/im
0. 4 /im
0. 45/im
0.45/im
0. 45jum
0.4 Mm
0.45 /urn
1.2 /im
A
B

A
C
D
A
E
A
B

B
B
D
A
F
Mod B&R
Yamate

AHERA
B&R
Yamate
AHERA
C/J
OTS
OTS

OTS
OTS
Wash
ETC
LTA/C/J
s/cc
f/cc

s/cc
s/cc
f/ccb
s/cc
•p
f/cc
f/cc

f/cc
f/cc
f/ccb
f/cc
•p
aln  this  study  the  PC  filter was  analyzed  by  direct  TEM and the MCE
filter was analyzed by indirect TEM.  In the remaining studies both
methods were applied to the same filter.
Key:
   Filter Type
   Laboratory
   Protocol
   Counting rule
MCE       mixed cellulose ester
PC        polycarbonate

A         RJ Lee Group (formerly ETC)
B         Battelle
C         UK Health and Safety Executive
D         Chatfield
E, F       Unnamed

Mod B&R   modified Burdett and Rood
OTS       OTS method (USEPA 1983)
Yamate    Yamate (1984)
AHERA     AHERA proposed rule (52 FR 15820)
B&R       Burdett and Rood (1983)
Wash      Washing without ultrasonic treatment
ETC       Ashed for 75 minutes,  details not given
C/J       Carbon coated,  Jaffe wick
LTA/C/J   Low-temp asher, carbon coated, Jaffe wick

s/cc      structures/cc  (includes BCM)
f/cc      fibers/cc (excludes BCM)
f/ccb     Bundles with a 3:1 aspect ratio or greater
          were counted as one fiber.  To the extent
          possible, individual components of a cluster
          were counted.  (E. Chatfield, pers. comm.)
                                  28

-------
Estimates of a and 6 are given in Table 7 together with their 95%
confidence intervals.  For Study 2 the analysis was repeated with
one extreme point (see Figure 2)  excluded.  Subsequent discussion
refers to the analysis of the reduced data set.  Confidence
intervals for a all include 0.  All seven studies have estimates
of the proportionality parameter, B, which are greater than 1
(Figure 8), indicating that estimates based on indirect transfer
measurements are larger than those based on direct transfer
measurements.  R is significantly greater than 1 (p < 0.05) in
four of the seven studies.

The Spearman correlation coefficient indicates a statistically
significant positive relationship between indirect and direct
measurements for all studies except Study 6.  A statistically
significant linear relationship is indicated by the Pearson
correlation coefficient in Studies 3 through 6.  The apparent
inconsistency between the two correlation coefficients for Study
6 is most likely due to one sample that had the largest
concentration measured by both direct and indirect methods (see
Figure 6).  This sample would tend to increase the Pearson
correlation coefficient, but have less effect on the Spearman
correlation coefficient.  A significant correlation is less
likely to be obtained when there is a small number of samples in
the data set.

The proportionality parameter, /3, varies considerably between
studies.  The smallest value of 3.8 may reflect the type of
asbestos involved—amphibole rather than predominantly
chrysotile.  All analyses in Studies 1 and 6 were done by
Laboratory A.  Therefore differences in the estimated values of /3
between these two studies cannot be attributed to laboratory
differences.  Protocols did differ slightly, however.
Concentrations measured in Study 1 were low by both direct and
indirect transfer methods.  Over 80 percent of the original
direct analyses were zero.  Consequently, the range of
concentrations in Study 1 may not be sufficient to gain a
quantitative estimate of the relationship between the two
transfer methods.

