Air Quality Criteria for Lead Volumes I ,II,III,IV


                                                      EPA-600/8-83-028A

                  UNITED STATES ENVIRONMENTAL PROTECTION AGENCY
                       Environmental Criteria and Assessment Office (MD-52)
                           Research Triangle Park, North Carolina 27711

   DATE:  November 30, 1983

SUBJECT:  Corrigenda for the First External Review Draft of 1983 Revised
          EPA Criteria Document, Air Quality Criteria for Lead

   FROM:  Lester D.  Grant, Director
          ECAO/RTP,  U.S.  EPA (MD-52)

     TO:  Recipients of the subject first external review draft of the
          1983 revised EPA Lead Criteria Document

     Copies of the first External Review Draft for the EPA Document Air
Quality Criteria for Lead were recently made available for a ninety-day
public comment period (October 15, 1983 - January 15, 1983), as announced in
the Federal Register.

     The External Review Draft of the Lead Document circulated to you and
other recipients did not contain Appendix 12-C, a detailed report of the
Expert Committee on Pediatric Neurobehavioral  Evaluations (which was con-
vened by ECAO/RTP to provide in-depth evaluations of studies of associations
between low-level lead exposure and neuropsychologic deficits in children
reported by Drs. Ernhart, Needleman, and their respective colleagues).
Copies of Appendix 12-C were withheld pending reconvening of the "Neuro-
behavioral Evaluations" Committee in order to consider comments by Drs.
Ernhart and Needleman on a preliminary draft of the Committee's report and
in order for the Committee to take into account newly available published
and unpublished information pertaining to the subject studies in carrying
out final revision of their report.  A copy of the Committee's final report
is enclosed for insertion as Appendix 12-C following other Chapter 12
materials in the External Review Draft of the Lead Document recently provided
to you.

     Another committee was convened by ECAO/RTP in late September to evaluate
certain German studies (by Drs. Kirchgessner and Reichlmayr-Lais) reporting
evidence interpreted by those investigators as being indicative of beneficial
effects of lead at very low exposure levels.  The "Essentiality" Committee
has recently completed their report evaluating the subject studies, and the
report is enclosed herein. That report constitutes Appendix 12-A and is to
replace the existing critique of the subject studies, which appeared as
Appendix 12-A in the recently circulated External Review Draft of the Lead
Document.

-------
     In addition to the above appendices, a corrigenda is also enclosed
which lists corrections/notations for text and tables in the circulated four
volumes of the subject draft Lead Criteria Document.  The corrections noted
are restricted to those thought to be crucial for accuracy or understanding
of the information presented and to indicate changes apropos to the insertion
of the above new appendices.  Minor typographical errors or editorial changes
are generally not included.   There are no changes indicated to be made for
Chapters 2-6 or 8 for the document.

     We apologize for the unfortunate delay in our being able to circulate
the above Appendix materials (for both Appendix 12-A and 12-C) to you.   We
also recognize the importance of their contents in terms of their crucial
utility in helping to resolve certain key issues of much relevance for the
development of criteria and standards for lead.  In view of our delay in
circulating these important materials, we are currently processing a Federal
Register notice announcing a one-month extension of the Public Comment
Period for that draft document to February 15, 1984. We do not anticipate
any problems in having that extension and its announcement approved in time
for publication in the Federal Register during the first or second week of
December, 1983.  We hope that this information will assist in your planning
of work efforts connected with preparation of public comments.

-------
                                                                 11/30/83
                     First External Review Draft for EPA
                  Lead Criteria Document (1983):  Corrigenda

                                   Chapter 1
Page

1-35



1-36

1-37

1-42


1-48


1-50

1-52




1-57

1-65

1-66

1-69

1-98

1-116


1-141



1-52
Line

Table 1-4
footnote.
6-7 up*

9 up

13 up


2 up


1

23-24




7 up

1 up

3,7,14 up

20

18

2 up
1 up

Table 1-20



12 up
     Correction/Notation

References cited in Table 1-4, but not provided in
Section 1.14 (reference list for Chapter 1), are availa-
ble in the cited Nriagu (1978b) paper.

Delete "..., due complex" from end of sentence.

Replace "...mg/g..." with "...ug/g..."

Change to read:  "...lead contribute differently to
each of these dietary groups (Figure 1-1)."

Change to read:  "...1971 Annual Report of the United
Kingdom..."

Replace "...Association..." with "...Organization..."

Change to read:  "...about 100 ug of lead is consumed
daily by each American.   For all Americans, this
amounts to only 8 tons/year, or 0.001-0.01 percent
of the total environmental contamination."

Change "...(Nriagu, 1978)" to read:   "...(Nriagu,  1978a)"

Change "...(Nriagu, 1978)" to read:   "...(Nriagu,  1978a)11

Change "mg" to "ug"

Change "primary" to "non-circumpulpal"

Change "1000 ug/dl" to "1000 ug/g"

Change "human" to "pediatric"
Change "human studies" to "studies of children"

Downward arrow should be inserted below "Vitamin D
metabolism interference" entry, indicating that the
vitamin D effect occurs down to 10-15 |ug/dl blood lead.

"maxim safe level"  should read "maximum safe level"
*Number of lines up from bottom of page (other entries are for number of lines
 from top of page).

-------
                                                                  11/30/83
                     Lead Document Corrigenda  (continued)

                                   Chapter 7


Page      Line                Correcti on/Notati on
                                  3     3
7-63      4              Change mm  to m
                                   3        3
7-64      21             Change u/m  to ug/m

7-69      1-2 up         Change last sentence  of page to read:  "For all
                         Americans, this amounts to only 8 tons/year or 0.001-
                         0.01 percent of the total environmental contamination."


                                   Chapter 9

Page      Line                Correcti on/Notati on

9-12      8              Change "secondary (circumpulpal)" to "circumpulpal"


                                  Chapter 10

Page      Line                Correction/Notation

10-5      13 up          Change "often" to "after"

10-15     12             Line should read:  "difficult to measure, and reliable
                         values have become available only recently (see Chapter 9)."

10-20     19             Add "Shapiro et al.,  1978" after "Winneke et al. 1981"

10-42     3-7 up         Sentence should read:   "In a related study (Grant et al.,
                         1980), rats were exposed to lead jjn utero, through
                         weaning, and up to 9  months of age at the dosing range
                         used in the Kimmel et al. study (0.5 to 250 ppm in the
                         dams'  drinking water  until weaning of pups; then the
                         same levels in the weanlings'  drinking water).  These
                         animals showed a blood lead range of 5 to 67 ug/dl."

10-45          23        Line should read:  "remains the one readily accessible
                         measure that can demonstrate in a relative way the rela-
                         tionship of various effects to increase in exposure."


                                  Chapter 11

Page      Line                Correcti on/Notati on

11-104    17             Change "statistical  relationship" to "significant
                         relationship"

11-110    2 and 6 up     Change "per mg/g" to  "per 1000 ug/g"

                                     2

-------
                                                                  11/30/83
Page

11-142

11D-23
                  Chapter 11  (continued)

                    Correction/Notatjon

               Numerical table entries are in |jg/dl units.

               Insert revised Appendix 03 (attached single sheet)
               in place of comparable page in Appendix 11-D.
Page

12-1
Line
19-23
12-48



12-60


12-62
11
11-12 up


4


18-on
12-64

12-65
5

7-on
         Chapter 12

     Correction/Notation

Replace  last two sentences of second paragraph with:
"An evaluation of these studies by an expert committee
convened by EPA in September, 1983, is contained in
Appendix 12-A.  The committee's report notes metho-
dological problems with the studies, which preclude
acceptance of the reported findings as demonstrating
the essentiality of lead.  These studies are, therefore,
not considered further in the present document.

Change "(10 to 20)" to "(10 of 20)."
Change "Reports of low blood levels..." to read
"Reports of effects at low blood levels..."

Change "...history of pica, as well..." to read
"...history of pica for paint and plaster, as well as..."

Replace the entire last paragraph of page 12-62 with:
"The Peri no and Ernhart (1974) and Ernhart et al. (1981)
studies were evaluated by an expert committee convened
by EPA in March, 1983.  The committee's report (see
Appendix 12-C) notes methodological problems which
preclude acceptance of the analyses and findings pub-
lished by Peri no and Ernhart (1974) and Ernhart et al.
(1981).  The committee's report, further, recommends
that the Ernhart data set be reanalyzed, including
longitudinal analyses of data for subjects evaluated
in both the Peri no and Ernhart (1974) and Ernhart et al.
(1981) studies.   Pending resolution of methodological
problems with the Ernhart data set and/or publication of
adequate reanalyses, the subject studies are not con-
sidered further in this document."
Change "primary"
to "non-circumpulpal"
After the first sentence ending with "...(see Appendix
12-C)," replace the rest of the paragraph with the
following:  "The committee's report notes methodological
problems which preclude acceptance of the published
analyses and findings reported either (1) by Needleman
et al.  (1979) or (2) in subsequent papers by Needleman

-------
                                                                  11/30/83
Page

12-65
Line

7-on
12-67     16 up

12-69     1 up



12-83     Table 12-3




12-84     Table 12-3
12-87     Table 12-3
12-88     Table 12-3

12-90     Table 12-4
12-91     5
          9

12-101    Table 12-5
12-115    3
          5

12-125    5


12-216    1
   Chapter 12  (continued)

     Correct1on/Notat1on

and coworkers  concerning additional analyses of the
same data set.  The committee's report also recommends
that the Needleman data set be reanalyzed.  Pending
resolution of  methodological problems with the
Needleman data set and/or  publication of  adequate re-
analyses, the  subject studies are not considered further
in this document."

Change "-0.06" to "+0.06"

Add the following sentence:  "The Landrigan et al.
(1975) and McNeil and Ptasnick (1975) studies are,
therefore, not considered  further in this document."

For Overmann (1977), Pb concentration should read:  "5,
15, or 45 mg/kg."
For Winneke et al. (1977), Pb concentration should read:
"745 mg/kg (diet)."

For Dietz et al. (1978), Pb concentration should read:
"0.025%."
For Cory-Schlecta and Thompson (1979), Pb concentration
should read:    "(1)0.0025,  (2)0.015, or (3)0.05%."
For Cory-Schelacta et al.  (1981), Pb concentration should
read:  "(1)0.005 or (2)0.015%."

For Milar et al. (1981), Pb concentration should read:
"mg/kg b.w.  (gavage)."
For Nation et  al. (1982),  Pb concentration should read:
"mg/kg b.w."
For Winneke et al. (1982), Pb concentration should read:
"0.08, 0.025,  or 0.075%."

Under Abbreviations, add:  "b.w.  body weight"

For Rice and Willes (1979), Pb concentration should read:
"Mg/kg b.w."  Under abbreviations, add:   "b.w.  b«dy weight"

Change "or to  "of"
Change "change level" to "chance level"

Under exposure protocol, the 3rd entry should end with
"PND 20" rather than "PNDO"

Change "human" to "pediatric"
Change "human  studies"  to "studies of children"

The 4th sentence of paragraph should read:  "Proteinuria
occurred in two patients."

Change "as much as" to "inasmuch as"

            4

-------
                                                                 11/30/83
Page

12-229


12A-1



12C-1
Line

7
8
    Chapter 12 (continued)

     Correcti on/Notation

Change "human" to "pediatric"
Change "human studies" to "studies of children"
Appendix 12-A  Replace Appendix 12-A (pp.  12A-1 to 12A-7) with enclosed
               copy of "Essentiality" Committee's report labeled as
               Appendix 12-A.

Appendix 12-C  Insert enclosed copy of "Neurobehavioral Evaluations"
               Committee's report labeled as Appendix 12-C.
Page
13-2
13-22
13-24
13-25
13-26
13-32
Line
4 up
5
Table
Table
Table
Table



13-6
13-7
13-8
13-10
13-44
12
         Chapter 13

     Correction/Notation

Delete "which" from line

Change "...about 2, may represent..." to "...about 2 ug/dl
per 1000 ug/g may represent..."

Entries for PbB values are in  |jg/dl  units

Entries for PbB values are in  ug/dl  units

Entries for PbB values are in  ug/dl  units

Under first column, "10" should be "10 ug/dl."
Under last column,  a downward  arrow should be added
immediately below the "Vitamin D metabolism inter-
ference" entry, to indicate that the vitamin D effect
occurs down to 10-15 ug/dl blood lead.

In line 7 of the second conclusion,  change "...lead
contribution can be..." to read "...lead contribution
to human blood lead levels can be..."
          1-2 up
               Change "maxim safe level"  to "maximum safe level"

-------
                                   Appendix D3
             List of Attendees at March 10-11 and March 30-31,  1983
                                   meeting of
                   NHANES II TIME TREND ANALYSIS REVIEW GROUP
                               Committee Members
Joan  Rosenblatt (Chairman)
National  Bureau of Standards

J.  Richard Landi's
University of Michigan

Roderick  Little
Bureau  of the Census
Joel Schwartz
U.S. EPA

J. Lee Annest
NCHS

Jean Roberts*
NCHS

James Pirkle
CDC

Vernon Houkf
CDC

EPA Staff
 Richard  Royal!
 John  Hopkins  University

 Harry Smith,  Jr.
 Mt. Sinai  School  of Medicine
                               Invited Discussants
Ben Forte
Ethyl Corporation

Chuck Pfieffer
DuPont

Ron Snee
DuPont

Asa Janney
ICF
Observers
David Weil (Meeting Co-ordinator)
U.S. EPA

Dennis Kotchmar*
U.S. EPA

Vic Hasselblad
U.S. EPA

Allen Marcus
U.S. EPA
xattended March 10-11 meeting  only.
fattended March 30-31 meeting  only.
Earl Bryant*
NCHS

Trena Ezzote*
NCHS

Mary Kovar*
NCHS

Bob Casadyx
NCHS

Robert Murphy
NCHS

Jack Pierrardx
DuPont

Kathryn Mahaffeyx
FDA
                                                              4 U.S. GOVERNMENT PRINTING OFFICE: 1983-459-017/7239

-------
                                           EPA-600/8-83-028A
       INDEPENDENT PEER REVIEW OF SELECTED STUDIES
        BY DRS.  KIRCHGESSNER AND REICHLMAYR-LAIS
CONCERNING THE POSSIBLE NUTRITIONAL ESSENTIALITY OF LEAD:

  Official Report of Findings and Recommendations of an
        Interdisciplinary Expert Review Committee
                      Presented by


      Expert Committee on Trace Metal Essentiality



                           to
              Dr.  Lester D.  Grant, Director
      Environmental Criteria and Assessment Office
      United States Environmental Protection Agency
         Research Triangle Park, North Carolina
                     November, 1983

-------
     The materials  contained in  this report were generated  as  the result of
critical evaluations  and deliberations  by the members  (listed  below) of the
Expert  Committee  on Trace  Metal  Essentiality.  All  members concur  with and
endorse the  findings and  recommendations  contained  in  the  present report as
representing the collective sense of the Committee.
Dr. F. William Sunderman, Jr. (Chairman)
Professor, Departments of Laboratory
  Medicine and Pharmacology
University of Connecticut School of
  Medicine
Farmington, CT   06232

Dr. M. R.  Spivey Fox
Chief, Nutrient Interaction Section
Division of Nutrition
U.S. Food and Drug Administration
Washington, DC   20204

Dr. Katnryn Mahaffey
Chief, Priorities and Research Analysis
National Institute of Occupational
  Safety and Health
Cincinnati, OH   45226

Dr. Forrest Nielsen
Research Chemist
Human Nutrition Research Center
U.S. Department of Agriculture
Box 7166 University Station
Grand Forks, ND   58202
Dr. Orville Levander
Research Chemist
Beltsville Human Nutrition
  Research Center
U.S. Department of Agriculture
Beltsville, MD  20705

Or. Walter Mertz
Director, Beltsville Human Nutrition
  Research Center
U.S. Department of Agriculture
Beltsville, MD  20705

Dr. Eknard Ziegler
Professor, Department of Pediatrics
University of Iowa Hospital
Iowa City, IA  52242
                                       ii

-------
                               TABLE OF CONTENTS







                                                                         PAGE





PREFACE	     1



INTRODUCTION	     2



CRITICAL COMMENTS 	     2



RECOMMENDATIONS AND CONCLUSIONS 	     3



ATTACHMENT 1	     4



ATTACHMENT 2 	     7

-------
                                    PREFACE
     The Expert Committee on Trace Metal Essentiality was appointed in August,
1983 by  the Environmental  Criteria and Assessment Office  (ECAO)  of EPA, to
evaluate the studies  of Drs.  M. Kirchgessner and A. M. Reichlroayr-Lais on the
possible nutritional  essentiality of  lead.   The  Committee  was provided with
all relevant papers  by the authors, the critiques of these papers prepared by
Or. Paul  Mushak in  his capacity as  a consulting  author  of the  revised Air
Quality Criteria  Document for  Lead,  and all correspondence between  ECAO and
Drs. Kirchgessner and  Reichlmayr-Lais.   Attachment 1 contains a complete list
of the materials reviewed by the Committee.