Studies 3, 4, and 6 involved samples from the same original study
(USEPA 1983), although not necessarily the same samples.
Assuming that the nature of the airborne asbestos material was
similar across the three studies, differences between the
estimates of ft (1,670, 755, and 109) may reflect mainly
differences between laboratories and protocols.
                                29

-------
Table 7.  Parameter Estimates for Model Y.  = a + BYD, where
YD  are  Measurements  Obtained Using the Indirect and Direct
Transfer Methods Respectively (95% Confidence Intervals in
Parentheses)
                                                              and
Study Sample
Size
I 28

2 22

2t 21

3 45

4 15

5 8

6 8

7 12

a
-0.07
(-0.23, 0.10)
11
(6, 16)
2.6
(-11.6, 16.9)
-4.3
(-17.8, 9.3)
-2.6
(-25.4, 20.3)
-0.6
(-1.3, 0.05)
-1.2
(-3.8, 1.4)
-0.20
(-0.61, 0.22)
/? Pearson
Correlation
24
(-11, 59)
0.09
(-3.0, 3.1)
110
(-4, 220)
1,700
(1,000, 2,300)
760
(220, 1,300)
28
(18, 37)
110
(28, 190)
3.8
(0.4, 7.2)
0.

-0.

0.

0.

0.

0.

0.

0.

26

07

41

65***

61**

92***

74*

52

Spearman
Correlation
0.37*

0.41*

0.55**

0.67***

0.72**

0.81**

0.32

0.57*

t Sample 85-324 excluded

*   p < 0.05
**  p < 0.01
*** p < 0.001
                                30

-------
        2500
        2000
      0)
        1500
      CO
      LU

      0) 100°
      •4—"
      0)

      cc
      CO
      Q_
         500
          0 -
        -500
                     I
                            3      4
                                 Study
Figure 8.   Estimated values  of the proportionality parameter,
/?, for Studies 1 through 7.   Vertical lines  indicate the 95
percent  confidence interval.
                                31

-------
                         VI.  DISCUSSION
An analytic method should be sufficiently accurate for its
intended purpose.  Accuracy has two components: bias and
precision.  Bias refers to a systematic deviation of the measured
value from the true value of the quantity being measured.  In
this case the objective is to characterize exposure in a
biologically meaningful way, that is, in terms of the number and
type of structures that are inhaled.  Precision refers to the
uncertainty associated with repeated measurements of the same
quantity.  The direct transfer method is often characterized as
being less biased than the indirect transfer method, whereas the
indirect transfer method is considered more precise by some
researchers.  Neither of these claims is supported by extensive
data.  Bias and precision are discussed in turn below, together
with suggestions for further research that could assist in
selecting the appropriate analytical method for a given
situation.

A. Bias

Bias must be considered within the context of the application.
If measurements are to be used in a comparative manner (e.g.,
comparing airborne asbestos levels inside and outside a
building), a bias that applies equally to both sets of
measurements may not affect the comparison.  If, however, the
objective is to measure exposure in order to assess risk, a bias
may have a significant impact on the interpretation of the data.
Although the details are controversial, it is thought that the
dimension of asbestos structures is important in determining the
incidence of disease.  Special attention should be devoted to
minimizing bias with respect to asbestos structures that
contribute most to disease incidence.  (Note that the
contribution is determined not only by relative potency of
asbestos structures of different sizes, but also by their
relative abundances.)  An ideal measurement method would mimic
the effect of respiration, etc. on complex structures (BCM) so
that those that readily disintegrate would be represented by
their individual components, while those that are firmly linked
would be counted and sized as single structures.


The studies considered in this paper all support the generally
accepted belief that airborne asbestos concentrations estimated
by an indirect transfer method are larger than those estimated by
a direct transfer method.  Breakdown of larger structures during
the ashing, sonication, and resuspension steps is assumed to be
the main explanation for the difference.  Fiber size information
from Studies 1 and 5, however, does not provide strong support
for this hypothesis.  Although more small fibers are counted
using an indirect transfer method, there is not a corresponding

                                33

-------
decrease in the number of large fibers and BCM, nor in the size
of the BCM.

Chatfield  (1986)  provides two additional hypotheses for the
larger structure counts obtained with an indirect transfer
method.  First, with the direct transfer method, structures may
be hidden by organic debris.  (This hypothesis was also suggested
by Sebastien et al, 1984.)  The effect is likely to be greatest
for small structures, but applies to structures of all sizes.
During indirect transfer the debris is removed by ashing, thereby
improving visibility and increasing the structure count.  Second,
with the direct transfer method, small structures loosely
associated with larger structures (for example, touching but not
bonded) are counted as a single structure.  During indirect
transfer, these structures are disassociated from the larger
structures and are counted as individual structures.