     The Committee  convened on  September 29,  1983  at  the  ECAO facilities in
Research Triangle Park, NC.  Present at the meeting were all but one member of
the Committee (K.M.), Drs. Anna Reichlmayr-Lais and E. Grassmann (substituting
for Professor Kirchgessner), Dr. Mushak, EPA staff, and observers from various
interest groups.  A  complete  list of attendees may be found in Attachment 2.

     Following  a  presentation by  Dr.  Reichlmayr-Lais,  in which  she  reviewed
her published data  as  well as experiments in progress,  all  meeting attendees
were given  an opportunity to  address the Committee.   The  Committee then pur-
sued specific lines  of questioning to its satisfaction  and  retired to execu-
tive session to draft its final report.

     The Committee   was  charged  with  critically evaluating  the  studies  of
Kirchgessner and Reichlmayr-Lais and determining whether or not they supported
the concept of  a  nutritional  essentiality of lead.  Their findings and recom-
mendations   are  contained  in  this  consensus report;  views  expressed  by the
members of  the  Committee in this report are their own and are not necessarily
those of the institutions with which they are affiliated.

-------
                                 INTRODUCTION
     The  Committee commends Drs.  Kircbgessner  and Reichlmayr-Lais  for their
pioneering work, which  is  at the  frontier of  current research on trace metal
nutrition, and wishes to express their appreciation to them for their coopera-
tion in the Committee's efforts to assess their findings.
                               CRITICAL COMMENTS


     The Reichlmayr-Lais and  Kirchgessner  data that were available for review
were derived from two experiments.   Based upon the published and oral descrip-
tions  of  the  experiments,  members  of  the  Committee  expressed  reservations
about  specific  facets of experimental  design, execution,  and documentation,
including the following:


     (1)  No selenium or chromium  was added to the  basal  diet,  nor were
          the concentrations of selenium or chromium measured in the diet
          to indicate  nutritional  adequacy of  those essential elements.
          (In discussion, Or. Reithlwayr-Lais  indicated that Se,  Cr, and
          other essential elements are  being  added to the  diets  in two
          experiments that are currently in progress.)

     (2)  The sole source of fat in the basal  diet was coconut oil, which
          might render the  rats  deficient  in essential fatty acids.  (In
          discussion, Or.  Reichlmayr-Lais indicated that linoleic acid is
          being added  to  the diets  in the experiments that  are  in pro-
          gress.)

     (3)  The possibility exist? that chelant residues (EDTA and APDC)
          may persist in the  basal diet despite the extensive extraction
          procedures that were  employed.  Documentation  of- the  EDTA or
          APDC  concentrations was lacking  and the  Committee considered
          that HPLC or radiotracer experiments would be advisable in order
          to address these concerns.

     (4)  As the basal diet was  prepared,  iron supplements were added in
          an aqueous mixture with several  other  inorganic ingredients
          (e.g.,  KI, CuS04).   Under  the conditions  of drying at 50°C,
          oxidation-reduction reactions  could occur  that might  affect
          the bioavailability of  iron.   The  Committee considered that
          this  potential problem should be  addressed  in  future  experi-
          ments .

     (5)  The method of blood  collection  by  decapitation  and  draining
          into  a  test tube via a funnel  raised  concerns owing  to  the
          potential for contamination by other body fluids  (e.g.,  gastric
          fluid, spinal  fluid, lyaph).

-------
     (6)  The  results  of lead  analyses of blood and  tissues  of the ex-
          perimental  animals have  not been  reported.   (Dr. Reichlmayr-
          Lais  indicated  that such analyses are being attempted in cur-
          rent experiments.)

     (7)  The  possibility that  lead  supplementation  of the  basal  diet
          might affect  its  palatability was not addressed in the experi-
          ments, either by pilot trials or by measurements of food intake.

     (8)  Lead supplementation  of  the basal diet was performed only at a
          single  (relatively  high)  concentration  of  1 ppm.   Further
          experiments at  graded levels of Pb supplementation are desira-
          ble in order to establish a dose-effect relationship.

     (9)  The statistical methods  that were used to analyze the data in
          the two experiments were not described in sufficient detail; the
          application of multiple t-tests may be a cause for concern, and
          the  various  reports  contain  inconsistencies in  numbers  of
          experimental animals per group.  These matters might advantage-
          ously be clarified in a consolidated report of each experiment.
          (Dr. Reichlmayr-Lais  indicated  that  such  a consolidated report
          is in press.)
                        RECOMMENDATIONS AND CONCLUSIONS
     In  view  of  the concerns  that are  listed  above,  the  Committee  reached
the following conclusions and recommendations:


     1.   The Kirchgessner and Reichlmayr-Lais data furnish evidence that
          is consistent with  and,  in some opinions,  indicative  of  a nu-
          tritional essentiality of lead for rats.

     2.   The evidence is  not sufficient to establish nutritional essen-
          tiality of lead for rats.

     3.   To address the basic issue of nutritional essentiality of lead,
          additional evidence needs to be obtained under different condi-
          tions  in  the  laboratory  of  Kirchgessner-Reichlmayr-Lais,  as
          well  as by independent investigators; additional species should
          also be examined.
     The Committee  emphasizes the  difference  that apparently  exists between
lead  concentrations  that  are of  concern  from a  toxicologic  viewpoint  and
those that might possibly be of nutritional concern.  Hence the Committee does
not perceive any practical incompatibility between (a) efforts to reduce Pb in
the human environment  to  concentrations that are unassociated  with  toxic  ef-
fects  and  (b) efforts  to define  the  potential  nutritional essentiality  of
lead.    The   Committee   recognizes  that  current  public  health concerns  for
humans are those of lead toxicity.

-------
                                 ATTACHMENT 1


     The following materials were considered by the Committee in their deli-
berations:


 1.   Reichlmayr-Lais, A. M. and Kirchgessner, M. (1981) Zur essentialitat von
     blei fur das tierische Wachstum. [Why lead is essential for animal growth.]
     Z.  Tierphyslol. Tierernaehr.  Futtermittelkd. 46:1-8.

 2.   Reichlmayr-Lais, A. M. and Kirchgessner, M. (1981) Depletions studien zur
     essentialitat von blei an wachsenden ratten.  [Depletion studies on the
     essential nature of lead in growing rats.]  Arch. Tierernaehr., 31:731-737.

 3.   Reichlmayr-Lais, A. M. and Kirchgessner, M. (1981) Eisenkupfer- und
     zinkgehalte in neugeborenen sowie in leber und milz wachsender ratten bei
     alimentarem blei-mangel.  [Iron-, copper- and zinc contents in newborns
     as  well as in the liver and spleen of growing rats in the case of alimen-
     tary lead deficiency.]  Z.  Tierphysiol.  Tierernaehr. Futtermittelkd.
     46:8-14.

 4.   Kirchgessner, M. and Reichlmayr-Lais, A.  M. (1980) Lead deficiency and
     its effects on growth and metabolism.  Presented at TEMA-4 Meeting; May;
     Perth,  Australia.

 5.   Reichlmayr-Lais, A. M. and Kirchgessner, M. (1981) Activities-veranderungen
     verschiedener enzyme im alimentaren blei-mangel.  [Activity changes of
     different enzymes in alimentary lead deficiency.]  Z. Tierphysiol.
     Tierernaehr.  Futtermittelk 46:145-150.

 6.   Kirchgessner, M. and Reichlmayr-Lais, A.  M. (1981) Changes of iron concen-
     tration and iron-binding capacity in serum resulting from alimentary lead
     deficiency.  Bigl.  Trace El em.  Res.  3:279-285.

 7.   Kirchgessner, M. and Reichlmayr-Lais, A.  M. (1981) Retention, absorbier-
     barkeit und intermeditare neifugbarkeit von eisen bei alimentarem bleimangel.
     [Retention, absorbability and intermediate availability of iron in the
     case of alimentary lead deficiency.]  Jjvt. J.  Vitam. Nutr.  Res.  51:421-424.

 8.   Reichlmayr-Lais, A. M. and Kirchgessner, M. (1981) Katalase- und coerulo-
     plasmin -acktivitat im blut bzw. serum von ratten in blei-mangel.   [Catalase
     and coeruloplasmin activity in blood and serum of rats with lead defici-
     ency.]   Zejrtralb].  Veterinaermed. Reihe A 28:410-414.

 9.   Kirchgessner, M. and Reichlmayr-Lais, A.  M. (1982) Konzentrationen ver-
     scheidener stoffwechsel-metaboliten im experimentellen bleimangel. [Concen-
     tration of different metabolites resulting from experimental lead defici-
     ency.]   Ann.  Nutr.  Metab. 26:50-55.

10.   Reichlmayr-Lais, A. M. and Kirchgessner, M. (1981) Heraatologische veran-
     derungen bei  alimentarem blei mangel.  [Hematological changes in the case
     of  alimentary lead deficiency.]  Ann. Nutr. Metab. 25:281-288.

-------
11.  Schwarz, K. (1973) New essential trace elements (Sn, V, F, Si):  progress
     report and outlook.  Proceedings International Conference Trace Element
     Metabolism in Animals (TEMA) II.  Madison, Wisconsin.  Edited by W. G.
     Hoekstra, J. W. Suttie, H. E. Gantner, and W. Mertz, University Park
     Press, Baltimore, MD.

12.  Pallauf, J., and Kirchgessner, M. (1971) Herstellung der gereim'gten
     halbsynthetischen diat.  [Production of the purified semi-synthetic
     diet.] Z. Tierphysiol. Tierernaehr. Futtermittelkd. 23:128-139

13.  Schnegg, A. (1975) Dissertation, T. U. Munchen.  Excerpt on diet from
     Mr. Schnegg1s dissertation, provided by Professor Kirchgessner.

14.  Kirchgessner, M. and Schwarz, W. A. (1976) Zum einfluss von zinkmangel
     und unter-schiedlichen zinkzulagen auf resorption und retention des zinks
     bei milchkuhen.  [Concerning the influence of zinc deficiency and different
     zinc additions on resorption and retention of zinc in milk cows.]  Arch.
     Tierernaehrung. 26:3-16.

15.  Mertz, Walter (1981) The essential trace elements.  Science (Washington.
     D.C.) 213:1332-1338.

16.  Mushak, P.  (1982) [Appendix 11-A, draft Air Quality Criteria Document for
     Lead].  August 9.  Assessment of studies reporting the potential essential-
     ity of lead.  Available for inspection at U.S. Environmental Protection
     Agency, Environmental Criteria and Assessment Office, Research Triangle
     Park, N.C.

17.  Kirchgessner, M. and Reichlmayr-Lais, A. M. (1982) [Rebuttal to Appendix
     11-A].  September 2. Available for inspection at U.S. Environmental
     Protection Agency, Environmental Criteria and Assessment Office, Research
     Triangle Park, N.C.

18.  Weil, D. (1982) [Letter to M. Kirchgessner].  October 14.   Available for
     inspection at U.S. Environmental Protection Agency, Environmental Criteria
     and Assessment Office, Research Triangle Park, N.C.

19.  Kirchgessner, M. (1982) [Reply to D.  Weil].  October 26.   Available for
     inspection at U.S. Environmental Protection Agency, Environmental Criteria
     and Assessment Office, Research Triangle Park, N.C.

20.  Mushak, P.  (1983) [Appendix 12-A, draft Air Quality Criteria Document for
     Lead].  January 5.  Assessment of studies reporting data regarding the
     potential essentiality of lead.  Available for inspection at U.S. Environ-
     mental Protection Agency, Environmental Criteria and Assessment Office,
     Research Triangle Park, N.C.

21.  Grant, L. D.  (1983) [Letter to M. Kirchgessner].   February 15.   Available
     for inspection at U.S. Environmental  Protection Agency, Environmental
     Criteria and Assessment Office,  Research Triangle Park, N.C.

22.  Kirchgessner, M. (1983) [Reply to L.  D. Grant].  March 28.   Available
     for inspection at U.S. Environmental  Protection Agency, Environmental
     Criteria and Assessment Office,  Research Triangle Park, NC.

-------
23.  Grassmann, E.,  Kirchgessner, M. and Hampel, G. (1970) Zur kupferdepletion
     bei ratten und kuten mil athylendlamin-tetraazetat und adenin,  [Copper
     depletion In rats and chicks as produced by ethylenedlamine-tetraacetate
     and adenine.]  Arch. Tierernaehr. 20:537-544.

24.  Grassmann, E. (1976).  Zur verwertung verschiedener eisenverbindungen bei
     der ratte.  [The utilization of various iron compounds in the rat.]
     Zentralbl. Veterinaermed. Reihe A 23:292-306.

25.  Schnegg, A.  and Kirchgessner, M. (1977) Zur differentialdiagnose von Fe-
     und Mi-mangel durch bestimmung einiger enzymabtivitaten.  [Differential
     diagnosis of Fe and Ni deficiencies by determining some enzyme activities.]
     Zentralbl. Veterinaermed. Reihe A 24:242-247.

26.  Schnegg, A.  and Kirchgesser, M. J.  (1977) Aktivitatsanderungen von enzymen
     der leber und xiere im rtickel-bzw.  eisin-mangel.   [Changes in liver and
     kidney enzyme activities during nickel or iron deficiency.]  Z.  Tierphysj^l.
     Tierernaehr.  Futtermittelkd. 38:300-205.