All three mechanisms may play a role to a varying degree under
different circumstances.  Note that predictions depend on the
size distribution of asbestos structures in the sampled air.
When only small fibers are present, the breakdown hypothesis
would predict little difference between direct and indirect
preparations whereas the debris hypothesis would predict higher
measurements with the indirect preparation.  When the majority of
structures are complex, the breakdown hypothesis would predict
higher measurements with the indirect preparation whereas the
association hypothesis would predict little difference.

Given that measurements by indirect TEM are generally higher than
those by direct TEM, it is important to determine whether
indirect measurements incorporate a positive bias  (because, for
example, the additional preparation artificially inflates the
number of  fibers) or the direct measurements incorporate a
negative bias  (because, for example, fibers are covered by
debris).   Fiber size data should be available  for Studies 2, 3,
and 4, and could be analyzed to distinguish between competing
hypotheses.  The number of structures counted, particularly those
in the larger size categories, could limit the investigation.  A
designed experiment in which samples were prepared according to
carefully  specified protocols would provide more conclusive
information.  Experimental factors include preparation method,
filter loading (low to high), and prevalence of complex
structures.

B. Precision

Other considerations being equal, the method with the highest
precision  is preferable.  For TEM analysis of  airborne asbestos,
the spatial distribution of asbestos structures on the surface of
a filter is important in determining precision.  Only a tiny
fraction of the original filter area is examined with the
electron microscope.  It is assumed that the area is

                                34

-------
representative of the entire filter surface in order to estimate
the concentration of asbestos in the sampled air.   (Other aspects
of the protocol including counting rules,  filter loading, and
area of filter examined also affect precision.  These are not
discussed further here because they can be varied independently
of the transfer method.  The effect of procedures such as ashing
and resuspension that are uniquely associated with indirect
transfer method would be included in any overall study of
precision.)

Chatfield (1984, 1986) has argued that the spatial distribution
of asbestos structures on the filter is closer to random (i.e.,
follows a Poisson distribution)  when an indirect transfer method
is used.  If structure counts per grid opening are available for
Studies 1, 2, 3 and 4, Chatfield's claim can be tested.  Efforts
are underway to obtain these data.  The question may also be
addressed experimentally by preparing samples by both techniques
and examining the filter in greater detail than is done during
routine analysis.  A relatively simple statistical design and
analysis would be sufficient to detect marked differences and
could provide a definite recommendation.  A more sophisticated
experiment is needed to explore heterogeneity on various spatial
scales in order to determine the advisability of preparing more
than one portion of the filter or analyzing multiple grids.

Since breakup of structures (resulting in a positive bias)  and
uneven spatial distribution of structures on the filter
(resulting in decreased precision) are claimed to be the major
disadvantages of the indirect and direct transfer methods
respectively, further research to support or reject these claims
would be a valuable and relatively low cost contribution to the
continuing discussion over the choice of analytical protocol.
                                35

-------
                            REFERENCES
Anderson KL, Theys RO, Dunmyre GR. 1989.  Sources of
contamination using indirect sample preparation techniques.
National Asbestos Council Journal, Summer 1989: 27-30.

Bishop TA, Collier RP, Kurth RE.  1981.  Statistical analysis of
ECC bypass data using a nonlinear constrained maximum likelihood
technique.  Nuclear Engineering and Design, 64: 87-91.

Britt HI, Lueke RH.  1973.  The estimation of parameters in
nonlinear, implicit models.  Technometrics, 15: 233-283.

Burdett GJ.  1985a.  Inter-laboratory comparison of the U.S.
Environmental Protection Agency school samples by the UK Health
and Safety Executive.  UK Health and Safety Executive.  Report
No. IR/L/DI/86/03.

Burdett GJ.  1985b.  The measurement of airborne asbestos
releases from damaged amosite insulation subjected to physical
attrition.  In Asbestos Fibre Measurements in Building
Atmospheres, Proceedings.  Chatfield EJ, editor.  Mississauga,
Ontario:  Ontario Research Foundation.

Burdett GJ, Rood AP.  1983.  Membrane filter, direct-transfer
technique for the analysis of asbestos or other inorganic
particles by transmission electron microscopy.  Environmental
Science and Technology, 17: 643-648.

Chatfield EJ.  1984.  Chatfield Technical Consulting Limited.
Measurement and Interpretation of Asbestos Fibre Concentrations
in Ambient Air.  Pre-publication copy of paper presented at 5th
AIA Colloquium in Johannesburg, October 29-31, 1984.

Chatfield EJ.  1985.  Airborne asbestos levels in Canadian public
buildings.  In Asbestos Fibre Measurements in Building
Atmospheres, Proceedings.  Chatfield EJ, editor.  Mississauga,
Ontario:  Ontario Research Foundation.

Chatfield EJ.  1986.  Asbestos measurements in workplaces and
ambient atmospheres.  In Electron Microscopy in Forensic,
Occupational and Environmental Health Sciences.  Basu S and
Millette J, editors.  New York, NY:  Plenum Publishing
Corporation.

Cook PM, Marklund DR. 1982.  Sample preparation for quantitative
electron microscope analysis of asbestos fiber concentrations in
air.  National Bureau of Standards Special Publication 619: 53-
67.
                                37

-------
Hatfield J, Leczynski B, Chesson J et al.   1987.  Battelle
Columbus Division.  Public buildings study quality assurance
plan.  Final report.  Washington, DC: Office of Toxic Substances,
U.S. Environmental Protection Agency.  Contract No. 68-02-4243.

Lee RJ.  1987.  The Constant study revisited:  A comparison of
the airborne asbestos fiber concentrations in schools as
determined by direct and indirect sample preparation techniques.
Monroeville, PA: Energy Technology Consultants.

Lehmann EL. 1975.  Nonparametrics.  San Francisco: Holden-Day,
Inc.

NIOSH. 1985.  National Institute for Occupational Safety and
Health.  Manual of Analytical Methods, 3rd edition, first
supplement.  Edited by Eller PM.  Cincinnati, OH:  Division of
Physical Sciences and Engineering, National Institute for
Occupational Safety and Health.

Ortiz LW, Isom BL. 1974.  Transfer technique for electron
microscopy of membrane filter samples.  American Industrial
Hygiene Association Journal, 35:423.

Peters ET, Doerfler TE. 1978.  Amphibole mineral fiber analysis
by electron microscopy: comparison of sample preparation
procedures.  In: Electron Microscopy and X-ray Applications of
Environmental and Occupational Health Analysis, Russell PA,
Hutchings AE, editors.  Ann Arbor Science Publishers, Inc. 189-
203.

Sebastien P, Plourde M, Robb R, Ross M.  1984.  Ambient air
asbestos survey in Quebec mining towns, Part 1 — Methodological
study.  Montreal, Quebec: Canadian Environmental Protection
Service.  EPS 3/AP/RQ/1E.

Steen D, Guillemin MP, Buffat P, Litzistorf G.  1983.
Determination of asbestos fibres in air: Transmission electron
microscopy as a reference method.  Atmospheric Environment 17:
2285-2297.

Tuckfield RC, Tsay Y-L, Margeson DP et al.  1988.  Battelle
Columbus Division.  Evaluation of asbestos abatement techniques
phase III:  removal.  Draft final report.  Washington, DC: Office
of Toxic Substances, U.S. Environmental Protection Agency.
Contract No. 68-02-4243.

USEPA 1975.  U.S. Environmental Protection Agency.  Asbestos
contamination of the air in public buildings.  Research Triangle
Park, NC, U.S. Environmental Protection Agency.  EPA 450/3-76-
004.
                                38

-------
USEPA 1978a.  U.S. Environmental Protection Agency.  Evaluating
and optimizing electron microscope methods for characterizing
airborne asbestos.  Research Triangle Park, NC: Office of
Research and Development, U.S. Environmental Protection Agency.
EPA 600/2-78-038.