27.  Schnegg, A.  and Kirchgessner, M. (1977) Konzentrationsanderungen einiger
     substrate in serum und leber bei Ni-bzw.  Fe-mangel.   [Concentration
     changes in some serum and liver substrates with Ni and Fe deficiency.]
     Z.  Tierphysiol.  Tierernaehr. Futtermittelkd. 39:247-251.

28.  Schnegg, A.  and Kirchgessner, M. (1977) Alkalische und saure phosphatase-
     ackivitat in leber und serum bei Ni-bzw.  Fe-mangel.   [Alkaline and acid
     phosphatase  activity in the liver and serum with Ni  versus Fe deficiency.]
     Int.  Z. Vitam.  Ernaehrungsforsch. 47:274-276.

29.  Nielsen, F.  (1983) [Letter to M. Davis].  May 19.  Available for inspection
     at U.S. Environmental Protection Agency,  Environmental Criteria and
     Assessment Office, Research Triangle Park, NC.

30.  Mushak, P. (1983) [Appendix 12-A, draft Air Quality Criteria Document for
     Lead].   July 1.   Assessment of studies reporting the potential essen-
     tiality of lead.   Available for inspection at U.S. Environmental Protection
     Agency, Environmental Criteria and Assessment Office, Research Triangle
     Park, NC.

-------
ATTACHMENT 2

List of attendees at September 29, 1983 meeting of the Expert Committee on
Trace Metal Essentiality:

PANEL MEMBERS

Dr. F. W. Sunderman, Jr. (Chairman) Dr. Walter Mertz
University of Connecticut, School of USDA
Medicine

Dr. Forrest Nielsen Dr. Ekhard Ziegler
USDA University of Iowa

Dr. M. R. Spivey Fox Dr. Kathryn Mahaffey*
FDA NIOSH

Dr. Orville Levander
USDA

INVITED DISCUSSANTS

Dr. Paul Mushak Dr. Anna M. Reichlmayr-Lais
University of North Carolina Technical University of Munich
Federal Republic of Germany

Dr. E. Grassmann (substituting for Dr. M. Kirchgessner*)
Technical University of Munich
Federal Republic of Germany

EPA STAFF PUBLIC OBSERVERS

Dr. David Weil (Meeting Coordinator) Dr. Gary Ter Haar
EPA/ECAO Ethyl Corporation

Mr. Jeff Cohen Dr. Elizabeth Lightfoot
EPA/OAQPS Ethyl Corporation

Dr. J. Michael Davis Dr. Jerry Cole
EPA/ECAO ILZRO

Dr. Robert Elias Dr. Magnus Piscator
EPA/ECAO Karolinska Institute

Dr. Lester Grant
EPA/ECAO

*not present at meeting

7 * US GWEKNNfm PRINTING OFFICE: UK-K»-Hl/im

-------
                                                       EPA-600/8-33-028A
                                APPENDIX 12-C
           INDEPENDENT PEER REVIEW OF SELECTED STUDIES CONCERNING
NEUROBEHAVIORAL EFFECTS OF LEAD EXPOSURES IN NOMINALLY ASYMPTOMATIC CHILDREN:
               OFFICIAL REPORT OF FINDINGS AND RECOMMENDATIONS
               OF AN INTERDISCIPLINARY EXPERT REVIEW COMMITTEE
                                Presented by

                        Expert Coonittee on Perilatric
                                     To:
                        Dr.  Lester fi. Grant, Director
                Environmental  Criteria  and Assessment Office
                United States  Enwironmental Protecttoa
                   Research  Triangle Park, North Carolina
                              November 14,  1983

-------
The materials contained in this report were generated as a result of critical evaluations

and deliberations concerning the subject studies in the course of review of them by members of

the Expert Committee on Pediatric Neurobehavioral Evaluations. The members of the Committee

(listed below) unanimously concur with and endorse the findings and recommendations contained

in the present report as representing the collective sense of the Committee.
Expert Committee on Pediatric
Neurobehavioral Evaluations
Dr. Lyle Jones,
Alumni Distinguished Professor
Dept. of Psychology and
Director, L. L. Thurstone
Psychometric Laboratory
University of North Carolina
Chapel Hill, NC 27514
Dr. Richard Weinberg, Professor
Dept.of Educational Psychology
and Co-Director, Center for
Early Education and Development
University of Minnesota
Minneapolis, MN 55455
Dr. Llpyd Humphreys, Professor
Dept. of Psychology and
Educational Psychology
University of Illinois
Champaign, IL 61820
Dr. Larry Kupper, Professor
Dept. of Biostatistics
School of Public Health
University of North Carolina
Chapel Hill, NC 27514
Dr. Paul Mushak, Associate Professor
Dept. of Pathology and Co-Director,
Environmental Toxicology Research Program
University of North Carolina
Chapel Hill, NC 27514
Dr. Sandra Scarr, Commonwealth
Professor, Dept. of Psychology
University of Virginia
Charlottesville, VA 22901
if

-------
                                            PREFACE

     As part of the periodic (5-year) review and revision of criteria for the National Ambient
Air Quality Standards (NAAQS) for lead established in 1978, the EPA Environmental Criteria and
Assessment Office  (ECAO/RTP)  initiated  in 1982 an intensive, critical evaluation of pertinent
scientific information concerning  health  effects associated with lead (Pb) exposure.   Of con-
siderable  importance  in  that  regard are  certain published  (and  related  unpublished) studies
from  several  different  research groups,  which  provide  data that  have  been  interpreted  as
demonstrating significant associations  between  neuropsychologic  deficits  (e.g., impaired cog-
nitive development)  or other  neurobehavioral  effects (e.g.,  poorer classroom  behavior)  and
lead exposures in otherwise apparently asymptomatic children.  The findings and interpretation
of  such  studies  have  become  a  matter  of great controversy, especially  among  those  research
scientists directly involved in the conduct and reporting of the  subject studies.
     In an effort to resolve major points  of controversy concerning some of the most important
and controversial  of  the  subject studies, an interdisciplinary  Expert  Committee on Pediatric
Neurobehavioral Evaluations was  convened  by Dr. Lester D.  Grant  (Director of ECAO/RTP) start-
ing in March,  1983,  to provide independent peer review of selected studies and to make recom-
mendations concerning  how particular study results should  be most  appropriately interpreted
or, possibly, reanalyzed  before final interpretation.   The Committee comprised internationally
recognized experts  in the  areas of:  child development,  psychometric  techniques,  biostatis-
tics,  lead exposure measurement  techniques, and overall aspects  of  lead  pharmacokinetics and
toxicology.  The present  report  contains  a series of critiques of interrelated sets of selec-
ted studies conducted during the 1970s and early 1980s.
     The Committee focused on answering  the following four general questions in reviewing each
of the sets of studies:

     (1)  Were the  studies  appropriately  designed and conducted  (including data collec-
          tion and statistical analyses)  so as to allow for scientifically sound testing
          of the main hypotheses posed regarding possible associations between lead expo-
          sure and neurobehavioral  effects (e.g., poorer classroom behavior, IQ deficits,
          etc.) in children?
     (2)  To what  extent  do  the particular data, statistical analyses,  and  results  ob-
          tained support  the conclusions stated in the published  papers (or other related
          materials regarding  each study),  and what caveats or  limitations  should most
          appropriately be stated as applying to such conclusions?
     (3)  Are there other conclusions that might be appropriately drawn (given the parti-
          cular design, data collection, and statistical  analyses employed in each study)
          and/or are there  other appropriate approaches to the analysis of the data col-
          lected that would be  expected to yield further meaningful  and important infor-
          mation concerning the  hypothesis  that low-level  lead exposure  leads  to neuro-
          behavioral deficits in children?
                                             iii

-------
     (4)  To what extent  do the published studies allow for meaningful conclusions to be
          drawn  regarding  quantitative exposure-effect  or  dose-response  relationships
          between any observed  neurobehavioral  effects and specific levels of lead expo-
          sure (as defined  by either dentine or  blood lead concentrations as indices of
          exposure)?

     In the course of  deliberating on general issues such as those posed above,  the Committee
considered more  specific  questions or points as appropriate for each of the studies reviewed.
Many of the specific  questions posed were presented in letters from Dr. Grant to the investi-
gators (see attachment  to this report, for a listing of letters).   At initial meetings of the
Committee  in   March, 1983,  these  and  other questions  were  discussed with  the  senior  in-
vestigators responsible for the conduct of particular studies, and some additional, unpublish-
ed information was provided by the investigators to the  Committee to assist in  accomplishing
as complete an evaluation  of each study  as possible  at  the time of  review.   A preliminary
draft of the Committee's report was provided to Drs.  Ernhart and Needleman in September,  1983.
The  Committee reconvened  in October, 1983,  at  which time written  comments   submitted  by
Drs.  Ernhart and Needleman were considered by the Committee in making revisions in the report.
     The Committee members  thank  the  investigators for taking time to meet with  us, for  their
assistance  in providing  and discussing  information  beyond  that  included in  the published
reports of their studies,  and for calling to our attention certain factual errors in the pre-
liminary draft of  our  report.   The Committee  hopes that  the  ensuing critiques  of specific
studies both  (1) help to  resolve  legitimate controversy regarding the most appropriate inter-
pretation^) of the subject study results and (2) provide  constructive criticisms and recom-
mendations that  are  of value in   carrying out  reanalysis  of certain subject  data sets  which
hold promise  for providing  more  definitive outcomes  than those thus  far reported  for  the
studies in the published literature.
                                              iv

-------
                                        TABLE OF CONTENTS

SECTION                                                                                 PAGE

      SUMMARY 	     v

I.    INTRODUCTION 	     1

      A.  Alternative Research Designs 	     1

          1.  Randomized Clinical Trials	     1
          2.  Cross-Sectional Designs 	     1
          3.  Longitudinal Designs	     4
          4.  Time-Series Designs 	     5

      B.  Additional Remarks 	     5

II.   REVIEW OF STUDIES BY DR. CLAIRE ERNHART AND COLLEAGUES  	     6

      A.  Background Information 	     6
      B.  Comments on Perino and Ernhart  (1974) and Ernhart et al.
          (1981) Studies 	     8

          1.  Indicators of Lead Exposure	     8
          2.  Psychometric Measurements and Procedures  	     12
          3.  Statistical Analyses 	     15
          4.  Committee Conclusions and Recommendations  	     17

      C.  Comments on Yamins (1976) Dissertation Study  	     18

          1.  Indicators of Lead Exposure	     18
          2.  Psychometric Measurements and Procedures  	     19
          3.  Statistical Analyses 	,	     20
          4.  Committee Conclusions and Recommendations  	     21

III.  REVIEW OF STUDIES BY DR. HERBERT NEEDLEMAN AND COLLEAGUES	     22

      A.  Background Information 		     22
      B.  Comments on Need!eman et al. (1979) Study	     28

          1.  Indicators of Lead Exposure 			     28
          2.  Psychometric Measurements and Procedures	     31
          3.  Statistical Analyses	     33
          4.  Committee Conclusions and Recommendations	     37

      C.  Comments on Burchfiel et al. (1980) Study	'...	     38
      D.  Comments on Need 1 eman (1982) Report		     39
      E.  Comments on Bellinger and Needleman (1983) Study	     39
      F.  Comments on Needleman (1981) Report	     40

IV.   POSTSCRIPT		..     41

V.    REFERENCES	     44

VI.   ATTACHMENT I	     47
                                              v

-------
SUMMARY

The Expert Committee on Pediatric Neurobehavioral Evaluations reviewed two independent
sets of studies by: (1) Or. Claire Ernhart and colleagues and (2) Dr. Herbert Needleman and
colleagues. The studies evaluated possible associations between low-level lead (Pb) exposures
and neuropsychological deficits in children who were otherwise apparently asymptomatic.
The Perino and Ernhart (1974) study evaluated relationships between blood Pb levels in a
sample of 80 inner city black children (aged 3-5 yr) and IQ scores determined by the McCarthy
Scales of Cognitive Abilities. Small but significant associations between lead exposure and
lower IQ scores were reported, based on regression analyses. The Committee found the blood Pb
measures were of acceptable reliability, as were also the psychometric measures for children.
However, errors now have been discovered in the data analyzed for that report. In addition,
confounding variables may not have been adequately measured, and the statistical analyses did
not deal adequately with confounding variables. The Committee concludes, therefore, that the
study results, as published by Perino and Ernhart (1974), neither confirm nor refute the hypo-
thesis that low-level Pb exposure in children leads to neuropsychologic deficits.
Ernhart et al. (1981), in a follow-up study, reassessed blood Pb levels and neuropsycho-
logic function in a subset of the same children 5 years later. The McCarthy Scales were again
used, along with school reading tests and teacher ratings of classroom behavior. Small but
statistically significant negative correlations were found between school-age blood Pb levels
and scores on some McCarthy subscales, controlling for certain confounders. No significant
associations remained if results were deleted for one "outlier" with markedly elevated dentine
Pb beyond other values for the higher Pb group. The Committee found the psychometric measures
to be acceptable, but the blood Pb sampling method raised questions about the reliability of
the reported blood Pb levels. In addition, the statistical analyses did not adequately con-
trol for confounding factors. The Committee concludes, therefore, that the Ernhart et al.
(1981) results neither confirm nor refute the hypothesis that low-level Pb exposure in chil-
dren is partially responsible for neuropsychologic deficits. The Committee recommends that
longitudinal analyses be carried out, using data from both the Perino and Ernhart (1974) and
Ernhart et al. (1981) follow-up studies.
The Committee also reviewed a doctoral dissertation prepared fay J. Yamins (1976) under
Dr. Ernhart1s direction. The Yamins study attempted to replicate certain aspects of the find-
ings reported by Perino and Ernhart (1974), but used different psychometric measures and a
different population of children. A major problem was the method of blood sampling, i.e.,
collection onto filter paper, which requires correction for hematocrit. Hematocrit levels
vi