USEPA 1978b.  U.S. Environmental Protection Agency.  Preparation
of water samples for asbestos fiber counting by electron
microscopy.  Athens, GA: Office of Research and Development, U.S.
Environmental Protection Agency.  EPA 600/4-78-011.

USEPA 1979.  U.S. Environmental Protection Agency.  Sprayed
asbestos-containing materials in buildings:  A guidance document,
Part 2.  Washington, DC, U.S. Environmental Protection Agency.
EPA 450/2-78-014.

USEPA.  1980.  U.S. Environmental Protection Agency.  Measurement
of asbestos air pollution inside buildings sprayed with asbestos.
Washington, DC: Office of Toxic Substances, U.S. Environmental
Protection Agency.  EPA 560/13-80-026.

USEPA. 1983. U.S. Environmental Protection Agency.  Airborne
asbestos levels in schools. Washington, DC: Office of Toxic
Substances, USEPA. EPA 560/5-83-003.

USEPA. 1985. U.S. Environmental Protection Agency. Evaluation of
asbestos abatement techniques, phase 1: removal. Washington, DC:
Office of Toxic Substances, USEPA. EPA 560/5-85-019.

USEPA. 1986a.  U.S. Environmental Protection Agency.  Evaluation
of asbestos abatement techniques phase II: encapsulation with
latex paint.  Washington, DC: Office of Toxic Substances, U.S.
Environmental Protection Agency.  EPA 560/5-86-016.

USEPA. 1986b. U.S. Environmental Protection Agency.  Filter blank
contamination in asbestos abatement monitoring procedures:
proceedings of a peer review workshop.  Cincinnati, OH: Office of
Research and Development, U.S. Environmental Protection Agency.

USEPA 1988.  U.S. Environmental Protection Agency.  Assessing
asbestos exposure in public buildings.  Washington, DC: Office of
Toxic Substances, U.S. Environmental Protection Agency.  EPA
560/5-88-002.

Yamate G, Agarwal SC, Gibbons RD.  1984.  Methodology for the
measurement of airborne asbestos by electron microscopy.  Draft
report.  Washington, DC: Office of Toxic Substances, U.S.
Environmental Protection Agency.  Contract No. 68-02-3266.
                                39

-------
40

-------
Appendix A: Definition of Asbestos Structure Types

-------
Fiber:    Structure with an aspect ratio (length to width) of 3:1
          or greater with substantially parallel sides.

Bundle:   Structure composed of fibers in a parallel arrangement
          with each fiber closer than one fiber diameter.

Cluster:  Structure with fibers in a random arrangement such that
          all fibers are intermixed and no single fiber is
          isolated from the group or groups of fibers closely
          spaced and randomly oriented.

Matrix:   Fiber or fibers with one end free and the other end
          embedded or hidden by a particulate.  Combinations such
          as a matrix and cluster, matrix and a bundle, or bundle
          and a cluster are categorized by the dominant fiber
          quality-cluster, bundle, and matrix, respectively.
                                43

-------
Appendix B: Data Listings

-------
CO

00

en
04

w
t
4-»
W

 e
 o
 n
-p
 (0

 0)
               a

               *i
                £


               i
                a
                         a
b«ra
                 ,   i   =
CMAo»CMtOrHO»aia)dft«,-H*rcMr'*r-


—*»co-^irtCMco«»rowi'^ri-<»r^LO'—«.—t*rr—i-HLO^CM«o^-^-rOLO
U?         CO   (O  '~t
^H            CM  O<                    ^                ^H           p^   fO      U3  CO           r*4   C*t           C^





s«^o»««ts>isa>tB«(Ssq)«)a)s»9)a)«Snv«»««^


««««««s«9>m«»«a»a»(BSAmA«««»*v«ai




















                                                                                              B  «  to   ^H  ,-*   «  m











CD  •  co   ^-r-».«»o>CM^-r*-r^.CMm»««»»ma>a»a>a*«i—«coro«—'r-Hro«^mcoco>-o»




                                                                                                       t-H








««is«)«v«««»s«>tacx)«>«cBtsim««v«i«i»9««











cmVICDWAVIWVVIVIt&cOcocDCOCMCMtOcOcOCMcDCOr-tDcQ^CD
                                                  ^^*-»^H^^fOf5^-t«-H«-^fOi-^^-<^-i^>^-tO'^'
                                                                                                       .—(
























•^^^f^U>^StOCOU*i^5'*l^3fQ^^r*Ci/>^'^K>^'^fO^M
3

 5
 i
CQ
                     a.  a


                     

 0)
 (0

 EH
                                     47

-------
Table B-2 .   Data Listing of Results from Studies 1 Through 5 (See
         Text for an Explanation of Units of Concentration.)
Study
l
l
1
1
1
l
l
l
l
l
l
1
1
l
l
l
l
l
1
1
l
l
l
l
1
l
l
l
2
2
2
2
2
2
2
2
2
2
Direct Transfer
Sample ID s/cc
7159
7206
7275
7518
7542
7622
7958
7962
7964
8006
8110
6796
6823
6853
6855
7539
7584
7637
7859
7874
7905
8098
8216
6780
6868
7212
7258
7588
CD13
CD21
CD10
FB4
J23
J55
J58
J60
J62
85-451
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0030
0.0030
0.0040
0.0030
0.0090
0.0070
0.0030
0.0030
0.0020
0.0060
0.0030
0.0020
0.0330
0.0100
0.0130
0.0130
0.0080
0.0010
0.0000
0.0040
0.9980
0.0000
0.0020
0.0072
0.0000
0.0010
0.0010
Indirect Transfer
Sample ID f/cc
7159
7206
7275
7518
7542
7622
7958
7962
7964
8006
8110
6796
6823
6853
6855
7539
7584
7637
7859
7874
7905
8098
8216
6780
6868
7212
7258
7588
CD14
CD22
CD9
FB3
J24A
J56
J57
J59
J61
KE22
0.009
0.000
0.025
0.051 -_.
0.034
0.072
0.000
0.005
0.010
0.087
0.000
0.030
0.000
0.030
0.028
0.017
0.061
0.187
0.019
0.065
0.042
0.000
0.016
0.060
0.108
0.125
0.017
0.055
3.740
5.850
36.950
21.600
14.350
8.730
13.000
3.350
2.060
13. 100
                                 48

-------
Table B-2.  (continued) Data Listing of Results from Studies 1 Through  5
Study
2
2
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
Direct Transfer
Sample ID s/cc
85-484
85-450
85-445
85-435
85-389
85-308
85-324
85-319
85-307
85-82
85-87
85-361
73
72
150
153
52
53
49
59
60
61
66
68
118
117
80
76
137
138
140
96
101
111
135
128
134
0.0010
0.5930
0.0000
0.0000
0.0000
0.0841
3.7000
0.0000
0.0000
0.0460
0.0460
0.0020
0.0040
0.0630
0.0000
0.0820
0.0100
0.0000
0.0000
0.0000
0.0050
0.0000
0.0000
0.0080
0.0000
0.0000
0.0040
0.0840
0.0000
0.0000
0.0000
0.0050
0.0270
0.0070
0.0060
0.0110
0.0020
Indirect Transfer
Sample ID f/cc
KE23
KE24
KE36
KE37
X005-3-3
081-10M
081-11
081-12
081-4-3
102-11
102-13
102-4-3
1
4
19
20
23
24
26
30
31
32
33
35
41
42
43
45
51
52
53
57
58
59
61
62
63
7.075
31.900
1.595
5.240
0.096
5.890
2.200
26.500
0.010
7.060
16.000
23.000
123.173
96.908
21.729
29.487
12.548
0.228
0.019
12.282
25.163
0.170
1.479
45.464
0.365
0.007
0.179
100.916
0. 063
0.007
0.000
6.294
5.487
1.913
0.135
11.340
6.787
                                    49