-------
apparently are not available for data reanalysis.  Although Yamins reported small but signifi-
cant  effects of  lead  exposure on  some  indices  of  cognitive functioning  (taking  age into
account), the Committee found it difficult to  place much confidence in such findings because
of the failure to control adequately for confounding variables  (besides age).
     Results from an epidemiological study conducted by Needleman and colleagues were reported
or discussed  in:   Needleman et al. (1979), Burchfiel et al. (1980), Needleman (1981), Needle-
man  (1982),  Needleman  et  al.  (1982),  Bellinger and  Needleman (1983),  and Needleman (1983).
The main  set of analyses was presented by Needleman et al. (1979).  The study entailed neuro-
psychologic  evaluations for more  than 2000 first-  and  second-grade  (mainly white) students.
Lead  exposure  was indexed  by dentine  Pb  in  deciduous teeth.  The classroom behavior of each
child submitting  a  tooth was rated by the child's teacher.  Some children, falling within the
highest and  lowest  deciles for dentine Pb measured  in one or  more of  their teeth, underwent
more in-depth neuropsychologic  evaluations,  including use of  an individual  standardized mea-
sure  of  intellectual  abilities  (the  WISC-R)  to  estimate  IQ  levels  and tests  of  academic
achievement, auditory and language processing, visual-motor reflexes,  attentional performance,
and motor coordination.   Needleman et al.  (1979)  reported a  relationship between first tooth
dentine Pb values and percentages of students receiving poor classroom behavior ratings, which
he has  interpreted (Needleman,  1983)  as "a strong  dose-response  relationship."   Children in
the high-Pb group (top 10% of dentine Pb levels) were also reported to have statistically sig-
nificantly lower IQ scores (especially verbal  IQ) than the low-Pb group (lowest 10% of dentine
Pb values),  taking into  account five covariates  in an analysis  of  covariance.   The high-Pb
children were also reported to do more poorly on certain other  neurobehavioral tasks.
     The  Committee  concludes that  the relationship between  dentine  Pb  levels  and teachers'
ratings  of  classroom  behavior  cannot  be  safely  attributed  to the effects  of Pb,  due  to:
(1) reservations regarding the  adequacy  of classification of  subjects  into  Pb exposure cate-
gories using only  the  first dentine Pb value obtained for each child and (2) failure to con-
trol  adequately for effects  of confounding variables.   The Committee  also concludes that the
reported  results  concerning the effects of  lead on IQ  and other  behavioral neuropsychologic
abilities measured  for  the low-Pb and high-Pb  groups  must be  questioned, due to:  (1) errors
made in calculations of certain parental IQ scores  entered as a control variable in analyses
of covariance; (2)  failure  to take age and father's education into account adequately in the
analyses  of covariance;  (3)  the  failure to employ a reliable  strategy for the control of con-
founding variables; (4) concerns regarding missing data for subjects included in the analyses;
and (5) questions about possible bias  due to exclusion of large numbers of provisionally eli-
gible subjects from statistical  analyses.  The Committee concludes, therefore,  that the study
results,  as  published  by Needleman et al.  (1979), neither confirm nor  refute  the hypothesis
that low-level  Pb exposure in children leads to neuropsychologic deficits.
                                              vii

-------
     The publications  by Needleman  (1982),  Needleman et al.  (1982),  Bellinger  and Needleman
(1983),  and  Needleman  (1983)  describe  further  analyses  of  the same  data set  reported  by
Needleman et  al.  (1979).  Burchfiel  et al.  (1980) reported analyses  of certain psychometric
data together with  additional  data on electrophysiological  (EEG) measures for a subset of the
high-Pb  and  low-Pb children from  the  Needleman  et al.  (1979) study.  The  above reservations
regarding the basic analyses reported by Needleman et  al.  (1979) apply  also to the analyses
reported by Burchfiel  et al.  (1980), Needleman (1982),  Needleman et al.  (1982),  Bellinger and
Needleman (1983),  and Needleman  (1983).   Similar reservations  apply to analyses  of another
data set (Needleman, 1981).  The  Committee  recommends  that the entire  Needleman  data  set  be
reanalyzed,  correcting for errors in data  calculation  and entry,  using better  Pb exposure
classification,  and appropriately adjusting for confounding factors.
     In  addition  to evaluating the  studies  of Ernhart  and Needleman, the  Committee reviewed
available reports  (some published  and others as  yet unpublished) of other  studies from the
United States  and Europe.  Although an exhaustive, in-depth evaluation of the world literature
on  low-level  Pb exposure  was  beyond  the  current charge to  the Committee, we  note that new
studies  reported in the spring and summer of 1983, with only a few exceptions,  failed to find
significant  association between low-level  Pb exposure  and neuropsychologic deficits, once con-
trol variables were taken into account.
     From its review of the recent research  literature  covered  in this  report,  the Committee
concludes that:   (1) in the absence  of control  for  other variables, a negative association
between  Pb exposure and neuropsychologic  functioning  has been established;  (2)  the extent  of
this negative association  is  reduced or eliminated when confounding factors are  appropriately
controlled;  and (3)  the  Committee knows  of no studies that, to date, have validly established
(after proper control  for  confounding variables) a relationship between  low-level Pb exposure
and neuropsychologic deficits in children.
                                             vi ii

-------
                                         INTRODUCTION

     In  approaching  its task,  the Committee  was  faced  first  with establishing criteria for
research  studies,  the results  of  which  may be accepted  as  evidence pertinent to determining
the  influence  of Pb  exposures on cognitive functioning  in  apparently asymptomatic children.
Because children  live and  mature  in a complex socio-cultural milieu that affects them in many
diverse ways,  isolation of  a definitive  cause,  e.g., lead  exposure, for neuropsychological
problems  in children  is extremely  difficult.  Under these circumstances, what kind of research
design is necessary or adequate to produce pertinent evidence?
     The  problem of determining the effects of Pb on cognitive functioning is viewed as an in-
stance of a general class of dosage-response problems. Alternative research designs with which
to approach such problems include:

     (i)   randomized clinical trials;
     (ii)  cross-sectional  observational  study of  individuals  from  groups  known to  vary  in
           exposure (dosage);
     (iii) longitudinal study of the same individuals over time;
     (iv)  a time series of  observations on different sets  of  individuals who are members of
           groups known to differ  in exposure (dosage).

A.   Alternative Research Designs

     1.    Randomized Clinical Trials

     There is  no question that randomized clinical trials,  properly  conducted,  provide evi-
dence that is  highly relevant to  the  research question.   Neither is  there  any question that
the  experimental  administration  of  Pb  to  human  subjects  is unethical, and  not to  be con-
sidered.   This highly effective research design, then, simply cannot be adopted to address the
question of the effect of Pb  on human cognitive functioning.

     2.   Cross-Sectional Designs

     The  bulk  of published work assessing  Pb  effects on human cognitive  functioning  has en-
tailed the cross-sectional study  of  a sample  of children.   A serious complicating feature of
the  design  results  from typical   empirical  findings  of  association  between low  or  moderate
levels of lead exposure, on the one hand, and such background variables as parental IQ, paren-
tal education, quality  of  home environment, family size, etc.,  all  of which are  known  to  be

-------
correlated with  children's cognitive  performance.   Under what  conditions, then,  might this
design yield valid conclusions about the effect of Pb on cognition? Three possibilities appear
to exist, as follows:

          (a)  Were  low  or moderate levels  of Pb consistently found to  be negatively corre-
               lated with  cognitive performance, while  all  potential  confounding background
               variables were  negligibly  correlated with cognitive performance,  then  a valid
               conclusion would  be that  Pb  is responsible for the cognitive  deficits.   How-
               ever, the premise  generally appears  to be false:  published studies on Pb, con-
               sistent with  research  literature in child psychology, report sizable correla-
               tions between cognitive performance and a host of background variables.
          (b)  Were  low  or moderate levels  of Pb consistently found to  be negligibly corre-
               lated with  cognitive  functioning,  regardless  of  the  pattern  of association
               between cognitive  function and confounded background variables,  then it would
               be fairly safe to conclude that the differences in Pb levels are not important-
               ly  related  to  cognitive  performance.   Again,  however,  the  premise generally
               appears to be false:  most published studies  on Pb report significant correla-
               tions  between  Pb  and  cognitive test  scores  unadjusted  for  other key  con-
               founders.
          (c)  Consider a study designed  so  as to provide a  factor analysis of interrelations
               among variables.   It might be found that cognitive performance  is represented
               on  one  factor  along  with  noncognitive variables  that  are not  appreciably
               associated with cognitive performance in the absence of Pb.  In the presence of
               Pb, however, such noncognitive variables might be hypothesized to be associated
               with cognitive function.  Lead would be the primary defining variable on such a
               factor.   Noncognitive variables that would be appropriate  candidates for this
               factor  analysis include  sensory  discriminations  and  electroencephalographic
               (EEC) recordings.   This  finding would support  the hypothesis that Pb was a par-
               tial determinant of cognitive functioning.

     The research studies of Pb effects on cognition of which the  Committee is  aware generally
fail  to match any of the above conditions  (a), (b),  or (c).   Rather,  the studies mainly report
cross-sectional data for which:   (1)  Pb is correlated with cognitive  test scores by a nonzero
but modest  amount; (2) Pb is  correlated  with background variables; and  (3)  background vari-
ables  are  correlated with cognitive performance.   In  most such studies, efforts  are  made to
separate the influence of  Pb on cognitive functions  from  the  influence of confounding varia-
bles, using methods of statistical adjustment (e.g., regression analysis).

-------
Statistical adjustment for confounding variables may reduce the residual relation between
Pb and cognition to a negligible value. If so, however, it would not necessarily follow that
cognitive functioning is not influenced by Pb; the effects of Pb might be masked by one or
more of the confounding variables. The research design is generally incapable of providing
evidence which permits a clear separation of the magnitude of effect of Pb from effects of the
confounding variables.
Statistical adjustment for confounding variables may leave a significant residual corre-
lation between Pb and cognition. If so, however, it would not necessarily follow that cogni-
tive function is influenced by Pb. Perhaps other, background variables, not explicitly ad-
justed for, but correlated both with Pb and cognition, are the effective determinants of cog-
nitive differences. Or, perhaps, less fallible measures of the confounding variables would
have further reduced the correlation between Pb and cognition to a non-significant amount.
The research design is not sufficiently sensitive to provide guidance concerning which of
these alternative conclusions should be embraced.
the controversy over the interpretation of results from a series of recent studies may be
attributed to this intrinsic ambiguity regarding the assignment of causal status to the pre-
dictor variable of interest (Pb) or to confounding variables (e.g., home and parental mea-
sures). Some consider Pb to act as a surrogate for the cohfounding variables. Others con-
sider the confounding variables to act as a surrogate for Pb. There is no scientific basis
for accepting or rejecting either set of interpretations.
In view of these considerations, the Committee concludes that, no matter how carefully
designed and executed, cross-sectional studies of relationships between Pb and cognitive func-
tioning are not able to yield definitive conclusions regarding the influence of low-level Pb
exposures on human cognitive functioning, when measures of both are correlated with background
variables also known to influence cognitive development and performance. At best, such cross-
sectional studies may yield evidence suggestive of effects of low-level Pb exposures which
would need to be confirmed by studies using more definitive research designs.
The Committee has been charged with evaluating the research reports of Needleman and his
associates and of Ernhart and her associates. All of these reports concern essentially cross-
sectional studies (although some of the Ernhart data are amenable to longitudinal analysis, a
subject to which we return later). Each study is thus subject to the severe reservations ex-
pressed above: i.e., from cross-sectional studies with confounding variables, it is not possi-
ble to draw definitive conclusions about the role of low-level Pb exposures as a determinant
of cognition. Nevertheless, we do present more detailed critiques of these studies in sec-
tions below, recognizing the importance of attempting to resolve apparent inconsistencies in
the results and conclusions presented by these sets of investigators and recognizing, also,
the value of accumulating even small or suggestive indications of possible relationships, or
lack thereof, between Pb and cognitive deficits.
3

-------
We have judged that, to assess effects of Pb on cognition, (i) randomized clinical trials
cannot be conducted and (ii) cross-sectional observational studies cannot adequately disentan--
gle effects of Pb from effects of confounding variables. We now comment on strengths and weak-
nesses of certain other research designs, longitudinal studies and time-series analyses, which
may be capable of yielding more definitive conclusions than cross-sectional designs.

3. Longitudinal Designs

A longitudinal design is characterized as a study of the same individuals over a period
of time. For the topic at hand, primary interest would reside in changes over time in rela-
tive levels of cognitive functioning as a consequence of earlier levels of systemic Pb or of
changes in systemic Pb. Many (but not all) of the confounding variables are usually quite
stable over time, and thus may be assumed to have a similar influence on measures obtained at
different times. To that extent, the difficulty created by confounding variables will be at
least partially alleviated.
Documentation of the history of systemic Pb exposure should begin early, even prior to an
infant's birth. Cognitive performance also should be assessed early, as soon as 18 months
after birth. Measurements of both sets of variables should be repeated periodically for
several years, and other measures, e.g., dentine lead, might be obtained at appropriate times.
It is crucial that as complete a history as possible of Pb exposure (as indexed by changes in
internal indices) be obtained and that such indices of exposure be evaluated for relationships
to dependent variables indicative of cognitive/behavioral development both proximate and dis-
tant in time after the exposure measures are obtained. This is important both to increase
information on latency periods for Pb effects to be manifested and in regard to augmenting our
knowledge of reversibility/irreversibility of Pb effects.
In the study by Perino and Ernhart (1974) discussed below, the same sample of children
was assessed both for Pb exposure and cognitive performance at ages 4-5 years, and again five
years later, as reported by Ernhart et al. (1981). The authors did inquire about the relation-
ship between later cognitive measures and earlier Pb levels, but they failed to study the
possibly revealing relationship between change in cognitive score and change in Pb levels (see
comments on these studies in a later section of this report).
A longitudinal design reduces but does not necessarily totally avoid problems of con-
founding variables. Variables that remain stable over time for a given individual, while
creating difficulty in a cross-sectional design, may not be as much of a problem in a longitu-
dinal design. However, confounding variables that change over time would be troublesome in
longitudinal as well as in cross-sectional studies. Techniques for statistical adjustment may
(and usually should) be employed for such variables. To the degree that they are prominently

-------
related to Pb levels and cognition or to changes in these variables, they are as troublesome
in longitudinal as in cross-sectional studies. The hope is that their effects will be far
smaller in longitudinal studies.

4. Time-Series Design

Fortuitous events, from the research perspective, may occasionally provide the opportuni-
ty for a time-series study of the effects of Pb on cognitive functioning. For example, it
might be recognized that an environmental change is imminent in Community A, a change antici-
pated to have a large effect on typical systemic Pb exposure in that community. Prior to that
change, Pb and cognitive measures might be obtained for a random sample of 5-year-olds (or 7-
or 9-year-olds) in community A, and also in community B, considered similar to A except for
the impending change. Later, after the environmental change and its effects have had an
opportunity to be exerted, similar measures are again collected on random samples at the same
age, in both communities. Differential changes in cognition as a function of different levels
of Pb may strongly suggest that Pb has influenced cognition. A specific example of where
such a research approach might be applied is a situation whereby an imminent governmental or
industry action is anticipated that would lead to substantial reductions in Pb exposure in a
particular geographic area.

B. Additional Remarks

The Committee cannot conclude these general introductory remarks without presenting an
additional caveat regarding the interpretation of even an unambiguous finding of a signifi-
cant negative relationship between low Pb levels and measures of children's IQ or other
behavioral variables, based on epidemiological observations. If an investigator is able to
discount the influence of confounding variables, and if no flaws are found with the research
design employed or the conduct of the study, the temptation may exist to conclude that Pb is
responsible for the observed lowered IQ levels or other behavioral deficits. Note, however,
that such results are, in many cases, equally consistent with the conclusion that increased Pb
exposures and associated body burdens are a consequence of low IQ or other observed behavioral
deficits. Furthermore, knowledge external to the research study generally would not be such
so as to provide an obvious basis for preferring one of these conclusions over the other.