-------
Table B-2.  (continued)  Data Listing of Results from Studies 1 Through 5
                 Direct Transfer
Indirect Transfer
Study
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
Sample ID
133
125
126
63
55
70
71
88
85
81
92
91
162
43
45
46
40
93
106
157
145
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
s/cc
0.0070
0.0000
0.0000
0.0040
0.0950
0.0000
0.0040
0.0050
0.1220
0.0020
0.0120
0.0000
0.0040
0.0000
0.0000
0.0000
0.0130
23043.8800
0.0220
0.0050
0.0000
0.0066
0.0310
0.0000
0.0071
0.0000
0.0012
0.0000
0.0000
0.0084
0.1935
0.1171
0.0036
0.0000
0.0111
0.0287
Sample ID
64
65
66
75
78
81
82
85
89
94
96
97
103
112
114
115
117
119
122
124
126
9
27
36
74
84
86
87
88
93
95
105
115
117
119
122
f/cc
18.527
9.669
0.092
0.161
44.918
3.872
4.724
54.176
100.358
0.100
18.319
0.146
36.529
4.347
1.564
0.217
5.953
10.548
24.763
1.018
0.000
27.700
79.700
0.200
11.130
2.319
0.637
0.000
0.314
0.113
42.843
63.520
0.216
5.953
10.548
24.762
                                    50

-------
Table B-2.  (continued)  Data Listing of Results from Studies 1 Through 5
                 Direct Transfer
Indirect Transfer
Study
5
5
5
5
5
5
5
5
6
6
6
6
6
6
6
6
7
7
7
7
7
7
7
7
7
7
7
7
Sample ID
LA39
LA47
SH43
SH51
LA23
LA41
LA49
QP45
72
76
128
118
66
68
40
45
7144A
7144B
7144C
9040
9041
9042
9061
9062
9063
4221
4222
4223
s/cc
0.1400
0.0640
0.0720
0.1100
0.0160
0.0130
0.0180
0.0250
0.0630
0.0840
0.0110
0.0000
0.0000
0.0080
0.0130
0.0000
0.0990
0.1100
0.0910
0.1000
0.1600
0.2910
0.0740
0.2150
0.0200
0.0500
0.0700
0.0840
Sample ID
LA39
LA47
SH43
SH51
LA23
LA41
LA49
QP45
72
76
128
•118
66
68
40
45
7144A
7144B
7144C
9040
9041
9042
9061
9062
9063
4221
4222
4223
f/cc
2.600
0.230
1.300
3.000
0.110
0.170
0.039
0.120
0.047
7.723
0.702
0.323
0.421
0.667
0.120
0.037
0.262
0.235
0.178
0.513
0.448
0.516
0.033
0.071
0.076
0.158
0.099
0.230
                                    51

-------
Appendix C: Correlation Coefficients for Study l  (USEPA 1988)

-------
Table C-l.      Measures of Association Between Direct and
               Indirect TEM Measurements Obtained in Study 1 (The
               Asymptotic Standard Error is Given in
               Parentheses.)
Comparison
Indirect
Indirect
Indirect
Indirect
Indirect
Indirect
Indirect
Indirect
Indirect
Indirect
s/cc vs.
f/cc vs.
f/cc vs.
f/cc vs.
b/cc vs.
b/cc vs.
b/cc vs.
s/mra2 vs.
f /mm2 vs .
f/mm2 vs.
direct
direct
direct
direct
direct
direct
direct
direct
direct
direct
Pearson
Correlation
s/cc
s/cc
f/cc
b/cc
s/cc
f/cc
b/cc
s/mm2
s/mm2
f/mm2
0
(0
0
(0
0
(0
0
(0
0
(0
0
(0
0
(0
0.
(0
0.
(0
0.
(0
.21
.11)
.26
.12)
.16
.13)
.37
.12)
.08
.11)
.06
.11)
.07
.14)
15
.10)
20
.10)
12
.12)
Spearman
Correlation
0
(0
0
(0
0
(0
0
(0
0
(0
0
(0
0
(0
0.
(0
0.
(0
0.
(0
.35
.16)
.39
.16)
.31
.17)
.34
.17)
.31
.16)
.29
.16)
.11
.18)
29
.16)
34
.17)
28
.17)
                                55