-------
REVIEW OF STUDIES BY DR. CLAIRE ERNHART AND COLLEAGUES

A. Background Information

The Committee undertook detailed review of two studies published by Dr. Ernhart and col-
leagues (Perino and Ernhart, 1974; Ernhart et al., 1981) and, also, preliminary review of a
third study reported in the 1976 doctoral dissertation of J. Yamins at Hofstra University.
(The latter doctoral research was conducted under Dr. Ernhart1s direction but is not yet pub-
lished in the peer-reviewed literature).
In the first study, based on the doctoral research of J. Perino (under Dr. Ernhart's
direction), inner city black children of low socioeconomic status were recruited for study
based on blood Pb values obtained during screening for possible undue Pb exposure by the New
York City Health Department during 1972. Children were randomly selected to participate in
the study so as to represent a group of subjects with lead exposures ranging from low (<30
ug/di) to moderately elevated (40-70 ug/d£) according to then existing screening standards.
Because the study was designed to evaluate neuropsychologic deficits associated with moderate
lead exposures in non-overtly lead-poisoned children, children with histories of overt signs
or symptoms typical of Pb poisoning were excluded from the study. Eighty black children (41
boys and 39 girls) of preschool age (3 yr to 5 yr, 1 mo) from Queens, New York, were included
in the study. The McCarthy Scales of Children's Abilities (McCarthy, 1972) were administered
to the children in their homes by a trained school psychologist (J. Perino), to yield a
General Cognitive Scale score with norms obtained in the same manner as and roughly comparable
to IQ scores. The test also provided scores on several subscales, i.e. Verbal, Perceptual
Performance, Quantitative, and Motor Abilities, Parental IQ was measured by means of the
Quick Test (Ammons and Ammons, 1962) of gross intellectual level. Questions regarding other
covariates were administered to the parent by the school psychologist, following a standard-
ized format and recording answers on a typed questionnaire form. In general, the results of
the study were such so as to lead the authors to conclude that neuropsychologic deficits
(i.e., decreased cognitive, verbal, and perceptual performance abilities) were significantly
associated with Pb exposure in the otherwise asymptomatic children studied. The results have
also been interpreted (in Air Quality Criteria for Lead, U.S. EPA, 1977) as demonstrating such
deficits to be associated with Pb exposures resulting in blood Pb levels of 40-70 ug/dfL.
The study by Ernhart et al. (1981) is a followup study, in which 63 children (30 boys, 33
girls) from the original cohort of 80 black children studied by Perino and Ernhart (1974) were
reexamined 5 years later. Scores were obtained for these school-age children on the McCarthy
Scales of Children's Abilities, school reading tests, teacher ratings of classroom behavior,

-------
and  several  neurobehavioral  exploratory measures.  Hypothesized relationships between perfor-
mance  on these  neuropsychologic tests  and childhood  Pb  exposures were  first statistically
evaluated by means of  omnibus  multivariate tests (Hotelling's T*), comparing  test scores of
"low  lead"  versus  "moderate lead" children defined in terms of (1) pre-school blood Pb levels
(low  = 10-30 (tg/dl  ;  moderate =  40-70 ug/d£)  and  (2)school-age blood  Pb  levels  (low S26
ug/d£; moderate = 27-49 ug/d£).  Significant omnibus test results (p<0.05) were obtained only
for  reading  scores  related to both preschool  and school-age Pb levels.  Further multivariate
and  univariate  analyses were  conducted for these significant  neuropsychologic outcome vari-
ables.  Univariate  tests  for differences on McCarthy scores  between  low and moderate Pb sub-
jects  (ignoring  control  variables)  suggested that the moderate Pb group performed more poorly
on the  General  Cognitive Index (GCI) and 3 of the 5 McCarthy Test subscales.  However, multi-
variate  (regression)  analyses revealed  that sex  and parental IQ were  control  variables that
were  significantly  correlated  with  one or more outcome measures.  When these control variables
were  ignored  in  analyses including  Pb exposure measures, preschool Pb was significantly nega-
tively  related to  scores on the GCI, 4  of  5 McCarthy subscales,  and the reading tests.   When
sex  and parental IQ  were  taken  into  account, however, preschool  Pb was  not  related to any
neuropsychologic outcome measure and school-age Pb was  significantly  related only to  the
McCarthy GCI,  verbal  subscale, and motor  subscale scores (with the variance  attributable to
school-age Pb generally being less than half that found for the same outcome measures  when the
control variables  were  excluded).  Dentine Pb  levels  in shed deciduous teeth  of  33  children
were  not significantly  related  to any outcome measures, even  when control variables  were ig-
nored.
     The Yamins  (1976)  dissertation  study, in part,  attempted to  replicate  the  Perino  and
Ernhart (1974) findings, using a different study population and psychometric tests.   Preschool
children (aged 2 yr,  4  mo to 5 yr,  9 mo),  including 80 black (38  girls, 42 boys) and  20 White
(10  girls,  10 boys) children from  low  to  low-middle socioeconomic status  communities in the
Nassau County (Long Island) Department of Health Clinics lead-screening program catchment area,
were included in the study.   Children with overt signs or symptoms of Pb intoxication  were ex-
cluded.  Of  the  included  children,  54  black  subjects  had blood  Pb  levels below  37 ug/d£,
whereas 26 fell  in  a 38-70 pg/dJl range;  of the  20 white children,  19 had blood Pb levels  below
37 ug/d£.   Cognitive performance and language development of the children were assessed by the
following tests administered  in  a set sequence in the  home by a  school psychologist  blind as
to the children's Pb levels:   a "nonverbal"  IQ  test (Peabody Picture Vocabulary Test;  PPVT); a
general information test (Caldwell  Preschool  Inventory); a concepts test (Block Sort); a per-
ceptual-motor functioning test  (Copy  Forms);  and a sentence  repetition task designed for the
study.  The  Ammons  Quick  Test  was  used to measure  parental  IQ, and  data were gathered  on
several other  background variables  (parental  education and  occupation,  quality of  housing,

-------
child's medical history, number of siblings, etc.) by means of standardized questionnaire and
rating forms. Raw scores for all dependent variable measures were used in multiple regression
analyses, which took into account age as well as the other potentially confounding background
variables that were measured. For the black children, several such variables {e.g., parental
IQ and education, absence of father from home, etc.) were significantly negatively related to
children's Pb levels but positively related to each other and the children's cognitive and
language variables. Similar results were obtained for the white children. Stepwise multiple
regressions were then performed, excluding predictor variables contributing less than 1% of
the variance, entering the included predictor variables into the equation before Pb. With all
predictor variables controlled, for the black sample, Pb contributed 2.4% of the total vari-
ance for nonverbal intelligence (PPVT, p<0.05), 3.1% for general information (Preschool Inven-
tory, p<0.01), 2.4% for overall level of acquired syntax (Total Repetition Score, p<0.05), and
2.5% for ability to repeat nongrammatical verbal stimuli (Ungrammatical Stimuli Test, p<0.05).
Lead did not contribute significantly to conceptual level (Block Sort), perceptual-motor func-
tioning (Copy Forms), or ability to repeat grammatical stimuli (Grammatical Repetition) scores.
In order to evaluate critically the above studies, the Committee met with Dr. Ernhart at
EPA facilities in Research Triangle Park, NC on March 17-18, 1983. At that time, a summary
overview presentation was made by Dr. Ernhart on the objectives, design, data collection and
analysis procedures, and results for each of the studies. Certain listings of raw data values
(provided in coded form to protect the privacy of subjects) and other pertinent published and
unpublished materials were examined by the Committee and considered during discussions with
Dr. Ernhart regarding diverse aspects of the studies reviewed. Some additional, follow-up in-
formation was requested by the Committee and was provided to them subsequent to the March
17-18 meeting with Dr. Ernhart. See Attachment 1 for a list of materials examined by the Com-
mittee in connection with their review of the subject studies. The Committee's comments
regarding the most salient points of concern and controversy related to methodological and
other features of the above studies by Ernhart and associates are presented below.

B. Comments on Perino and Ernhart (1974) and Ernhart et al. (1981) Studies

1. Indicators of Lead Exposure

In the first two studies under consideration, Perino and Ernhart (1974) and Ernhart et
al. (1981), the major indicator of exposure was blood Pb, with additional use of erythrocyte
porphyrin (EP) measurements and a sub-group of dentine Pb samples in the follow-up study of
Ernhart et al. (1981).

-------
     On  the  basis of current criteria  of  methodology assessment for Pb  analyses  as noted in
Chapter  9  of  the EPA draft document Air Quality Criteria for Lead (U.S. EPA, 1983), it can be
said that the blood lead values in the Perino and Ernhart study are reasonably reliable.  With
the follow-up study of Ernhart et al. (1981), blood lead accuracy becomes potentially problem-
atical,  due both to:   (1) a combined positive  bias  of capillary blood sampling and choice of
analysis,  and  (2) a  bias of negative  direction  but of possibly variable size, owing to the
generally poorly  recognized  effect  of whole blood hematocrit/hemoglobin  on  blood  Pb measure-
ments  using  filter paper.  The latter  factor  requires making use of the hematocrit measure-
ments  for  the  subjects'  blood (which  presumably are  available,  since EP  measurements  also
require  knowing  the  hematocrit)  to  correct for differential  spread or diffusion of blood (and
concentration of Pb therein) on the filter paper matrix.
     Measurement of Pb in dentine of shed teeth from the subjects in the Ernhart et al. (1981)
follow-up study  was  carried  out in the  laboratory which  both pioneered the analysis of Pb in
teeth  and  probably  has  the most experience and proficiency with such analyses.   The method of
analysis for dentine  Pb  is reasonably reliable,  and  it appears  that analysis error in repli-
cate sampling is  at the  step of isolating  the  dentine zones from a given whole tooth sample.
     Measurement of erythrocyte porphyrin involved  blood collected on filter paper and subse-
quent  elution  and micro-fluorometric analysis.   Such a "wet" or laboratory  analysis  is  con-
sidered much more reliable than the  use of the  hematofluorometer when applied to blood samples
of modest EP content.   Analytical  error was noted to be less  than 15 percent.
     In the study of Perino and Ernhart (1974)  lead exposure  in the pediatric subjects was in-
dexed  by analysis of  venous  whole blood.  Characteristics of the specific procedures employed
and pertinent evaluative  comments  are as follows:

          (a)   Venous blood samples were collected  by trained technicians in a  lead screening
               program and analyzed  in  the laboratory facilities of the New York City Depart-
               ment of Health (NYCHD)  during the summer of  1972.   Sampling  involved standard
               precautions to minimize  sample  contamination  and collection during  the summer
               months, when  blood  lead values  in the city are known to  be  maximal.   Samples
               were analyzed  within  48 hours of collection and refrigeration.
          (b)   As a lead  analysis facility,  the NYCHD laboratory has processed  a large volume
               of samples for lead content over many years, an important consideration in  view
               of the fact that laboratory proficiency is directly related to the  level of Pb
               analysis  activity.
          (c)   The specific method of  blood lead analysis employed was  the  Hessel  extraction
               variation   of  atomic   absorption  spectrometry  (Hessel,   1968),  a macro  method

-------
using venous blood which still enjoys popularity up to the present time. The
periodic surveys by the Centers for Disease Control (CDC) of participating
laboratories in their proficiency programs indicate (see Boone et a!., 1979)
that the Hessel method is somewhat more accurate than the Delves cup technique
and more accurate than the other variations of atomic absorption spectrometry.
Precision tends to be less than for the Delves cup procedure, consistent with
the reported analytical error of ±5-6 |jg Pb/d£ for the NYCHD laboratory when
using the Hessel method (communication of B. Davidow to N.B. Schell, see
Ernhart, March 11, 1983: summary of conversation between N.B. Schell and C.
Ernhart). Compared to isotope dilution mass spectrometry (IDMS), the Hessel
method for the range of blood lead in the Perino and Ernhart study shows a
modest positive bias of 2.5 ug Pb/dJL
(d) At the time of data collection for the Perino and Ernhart study, it was the
practice of the NYCHD laboratory to report blood lead values rounded to the
nearest decile of blood lead. Hence, a subject blood Pb value of 40 \jtg/dA in
the Perino and Ernhart (1974) report would have resulted from a reading of some
value between 35 and 44 ug Pb/d£.
(e) Internal and external quality control protocols were in place in the NYCHD
laboratory at the time of the subject study. The latter consisted of partici-
pation in both the CDC and New York State proficiency testing programs, and the
NYCHD laboratory met acceptable proficiency standards.
(f) It appears that there are no major difficulties with methodological aspects of
the blood Pb data. The moderate positive bias in the Hessel procedure, if
corrected for among the Perino and Ernhart (1974) study data, would result in a
constant shift downward in all values. Since there was decile rounding, abso-
lute corrections for this bias would require having the original blood Pb
values.
(g) The full use of confidence bounds for the lead measurements, if employed in any
overall reanalysis of the data, would require having the original blood lead
values as well as taking into account the analytical error noted above.

Lead exposure levels for subjects in the follow-up study of Ernhart et al. (1981) were
indexed via measurement of Pb in capillary blood (applied to filter paper) and free erythro-
cyte protoporphyrin (FEP) determination. A sub-group of the follow-up population furnished
shed teeth for dentine lead analysis. Characteristics of the procedures used and evaluative
comments are as follows:
10

-------
(a) Capillary blood samples were collected on filter paper and analyzed In the same
NYCHO laboratories (as noted above) that assayed blood Pb by the Delves cup
variation of atomic absorption spectrometry.
(b) It Is now generally accepted that capillary blood samples, as compared to those
obtained by venous puncture, manifest a significant positive bias in lead level
even under relatively stringent sample collection conditions. The best data
base by which to estimate the magnitude of this positive bias is the NHANES II
survey, which indicated that for the Delves cup procedure used in the present
study under discussion capillary blood Pb was 6 ug/d£ higher than for venous
samples. An additional positive bias of approximately 3 ug/d£ exists for the
Delves cup analysis compared to the definitive IDMS method.
(c) A generally unrecognized problem with the analysis of lead in blood using
filter paper spotting has to do with the close dependence of blood flow (on
filter paper) on hemoglobin/hematocrit content. The study of Carter (1978)
makes it clear that blood flow is increased on filter paper with decreasing
hematocrit, resulting in proportionately lower blood lead values contrasted to
venous blood analysis.
(d) In view of the documented problem of using blood on filter paper without
correction for hematocrit, blood lead values in the Ernhart et al. (1981)
report would have to be appropriately corrected, if this was not already done
initially by the NYCHD. It should be noted that Carter (1978) observed under-
estimation of lead content of blood on filter paper at hematocrit values that
would be considered in a normal range.
(e) Dentine lead levels were determined in the laboratories of Dr. Irving Shapiro,
School of Dental Medicine, University of Pennsylvania, a facility which
pioneered such analyses and is recognized as having the most proficiency in
such measures. In this study, dentine was isolated from a given whole tooth
sample.
(f) Samples of isolated dentine were dissolved in perchloric acid, buffer added,
and lead measured by an electrochemical technique, anodic stripping voltam-
metry. At the levels of lead being measured in the dentine samples, this tech-
nique provides reasonably reliable results for the solubilized analyte.
(g) From data of Shapiro et al. (1973), duplicate analysis of dentine with low and
elevated lead exposure of subjects indicates that the analytical error in-
creases with concentration, suggesting greater variation in replicate section-
ing than in the instrumental measurement itself.
11

-------
(h) Identification of an "outlier" by Ernhart in further unpublished analyses of
data from the follow-up study (see following sections) was done on the basis of
a dentine lead value being 107.4 ppm Pb in the subject sample. A methodolo-
gical problem accounting for this high value has been discounted by Shapiro
(personal communication of I. M. Shapiro to C. B. Ernhart, see Ernhart,
February 3, 1983, letter to D. Weil), who indicated that any contamination of
the dentine section by inclusion of circumpulpal dentine, a region manyfold
higher in lead, would only account for 3-4 percent of the above dentine lead
level.
(i) Erythrocyte porphyrin analysis in the present study was carried out in the
NYCHD laboratory which simultaneously analyzed capillary blood for lead. It
was noted that analytical error was less than 15 percent, using a microfluoro-
metric analysis of blood samples eluted from the filter paper. This laboratory
employs internal and external quality control protocols for EP analyses, the
latter including the EP proficiency testing program of CDC.

2. Psychometric Measurements and Procedures

Comments on the psychometric measurements employed in the Perino and Ernhart (1974) study
and the results obtained are as follows:

(a) The study employed the McCarthy Scales to assess the intellectual development
of young children and the Ammons Quick Test to assess parents' IQ levels. The
Committee agreed that the McCarthy Scales were appropriate for assessing intel-
lectual abilities of the children in this sample, whose race, age range, and
socioeconomic status (SES) were represented in the standardization sample. The
Ammons Quick Test, however, has more questionable validity for two reasons:
(1) the content of the test is a very limited sample of adult intelligence,
and (2) the test was not standardized on low SES black subjects. The reliabil-
ity of the McCarthy Scales is satisfactory for the present research. The reli-
ability of the Quick Test, however, is only about .75, a low value for an adult
measure. More importantly, correlations with the Stanford-Binet and WISC range
from only .10 to .80, with a- median in the .40 range. Large discrepancies have
been observed between Quick Test Scores and individual IQ test scores (Sattler,
1982). Although the Quick Test is considered useful in large scale research
studies, where a simple and quick assessment of average ability is needed, the
measure is less adequate in the Perino and Ernhart study where it was used as a
12

-------
control for Individual differences. A measure such as the short form of the
WAI$ or the WAIS vocabulary scale would have provided a better estimate of
parental IQ.
(b) The administration of the tests by the first author, a school psychologist, was
blind with respect to the children's blood Pb levels. Although the tests were
administered in the home, under nonstandard conditions, the Committee concluded
that the assessments were generally valid. Because of the training and clini-
cal experience of the investigator, the Committee thought it unlikely that the
assessments were seriously compromised.
(c) Birth weight, history of birth risk factors, and maternal education were
reported by the mothers in an interview, but were not verified by checking of
appropriate records. Maternal occupation was recorded from clinic records em-
ploying a 1950 census classification that was inadequate to differentiate among
urban blacks. Thus, nearly all families in this sample fell within two occupa-
tional classes. It is not known whether a finer SES scale would have resulted
in greater relations between SES and other variables in the study than those
reported by Peri no (1973).
(d) In view of the limitations of the Quick Test and the measurements of some of
the control variables, it is especially important that new analyses of the data
(proposed below) be employed to maximize the efficiency of the control vari-
ables.
(e) Descriptions of quality control procedures by Dr. Ernhart regarding the check-
ing of data entries onto computer cards and/or tapes seemed to indicate that
reasonable care was taken to ensure accurate encoding of data for statistical
analyses. The Committee had no feasible way to confirm this independently, but
Dr. Ernhart, in responding to the Committee's request for reanalyses of the
data set, reported as follows: "In the course of conducting the reanalyses to
include parent education, we found several errors in the data and the previous
analyses. One child's age was incorrect by one year (38 months rather than 50
months). This changed his General Cognitive Index (GCI) from 50 to 86; other
scores were correspondingly affected. Another child's GCI was incorrect by one
point. The degrees of freedom used and reported in the tests of significance
of the lead effect (regression analyses) should have been 1 and 75, not 4 and
75."

Comments on the psychometric assessment procedures used by Ernhart et al. (1981) in the
follow-up study are as follows:
13

-------
(a)  Because this study is a follow-up of the Perino and Ernhart sample, many of the
     issues raised about  the original study apply  here.   Also,  use of the McCarthy
     Scales with age  groups  not included in the standardization sample is question-
     able.  More  than half of the  children  were beyond the age range  of the test,
     and  their  test  scores were determined by  linear  extrapolation.   The Committee
     believes that  a better  procedure would have  been the use of residuals after
     regressing raw McCarthy Scale  scores on age at testing.  The Committee acknow-
     ledges that no  subject  reached the ceiling  on the subtests  and also that some
     value  exists  in  re-administering  the  same measure  in a  longitudinal  study.
     However,  the  Committee  concludes  that  it  would  have been preferable  to have
     chosen an age-appropriate measure of intelligence, such as the WISC-R, which is
     based on a suitable  standardization sample and also taps cognitive performance
     of older children.  Despite reservations about the use of the McCarthy Scales in
     the  follow-up, the Committee  does not believe that the assessment of intellec-
     tual development was  seriously compromised.  Because the children in the study
     were functioning at levels well below those of most children of their ages, the
     test was more appropriate for them than for most children  of the same age but
     of average intelligence.
(b)  Reading test  scores  were  obtained from many  different tests.   The Committee
     concludes,  however,  that  the  combination of multiple measures  does  not neces-
     sarily compromise the assessment of reading achievement (Scarr and Yee, 1980).
     The age correction used was appropriate.
(c)  More  than  one psychologist  collected  the  follow-up McCarthy  data,  generally
     within the children's schools.   The Committee believes that the administration
     and  scoring of the protocols  were adequate.  The testers were both blind as to
     the children's Pb levels and well-trained in psychometric assessment.
(d)  Correlations between the  earlier  and  later  administrations  of  the McCarthy
     Scales, as reported  in  Ernhart et al.  (1980),  were in the moderate range (from
     .24 to .61).   The highest correlation is for the General Cognitive Index, which
     is  considered by the Committee  to be  the most important cognitive  measure in
     the  study.   In  light  of  developmental changes  from preschool  to  school-age
     skills, and the  use  of  extrapolated scores, this relationship suggests consid-
     erable reliablity for the McCarthy Scales in the follow-up study.
(e)  The same comments regarding quality assurance checks for data  entries  as were
     made under (e) above  for the  Peri no and Ernhart  (1974)  study also apply here.
                                    14

-------
3.   Statistical Analyses

     The  Committee  believes that  the treatment  of  confounding factors  can be  handled  in a
better way  than  as  described in the  1974  and 1981 Ernhart papers.  The  analyses performed by
Ernhart and  colleagues  can be questioned in light of currently accepted statistical practices
for dealing with interaction and confounding, and they should be reworked.
     The  proper  model  for assessing the effect of Pb exposure on IQ is one which contains, in
addition  to  the Pb  exposure variable,  all  factors  deemed to  be  potentially confounding.   A
potentially  confounding  factor  is  one which is correlated  (in  the population) with the expo-
sure variable  (Pb)  and  for which there  is  reasonable  evidence  from previous research experi-
ence and  knowledge  that  it is predictive  (in its  own  right)  of  the  outcome variable (IQ).
Thus,  a   procedure  which  chooses  potential  confounders  via a forward  selection  procedure,
ignoring  the Pb variable (as Ernhart did),  can be misleading and can actually incorrectly drop
important confounders from consideration.   A backward elimination strategy is designed to ob-
tain the  most  accurate  estimate of the  effects of  the exposure variable adjusted for all key
confounders, whereas a  forward  selection approach is designed to predict the outcome variable
but does  not assure an accurate estimate of the effects of the exposure variable.
     As an example,  parental  education was eliminated by a forward selection approach not in-
volving the  Pb exposure  variable  at all;   in  fact,  it is a potential  confounder,  and should
only be  eliminated  from  the full  model if its  elimination  does  not  alter the Pb  exposure
regression coefficient.   Similarly, SES should be handled in the same fashion.
     In theory, assuming  that  the  data set is free from bias (i.e., is a random sample of the
population under  study),  one can  lose  precision  but  not  validity by  including in  the  full
model  a true non-determinant of the  dependent variable  under  study.  An example in  the  data
under consideration is  the variable  sex, which is generally agreed to be a predictor of motor
skills but not of the  other outcome  measures  being  studied.   Its inclusion in the  full model
(using any outcome variable other than motor skills)  should not  cause a validity problem (even
though it is correlated with Pb exposure);  this  can be demonstrated (assuming  that  the  data
are representative  of  the  population)  by  dropping  the  sex .variable from the  full model  and
noting that  the  Pb  exposure coefficient does  not materially  change. With motor  skill  as  the
outcome variable, sex should be  expected to be manifest as a real  confounder in these  data and
hence  cannot be  dropped  from  the  model.  In summary,  then,  a  reanalysis  of the  Perino  and
Ernhart (1974) and Ernhart et al.  (1981) data should  be carried  out, based on a model  contain-
ing the Pb exposure  variable and all  available confounders measured.
     In general,  interaction effects between  the   Pb  exposure  variable and  the  covariates
should be assessed  before  confounding issues are  considered.  A qualitative assessment can be
                                              15

-------
done Initially by stratifying on each of the covariates and determining whether the relation-
ship between Pb exposure and IQ is reasonably constant across strata of the covariate under
consideration. Nonconsistency suggests the need for one or more interaction terras in the
model under study. Ultimately, the modeling of interaction involves the use of cross-product
terms in regression models. A strategy for dealing with interaction and confounding in obser-
vational epidemiologic data is described by Kleinbaum et al. (1982). Possible interaction
effects were not really examined by Ernhart and colleagues, and the presence of any inter-
action(s) would complicate data interpretation. However, given the highly variable nature of
the data and the limited sample size, it may be difficult to deal adequately with interaction
assessment for these data.

Comments on specific aspects of the statistical analyses employed in the 1974 and 1981
Ernhart studies include:

(a) Corrections for unreliability (i.e., measurement error) in the variables under
study should be made, especially concerning the Pb exposure indices. Such cor-
rections will adjust the observed correlations, and could have a major impact
on the ultimate conclusions drawn from the analyses. They provide a limit on
the strength of relationships by showing the relationship that would be expec-
ted were the measurements made and recorded without error.
(b) The outlier alluded to earlier in the Ernhart et al. (1981) data set should be
dropped from all analyses. The dentine Pb value provides enough evidence that,
in this case, the past Pb exposure was sufficiently high to conclude that the
subject may have earlier been overtly symptomatic but undiagnosed.
(c) It is important that the dependent variables be adjusted appropriately for age.
As a suggestion, one could use, for each child, the deviation from the linear
regression line of "raw score" on "age at testing".
(d) Dependent variable scores for a given individual on a series of intelligence
tests are obviously correlated. Multivariate analysis of covariance is one
option, but must be done very carefully. Certain assumptions (e.g., multivari-
ate normality) must hold at least approximately.
(e) Treating the Pb exposure variable as a continuous variable is preferable to
categorizing cases into high and low Pb groups.
(f) The Committee notes that, for the 62 children in the 1981 study (excluding the
outlier), significant correlations are seen between the identification number
16

-------
assigned to children and other variables of interest. Identification number
correlates -0.43 with 1972 level of blood Pb, 0.38 with parent IQ, 0.28 with
parent education, and 0.27 with 1972 GCI score from the McCarthy scale. These
results would be expected if identification numbers were assigned to children in
the order in which they were assessed (by Peri no in 1972) and if the children
with higher Pb levels tended to be the earlier ones assessed. Results of the
Ernhart studies, particularly the one by Perino and Ernhart (1974), could be
affected by this nonrandom ordering of assessments, to the extent that any
aspects of the assessment process systematically changed over the course of the
studies.

An alternative research question to those addressed in the Perino and Ernhart (1974) and
Ernhart et al. (1981) analyses can be asked by considering data from the cases assessed both
in 1972 and in 1977: namely, is there a relationship between differences in cognitive scores
and differences in blood lead concentrations from 1972 to 1977? Do children whose blood lead
levels were higher in 1977 than expected from their 1972 levels display cognitive scores in
1977 that differ from those expected from their 1972 cognitive scores?
This question could be addressed by regression analyses. The 1977 Pb level could be pre-
dicted from its regression on the 1972 Pb level and residual values obtained. Similarly, the
1977 IQ value could be predicted from its regression on the 1972 IQ value and residual values
obtained. The correlation between IQ residuals and Pb residuals would indicate an association
between change in IQ and change in Pb. That correlation also could be adjusted for confound-
ers such as parental IQ, parental education, and SES. We would urge such a reanalysis of the
data from Perino and Ernhart and Ernhart et al. (after correction of the erroneous values
recently discovered in those data). Interpretation of results from such analyses must depend
upon careful inspection of the cross-lagged correlations on which the correlation of residuals
depends. In addition, of course, even the residual values may be acting as surrogates for un-
recognized confounding variables.

4. Committee Conclusions and Recommendations

The Committee's conclusions and recommendations regarding the Perino and Ernhart (1974)
study can be summarized as follows. Blood lead levels, as the main index of exposure, appear
to be of acceptable reliability. The psychometric measures for children are also acceptable,
but confounding variables may not have been adequately measured. The statistical analyses do
not adequately deal with confounding variables. In the view of the Committee, the findings of
17

-------
this study, as reported, neither support nor refute the hypothesis that low to moderate lead
exposures are associated with cognitive impairments in apparently asymptomatic children. The
Committee recommends that the data from the Perino and Ernhart study be used in conjunction
with the 1981 Ernhart et al. follow-up study in longitudinal analyses.
The Committee's conclusions and recommendations concerning the Ernhart et al. (1981)
study include the following points. Given limitations in both the control and the outcome
measures, it is difficult to assess the possible role of less-than-ideal measures in contribu-
ting to the generally null results reported. When the authors conclude that there are no sig-
nificant effects, or very weak effects at best, then that outcome might also be reasonably
attributable to unreliable measures or other procedural problems. One major difficulty with
this study is the potential unreliability of the blood Pb level measurements, such that the
Committee recommends that the blood Pb values be corrected in the fashion specified earlier.
The psychometric data were adequately collected but should be readjusted for age. The cross-
sectional analyses suffer from the same problems as those of the previous (Perino and Ernhart,
1974) study. More importantly, the analyses failed to exploit fully the longitudinal aspects
of the study data set. In the view of the Committee, then, the findings of this study as
reported in the published form also neither support nor refute the hypothesis that the report-
ed blood lead levels are associated with cognitive impairments in children. The Committee
strongly recommends that longitudinal analyses of these data (from both Perino and Ernhart,
1974, and Ernhart et al., 1981) be carried out.