-------
Table C-l.(continued)    Measures of Association Between Direct
                         and Indirect TEM Measurements Obtained
                         in Study 1  (The Asymptotic Standard
                         Error is Given in Parentheses.)
                                Pearson     Spearman
       Comparison             Correlation  Correlation
Indirect f/mm2 vs.  direct b/mm2   0.28          0.27
                                 (0.11)        (0.19)

Indirect b/mm2 vs.  direct s/mm2   0.04          0.23
                                 (0.10)        (0.16)

Indirect b/mm2 vs.  direct f/mm2   0.04          0.26
                                 (0.10)        (0.17)

Indirect b/mm2 vs.  direct b/mm2   0.02          0.00
                                 (0.14)        (0.19)
s   = Structure Concentration
f   = Fiber Concentration
b   = BCM Concentration
cc  = Cubic Centimeter of air
mm2 = Square Millimeter of Filter
                                56

-------
50272-101
 REPORT  DOCUMENTATION
          PAGE
                             EPA 560/5-89-004
                                                                                      3. Recipient's Accession No.
4. Title and Subtitle
    Comparison of Airborne Asbestos Levels Determined by Transmission Electron
    Microscopy (TEM) Using Direct and Indirect Transfer Techniques
                                                                                      5. Report Date
-March, 1990
7. Author(s)

	Chesson J, Hatfield J.
8. Performing Organization Rept. No.
9. Performing Organization Name and Address

    Chesson Consulting, Inc., 1717 Massachusetts Ave, NW, Washington, DC 20036

    Battelle, Arlington Office, 2101 Wilson Boulevard, Arlington, VA 22201
10. Project/Task/Work Unit No.
11. Contract(C) or Grant(G) No.

(c> 68-02-4294
(G)
 12. Sponsoring Organization Name and Address
    U.S. Environmental Protection Agency
    Office of Toxic Substances
    Exposure Evaluation Division (TS-798)
    401 M Street, SW, Washin£tnn, DC 20460
 15. Supplementary Notes
13. Type of Report & Period Covered
Peer-reviewed report
u.
16. Abstract (Limit: 200 words)
    A subset of air samples from a 1988 EPA study were reanalyzed for asbestos by TEM using an indirect transfer
    technique. The samples were originally analyzed using a direct  transfer technique.  This document presents the
    results of the reanalysis and extends the discussion to include data from six other studies.  The development of
    the two  techniques and their respective advantages and disadvantages are described.

    The data support the general opinion that TEM analysis of air samples using indirect transfer methods tends  to
    provide  estimates of total airborne asbestos structure concentration  that are higher than those obtained using
    direct transfer methods.  There is no single factor that can be used to convert measurements made by one
    method  to a value that is comparable with measurements made by the other because the quantitative relationship
    is expected to depend on details of the sampling and analytical protocols and the nature of the asbestos in the
    air.  The ratio of indirect measurements to direct measurements ranges from 3.8 to 1,700 for the studies
    considered.

    Additional research is needed to determine which transfer technique more accurately reflects biologically
    meaningful airborne asbestos concentrations.  Breakdown of larger structures into smaller ones during indirect
    preparation does not appear to be sufficient to explain the difference in measured concentrations.  Alternative
    hypotheses involving interference by debris and association of unattached structures may also be important.
17. Document Analysts  a.  Descriptors
   b. Identifiers/Open-ended Terms
   c. COSATI Field/Group
18. Availability Statement
19. Security Class (This Report)
Unclassified
20. Security Class (This Page)
Unclassified
21. No. of Pages
68
22. Price
 ee ANSI-Z39.18)
                                                 See Instructions on Reverse
             OPTIONAL FORM 272 (4-77)
             (Formerly NTIS-35)
             Department of Commerce

-------