C. Comments on Yamins (1976) Dissertation Study

1. Indicators Of Lead Exposure

Comments on specific aspects of the Pb exposure measurement methodology used in the
Yamins (1976) dissertation study are as follows:

(a) Blood Pb was sampled by finger puncture, using established techniques for blood
lead sampling, and the blood samples collected on filter paper presumably pro-
vided by the New York City Health Department. Upon collection, samples were
transported to the New York City Health Department for lead measurement.
(b) The use of filter paper for blood collection raises the same question that is
of concern in the Ernhart et al. follow-up study, i.e., the blood lead value
must be corrected for hematocrit. The problem here occurs, actually, with each
of the three indices (highest blood lead, mean blood lead, or most recent blood
lead), so that it is not possible to assess the actual blood lead level in each
18

-------
case nor to determine the suitability of selecting one exposure index over
another, since a given subject may have had variable hematocrit over the multi-
sampling time period.
(c) Measurement of EP was also carried out in the laboratories of the New York City
Health Department, quality control details for which were discussed among com-
ments on the studies of Ernhart and coworkers (vide supra).

2. Psychometric Measurements and Procedures

In the Yamins dissertation, a variety of psychometric measurement procedures were employ-
ed, including some standardized instruments such as the Caldwell Preschool Inventory and
other, more-or-less experimental measures such as the verbal repetition tasks.
Specific comments on the psychometric measurements utilized by Yamins (1976) include:

(a) This study employed the Peabody Picture Vocabulary Test (PPVT) to assess the
intellectual development of the children. The PPVT is not a nonverbal measure
of intelligence; it provides a narrow assessment of verbal abilities. However,
the committee agreed that the PPVT was a reasonable measure of cognitive per-
formance to use in the study because the scores correlated well with the other
cognitive and experimental language measures, including the Caldwell Preschool
Inventory, which is an appropriate measure for this population.
(b) The Ammons Quick Test was used to assess parental IQ in this study. As noted
earlier, this measure has questionable validity. However, the pattern of cor-
relations between the Quick Test scores and other variables does establish some
credibility for its use in the current study.
(c) The experimental measures of language skill correlated with child's age, as one
would predict, and yet appear to measure skills already tapped by the cognitive
measures.
(d) There is no obvious explanation for the correlation of 0.30 between parent
Quick Test scores and child's age. However, when age is partial led out, the
correlation between child IQ and parental IQ is approximately 0.35, a value
close to that found in other studies.
(e) The author (Yamins) administered all of the cognitive and language measures in
the children's homes and was blind as to Pb levels at the time. She appears to
have been appropriately trained and competent to collect the psychometric test
data. Quality assurance checks regarding data collection (e.g., scoring,
coding, and keypunching) could not be ascertained but are assumed to be the
same as for the above Ernhart studies.
19

-------
3.   Statistical Analyses

     Many of the  same reservations expressed earlier regarding  the  analyses used in the 1974
and 1981 Ernhart studies also apply here.  Specific concerns include the following:

          (a)  Stepwise multiple regression was employed to choose the set of covariates to be
               included  in  the final regression  model  with mean lead  level.   In the Commit-
               tee's view,  this  is not the appropriate way to deal with potentially confound-
               ing factors.  A  backward elimination strategy starting with a model  containing
               lead and  all potential  confounders  is  recommended since  confounding involves
               association with both the outcome variable (e.g.,  a measure of learning perfor-
               mance) and the  exposure  variable (e.g.,  mean lead level).   A forward selection
               approach, as  was apparently  employed  by Yamins,   ignores  the relationship be-
               tween the potential confounders and the exposure variable in choosing an appro-
               priate  subset for  control,  and  hence  can  lead  to  inappropriate  adjustment.
          (b)  The strategy  for  analysis described  on page 79 of the Yamins (1976) disserta-
               tion is not appropriate for valid control of confounding effects (see preceding
               comment).   Although it may produce the same results  as  a  backward elimination
               approach, one cannot  know without trying both approaches.   In this case, back-
               ward elimination of variables does not  markedly alter the outcome of  the analy-
               sis.   One  Committee member (LH)  found that the association  between  Pb  and IQ
               remained after  controlling for father's  absence,  parental  IQ, parental  educa-
               tion,  birth  order,  and  birth weight,  using  a backward elimination  approach.
          (c)  The results displayed  in Table 9 (page 82)  of  the dissertation do not,  in the
               Committee's opinion, represent a  strong  indictment of lead exposure.  Finding a
               few significant  partial  correlations of lead exposure  with  various  dependent
               variables just by chance is not at all  unlikely when  performing several  analy-
               ses using mutually  correlated dependent  variables.  Apparently,  no multivariate
               (as opposed to multivariable)  analyses  were employed to account  for such corre-
               lations.
          (d)  The results in  Table  10 (page 85) of  the dissertation are based on  a compari-
               son of a  "low lead" group to a "moderate lead" group  (after dichotomizing lead
               exposure), with  adjustment only  for  the covariate  "age".   Given that  other
               potential  confounders were  apparently  ignored, and given that  several  compar-
               isons  were  made  involving correlated responses, not much importance can  be
               attached to the few significant findings  reported.

                                              20

-------
          (e)  Results  presented  in Tables 11  (page  86),  12 (page 88), 14  (page  91),  and 15
               (page 93)  of  the dissertation are also  based on controlling only for age. The
               Committee finds  it difficult to place much importance on the findings presented
               in these tables.

4.   Committee Conclusions and  Recommendations

     The Committee  concludes  that,  despite reservations expressed regarding psychometric mea-
surements employed  in  the Yamins study, the pattern of results obtained (including the inter-
correlations  between different measurement outcomes and parental  measures)  lends  credibility
to  the  psychometric results as reasonably reflective of the cognitive abilities  of  the sub-
jects tested.   The blood  lead data  require  correction for hematocrit because of  the  use of
filter paper  for  blood collection;  unfortunately, the  hematocrit  values  are not available to
make this  correction.   In addition,  the  Committee  finds  that  specific  statistical  analyses
employed (including stepwise  regressions and covan'ate analyses controlling only for age) are
not  the  most appropriate  for  analyzing the  Yamins data set.   Rather, multivariate  analyses
should have been  used  that included other  potential  confounders besides  age and,  also, back-
ward elimination  of variables  having negligible impact on the variance attributed  to Pb.  The
Committee further finds that the very  modest  residual  effects attributed to Pb based  on the
reported analyses controlling  only  for age are not  convincing evidence for a negative effect
of Pb on the cognitive abilities of the subject children.
                                              21

-------
REVIEW OF STUDIES BY DR. HERBERT NEEDLEHAN AND COLLEAGUES

A. Background Information

The Committee undertook detailed review of an epidemiclogical study published in 1979 by
Dr. Herbert Needleman and associates (Needleman et al., 1979). In addition, limited review
was carried out for several other reports (Burchfiel et al., 1980; Needleman, 1981; Needleman,
1982; Needleman, 1983; Bellinger and Needleman, 1983) published as follow-up analyses of the
same data set and/or new data constituting extensions of the 1979 study.
Approximately 3329 children attending first and second grades between 1975 and 1978 in
Chelsea and Somerville, Massachusetts, constituted the study population in the Needleman et
al. (1979) study. Children submitted shed teeth to their teacher, who verified the presence
of a fresh socket. The shed deciduous teeth were cleaned ultrasonically (discarding any con-
taining fillings), followed by dissection of a 1-mm central slice and subsequent analysis of
Pb in dentine tissue by anodic stripping voltammetry. Teeth were donated from 70% of the
population sampled. Almost all children who donated teeth (2146) were rated by their
teachers on an eleven-item classroom behavior scale. The results obtained for the rated chil-
dren were reported to demonstrate a dose-response relationship between increasing dentine Pb
levels and increasing percentages of students receiving negative (poorer) ratings on several
of the 11 categories of classroom behavior, as shown in figure 1 below.
Following the teachers1 ratings of classroom behavior, subsets of the rated students
(reported to represent polar groups of children with the lowest and highest 10 percent of den-
tine Pb levels) were recruited for further, extensive neuropsychological evaluation by means
of psychometric tests. Each subject whose initial tooth slice was in the highest 10th percen-
tile (>24 ppm) or lowest 10th percentile (<6 ppm) was provisionally classified as having high
or low lead levels, respectively. Repeat dentine lead samples from the same teeth were analy-
zed, when possible, and attempts were made to obtain and analyze other shed teeth from each
subject provisionally classified in either lead exposure group (with more than one analysis
being obtained for all but one subject). Parents of children provisionally classified as
having either high or low dentine Pb levels were invited to have their children participate
in further neuropsychological evaluations. Criteria were established for requisite agreement
between replicate dentine sample analyses before the data for a given subject were included in
the study; when requisite agreement was not found, then the subject was designated as "unclas-
sified" and excluded from data analyses. Other children were excluded from the study because:
(1) their parents were unable or unwilling to participate; (2) they came from bilingual homes;
22

-------

40

cc
UJ
3 30
>
m
a
|S 20
27.0
•

1
M


.
Ill
™
p




r\


1






__
,


ill
t


_



'



•mm


—





CLASS
           123456
                    123456    123456   123456   1234C6   1234S6    123456   123456    123456    123456
                                                                                                      123456
         DISTRACT-    NOT   DEPENDENT    NOT     HYPER-  IMPULSIVE   FRUS-     DAY     SIMPLE SEQUENCES   LOW
            IBLE   PERSISTENT         ORGANIZED  ACTIVE             TRATED  DREAMER   DIREC-            OVERALL
                                                                                    TIONS              FUNC-
                                                                                   UlNABLE TO TOLLOwJ TKJNING

           Figure 1. Distribution of negative ratings by teachers on eleven classroom behaviors in relation
           to dentine lead concentrations.
           Source: Needleman et  al. 11979).

-------
(3) they had been diagnosed as having been lead poisoned; or (4) their medical history indi-
cated a birth weight of <2500 g, delay in discharge beyond mother's discharge from hospital
after birth, or a record of noteworthy head injury (any of which can correlate with slower
neurobehavioral development). Table 1 from the Needleman et al. (1979) paper, reproduced
below, lists the number of provisionally eligible children, those excluded from neuropsycho-
logic testing, and those undergoing neuropsychologic testing who were retained or excluded
from data analyses.

TABLE 1. REASONS FOR EXCLUDING SUBJECTS AND DISTRIBUTION OF
FINAL DENTINE LEAD LEVELS IN INCLUDED AND EXCLUDED GROUPS
GROUP
Provisionally eligible subjects:
Excluded from neuropsychologic testing:
Bilingual home
Not interested
Moved'
Othert
Total
Subjects tested
Excluded from data analysis:
Later tooth discordant ...
Not discharged from nursery with
mother, possible head injury,
reported to have plumbism or
bilingual home
Total
NO. DENTINE LEAD LEVEL
LOW HIGH UNCLASSIFIED
524 258 187
254* 123 101
84
57
19
94
254
2701 135 86
112T 35 28
36
76
m
79
30

49
49

Cases scored and data analyzed 158 100 58

"Teachers' behavioral assessment available on 235.
tlnfant at home, two working parents, etc.
ITeachers' behavioral assessment available on 253.
Source: Needleman et al. (1979)

Mean dentine lead values for the 100 children included in the low-Pb and 58 in the high-
Pb exposure groups were not reported by Needleman et al. (1979). However, Bellinger and
Needleman (1983), who studied 141 of the 158 subjects of Needleman et al. (1979), reported
those subjects to have mean dentine Pb levels of 6.2 ppm and 31.4 ppm, respectively. Mean
blood Pb levels reported as having been assayed 4-5 years earlier for approximately 50% of
the children in these two groups were 23.8 ± 6.0 (jg/di vs. 35.5 ± 10.1 pg/d£, respectively;

-------
the  highest  blood-Pb  level  recorded was  54  ug/d£.   The  low-Pb and  high-Pb group children
underwent   a  comprehensive   neuropsychologic   evaluation,   beginning  with   the  Wechsler
Intelligence Scale  for Children-Revised (WISC-R), with the examiners blind to the Pb-exposure
status  of  the children.   In addition  to  the WISC-R, the children  were  administered,  in set
sequence, tests  of:   concrete operational  intelligence; academic achievement (in mathematics,
reading  recognition,  and  reading comprehension);  auditory  and  language  processing;  visual-
motor reflexes;  attentional  performance;  and motor  coordination.  While  each child was being
tested,  the  parents filled out a comprehensive medical and social history, received a 58-item
questionnaire on parent attitudes, and took the Peabody Picture Vocabulary Test (PPVT).   Also,
39 non-lead  variables  potentially affecting the children's development were scaled and coded,
e.g., estimation of parental  socioeconomic  status  (SES) by  a  two-factor Hollingshead index.
     The scores  of  the high-Pb and  low-Pb children  for each of 39 control variables were com-
pared statistically by the Student t-Test,  with the  two groups differing significantly on such
variables as age,  father's social class and father's education.  Scores from the neuropsycho-
logic evaluations of the  high-Pb and low-Pb  children were  then compared statistically, using
an analysis  of covariance  with dentine-Pb level as  the main independent variable and with the
following five covariates:  father's SES (composed of education and occupation score); mother's
age  at  subject's birth;  number  of  pregnancies;  mother's education;  and parental  IQ  score.
With the exception  of these variables and age,  the low-Pb  and high-Pb groups were similar in
regard to most of the non-Pb control factors.
     Results of the neuropsychologic evaluations for the low-Pb and high-Pb groups can be sum-
marized  as  follows:  children in  the  high-Pb group were reported to  have  performed signifi-
cantly less well  on the WISC-R (especially on the verbal items), on three measures of auditory
and visual processing,  on  attentional  performance as measured  by reaction time under varying
delay conditions, and on most items of the  teachers' behavioral  ratings.  The high-Pb children
appeared to be particularly less competent in areas of verbal performance and auditory proces-
sing, having  obtained lower scores, for example,  on tasks  requiring:  response  to verbal  in-
structions of increasing complexity, immediate repetition of previously  uttered  sentences  of
increasing complexity, and discrimination of tone sequences  of increasing complexity as either
alike or  different. Impaired focusing  of  attention (or distraction) of  high-Pb  children  was
also reflected by  a  significantly higher percentage of high-Pb children  rating items being
found to be  significantly  different (i.e., more negative) for high-Pb than low-Pb children at
p <0.05.  Overall, the sum score (mean) of  ratings of classroom behavior were found to be sig-
nificantly poorer for the high-Pb children  based on an analysis of covariance.
     Burchfiel  et al.  (1980),  using computer-assisted spectral analysis  of recordings  from a
standard EEC examination on  41  (22  low-Pb and  19  high-Pb)  children  from the Needleman et al.

                                              25

-------
(1979)  study,  reported significant  increases in percentages of  low frequency delta activity
and decreases in percentages of alpha activity in the spontaneous EEG of the'high-Pb children.
Percentages of alpha and delta frequency EEG activity and results for several psychometric and
behavioral  testing  variables (e.g., WISC-R  full-scale  IQ and verbal  IQ,  reaction time under
varying delay, etc.)  obtained  for the same children were then employed as input variables (or
"features") in direct  and stepwise discriminant analyses. The  separation determined by these
analyses for combined psychological and EEG variables (p<0.005) was strikingly better than the
separation  of low-Pb  from  high-Pb  children  using either  psychological  (p  <0.041)  or EEG
(p<0.079) variables alone.
     A more recent  paper  by Needleman (1982) provided a summary overview of findings from the
Needleman et  al.  (1979)  study and findings reported by Burchfiel et al. (1980) concerning EEG
patterns  for  a  small  subset  of  the  children included in the  1979  study.   Needleman (1982)
also summarized results of an additional analysis of the Meedleman et al.  (1979) data reported
elsewhere by  Needleman  et al.  (1982).   More  specifically, cumulative  frequency distributions
of verbal  IQ  scores for the low-Pb and  high-Pb  subjects from the 1979 study were reported by
Needleman et  al.  (1982)  and reprinted as  Figure  2 of the  Needleman  (1982)  paper,  as shown
below in  Figure 2.   One key point made by Needleman (1982) was that the average IQ deficit of
four points demonstrated  by the Needleman et  al.  (1979)  study reflected not just further im-
pairment of cognitive  abilities  of children with already  low  IQs but rather a shift downward
in the entire  distribution of  IQ scores across  all  IQ  levels in the high-Pb group, with none
of the children in that group having verbal IQs over 125.
     The Bellinger and Needleman (1983) paper provided still  further follow-up analyses of the
Needleman et  al.  (1979)  data  set, focusing  mainly on  comparison  of  the  low-Pb  and high-Pb
children's observed IQs versus  their  expected IQs  based on their mothers' IQs.  Bellinger and
Needleman  reported  that  regression analyses  showed that  the IQs  of  children with  elevated
levels of dentine Pb  (>20 ppm)  fell  below those expected based on their mothers'  IQs and that
the amount by which a child's IQ falls  below the expected value increases with increasing den-
tine Pb levels in  a nonlinear fashion (see Figure  3 below,  showing plots  of  IQ  residuals by
dentine Pb  levels as  illustrated in Figure 2 of the Bellinger and Needleman paper).  In fact,
examination of the  scatterplot  shown  in Figure 3  and  the discussion  of  results  provided by
Bellinger and  Needleman (1983)  indicate  that regressions for the 20-29.9 ppm group  did not
reveal  significant  associations  between  increasing  Pb  levels in  that  range and IQ residuals,
in contrast to statistically significant  (p<0.05)  correlations found between IQ residuals and
dentine Pb in the  30-39.9  ppm range.
     In order to evaluate  critically  the above studies, the  Committee  met with Dr. Needleman
at his University of  Pittsburgh  (Children's Hospital)  office facilities in Pittsburgh, PA, on
                                              26

-------
2

m
>

i
3
u
                                                                             20      30      40      50

                                                                             DENTINE LEAD LEVEL ppm
                                                                                                                 70
      50   60
                70
                     80    90   100   110   120  130

                        VERBAL I.Q.
       Figure 2. Cumulative frequency distributions of
       verbal IQ scores in high and low lead subjects.

       Source: Needleman (1982) and Needleman et al.
       (1982).
Figure 3. Scatterplot of children's dentine lead level versus
residual. Regression Hnes are shown for four ranges of
lead level: low lead. 0.9 to 9.9 ppm; elevated lead, 20.0 to
66.0, 20.0 to 29.9. and 30.0 to 39.9 ppm.

Source: Bellinger and Needleman (1983).

-------
March 30-31, 1983. During that meeting, Dr. Needleman presented an overview focusing mainly
on the objectives, design, data collection and statistical analysis procedures, and findings
for the original study reported by Needleman et al. (1979). Dr. N«edlemart also provided addi-
tional information regarding follow-up analyses or extensions of the 1979 study either pub-
lished in other papers referred to above or expected to be published in the near future. This
additional information included comments regarding the conduct of a separate study involving
the evaluation of teachers' ratings of classroom behavior of children in Lowell, MA (a dif-
ferent population from the one sampled in the 1979 study). Certain listings of raw data
(provided in coded form to protect the privacy of subjects), computer printouts summarizing
data entries for statistical analyses or results of such analyses, and miscellaneous other
pertinent materials were discussed with Dr. Needleman during the March 30-31 meeting. Addi-
tional information was requested by the Committee in order to clarify factual points or to
help resolve evaluative issues arising from the discussions in March with Dr. Needleman. A
portion of the information was provided during the 2-3 months following the March meeting.
(See Attachment 1 for a list of materials examined by the Committee in connection with the re-
view of the subject studies.) The Committee's comments regarding the most salient points of
concern and controversy related to methodological and other features of the above studies by
Needleman and colleagues are presented below.

B. Comments on Needleman et al. (1979) Study

1. Indicators of Exposure

In the principal study (Needleman et al. , 1979) as well as ir» subsequent reports on sub-
sets of subjects from the initial population (e.g., Burchfiel et al. , 1980; Bellinger and
Needleman, 1983), Pb exposure in the pediatric subjects was assessed by analysis of Pb in the
dentine of deciduous teeth. In contrast to blood Pb, which is an exposure marker for rela-
tively recent exposure, whole-tooth or tooth*-region analysis for Pb content yields an index of
cumulative Pb exposure of the subject up to the time of exfoliation.
In the report of Needleman et al. (1979), blood Pb levels as an additional index of prior
exposure were reported as only being available for some (approximately 50%) of the subjects in
the highest/lowest deciles, and were discussed only in terms of group means. These measure-
ments were reported to have been obtained as part of a blood screening program in the subject
communities 4-5 years prior to collection of tooth samples.
Observations and comments concerning specific aspects of the Pb exposure indices and
associated methodological procedures include:

-------
(a)  Dentine  was  isolated  from  each tooth  sample  by a  procedure described  in an
     earlier  report  in  which  the present principal  investigator (Dr.  Needleman) was
     also heavily involved (Shapiro et al., 1973). In that procedure,  very thin sec-
     tions  of tooth  were carefully  cut from the  central  sagittal  plane,  dentine
     (coronal plus secondary)  was  mechanically separated from enamel  and circumpul-
     pal  dentine,  and the  dentine samples were dissolved in  acid and subsequently
     analyzed by anodic stripping voltammetry (ASV)  for Pb content.
(b)  According to Dr. Needleman,  the type of tooth  selected for analysis was fairly
     consistent:   mainly  the   incisor,  and some bicuspids.   Hence, it appears that
     any variation in Pb  content which might arise  from random selection of diverse
     types  of dentition (due  to variation in Pb content across different  types of
     teeth) would have been minimal.
(c)  Unlike  blood  Pb,  there  is  no external  quality control  framework  by  which to
     evaluate the dentine  Pb  analyses such as were  done in  the subject studies.  One
     must therefore  consider   the  specific steps in the analysis  against  a general
     body of  information.   Two steps  in the  dentine  Pb measurement need to be con-
     sidered.  According  to Needleman,  the  homogeneity of  dentine in  terms  of Pb
     content  for a given  tooth can vary sufficiently that tooth sectioning was con-
     fined  to an  initial  central  sagittal sectioning  in  all  cases,  the  sectioned
     sample providing two  (replicate)  samples for analysis.    Once  dentine  was iso-
     lated,   its  subsequent analysis  by ASV  would  be expected to  be  achieved with
     good accuracy and  precision, given available  data for overall  performance of
     ASV  assays of Pb  in biological  matrices  and the fact  that such  Pb  levels  are
     relatively high.   Since  the  major  determinant of  variance  in  the  replicate
     (single  tooth)/duplicate  (multiple teeth)  analyses  was  the  dentine  isolation
     step,  Needleman et al.  (1979)  attempted to minimize unacceptable  variance by
     use  of  intra-sample  concordance  criteria  in  analyzing  relationships  between
     dentine Pb levels  and the results of the psychometric test battery.
(d)  The impression gained from close reading of the Needleman arid  other reports, as
     well as  discussions  with Needleman,  is  that use of dentine  Pb  values  entails
     methodological  skill  at  the  step  of dentine isolation.   From the  information
     available to the Committee  as to actual  variation in dentine  Pb  across a given
     tooth sample,  it appears  that ±15% represents a reasonable specific estimate of
     variability  for the  dentine  analysis  for subjects from the low-Pb  and high-Pb
     groups  included in   statistical  analyses  of   neuropsychologic   test  outcomes
                                    29

-------
(i.e., for subjects with the greatest concordance among their dentine Pb
values). Much greater variation existed among replicate or duplicate dentine
Pb values for individuals from the low-Pb and high-Pb groups excluded from the
statistical analyses. Examination of replicate/duplicate values for measure-
ments for all subjects in the study (including the 2000+ students for which
teachers' ratings were obtained) would be necessary to determine an overall
estimate of variance for dentine Pb measurements in the study.
(e) Whole-tooth analysis would be simpler technically and has been more often
employed than specific tooth-region analysis. However, one can also expect
that such a measure would be less sensitive as a biological index of Pb expo-
sure due to the inclusion of enamel, a region that contributes significiantly
to tooth mass but has relatively invariant low Pb content regardless of expo-
sure. In the case of whole-tooth Pb, it is known that Pb content is linearly
related to age of the subject and that the values of the slopes of Pb content
vs. age are better indicators of Pb exposure than just the Pb concentrations
alone (Steenhout and Pourtois, 1981). Shapiro and coworkers (1978) have also
reported that there is a better correlation of tooth Pb concentration/year with
either blood Pb or erythrocyte porphyrin than just Pb concentration unadjusted
for age. Expression of dentine Pb as a function of age in the present study
(M9 Pb • g dentine • yr ) would be desirable, especially because the mean
age of the high-Pb subjects was greater than that of the low exposure group.
While this method of expressing Pb exposure would have minimal effect on the
categorization of subjects into high- and low-Pb groups, it might be expected
to influence relationships between Pb and other variables, e.g., in Figure 3
(above) from Bellinger and Needleman (1983).
(f) The relative quality of the earlier blood Pb determinations for some low-Pb and
high-Pb subjects cannot be ascertained and must be considered more suspect than
the main exposure measure used (i.e., dentine Pb). At the time blood Pb levels
were measured in the subjects, the quality control for the community Pb-screen-
ing programs was minimal, with sampling being done by finger puncture and trans-
fer to capillary tube (communication of V. N. Houk to H. L. Needleman, see
Needleman, November 22, 1982: letter to L. D. Grant). These limitations on
the relative reliability of such measurements apparently were the reason for
Dr. Needleman's discussion of these values only in terms of group means for the
low-Pb and high-Pb subjects. Given present knowledge about the impact of sam-
pling protocols on the accuracy of blood Pb measurements, one can reasonably
say that finger puncture plus capillary tube versus venous puncture plus low-Pb
30

-------
blood tube would impart a significant positive bias to the blood Pb levels ob-
tained. Hence, the overall means reported for the low-Pb and high-Pb groups in
terms of blood Pb would, if anything, likely be higher than their true values
for the 50% of the low-Pb and high-Pb children sampled.
(g) Apart from the issue of reliability of the blood Pb measures under considera-
tion is the question of age of the children at the time of blood sampling rela-
tive to the known variation of blood Pb with age in children for a given expo-
sure setting: i.e., blood Pb levels in children generally tend to peak at 2-3
years of age and decline in subsequent years. Available information on the ages
of the children at the time of psychometric testing, the years when such test-
ing occurred, and the years when blood Pb measurements were made indicate that
the ages of some children at the time of blood Pb measurement were probably at
or not materially beyond the period of typical peaking in blood Pb, but others
may have been sampled at a later age within 1-2 years (during 1973-74) prior to
their participation in neuropsychologic testing while in first or second grade
(as early as during 1975-76). It would be necessary to know the ages of speci-
fic subjects when the blood lead determinations were done and their age at neu-
ropsychologic testing before reliable judgments could be made regarding the
representativeness of the reported mean blood Pb values for either the low-Pb
or high-Pb subjects.

2. Psychometric Measurements and Procedures

The study employed a comprehensive neuropsychological battery to assess the children's
behavioral functioning. The measures included the WISC-R, Piagetian tasks, and selected tests
of academic achievement, auditory and language processing, visual and motor performance, reac-
tion time, motor coordination, and teacher ratings of classroom behavior. Mothers' attitudes
toward child rearing and parental IQ (indexed by the PPVT) were also assessed. The PPVT is a
narrow assessment of mothers' intelligence, but their PPVT scores correlated in expected ways
with other variables in the study.
Dr. Needleman administered the PPVT to the mothers, and three other examiners adminis-
tered the WISC-R and other assessments of the children in a fixed order. The examiners were
blind as to children's Pb levels and scored the test immediately after the test session. It
is not known how qualified the examiners were to administer individual tests, but Dr. Needle-
man reported that the examiners were instructed on how to administer and score the test.
In a recent publication (Needleman, 1983, p. 243) additional details of the psychometric
procedures were reported. Children with low-Pb exposure were scheduled early in the study,
31

-------
because "I wanted my technicians to get some experience with normal children." In addition,
Dr. Needleman told the Committee during the meeting in Pittsburgh that the study began with
three technicians, one of whom left during the study and was replaced by a fourth tester.
Results of the Needleman et al. (1979) and related publications using the psychometric test
scores could be affected by this nonrandotn ordering of assessments.
Dr. Needleman reported that quality assurance procedures for ensuring the accuracy of
teachers' ratings and neuropsychologic test results used in statistical analyses included:
(1) summing of teacher ratings across items and entry of scores for each item and sum scores
onto computer cards, followed by verification and transfer onto magnetic tapes; (2) checking
of neuropsychologic test scores by a second examiner other than the one doing the original
scoring, followed by entry onto cards, verification, and transfer to magnetic tapes; (3) sub-
sequent 5% sampling of computer tape entries to check accuracy against original data listings,
with 12 errors in 15,600 columns of entries being found and corrected.
The Committee's inspection of raw data during the March visit with Dr. Needleman revealed
some problems of another kind, however. Printouts of parental IQ data for Tow- and high-Pb
subjects included in the statistical analyses (e.g., analyses of covariance) published in the
1979 article revealed errors in calculating parental IQ values for some subjects when their
fathers and mothers were both administered the Peabody Picture Vocabulary Test. Instead of an
average of mothers' and fathers' scores (midparent IQ), the parents' IQ scores were evidently
combined by taking one-half of one parent's score and adding that value to the other parent's
score. This erroneous procedure resulted in some parental IQ values that lie well outside
probable values. These errors were confirmed later by Or. Needleman in his letter of
October 4, 1983 to Dr. Bernard Goldstein. The impact of this is to introduce error into the
results of all of the published analyses of the data set where parental IQ was included as a
variable. Correcting these errors would alter the results of the analyses. The precise in-
fluence of the errors on the results can be determined only by reanalyzing the data, and the
Committee urges that this be done.
During inspection of raw data, the Committee also noted a seemingly higher proportion of
large discrepancies between the children's WISC-R Verbal and Performance Scale IQ scores than
would be expected in an unselected sample. The discrepancies seemed to be distributed across
the high-Pb and low-Pb groups. Neither sufficient time nor facilities were available during
the data inspection to carry out an adequate quantitative analysis of the relationship of the
verbal to performance IQs, but if the discrepancies are as large and/or numerous as they ap-
peared to be, this may raise questions about the validity of the WISC-R assessment as employed
in this study.
32

-------
3.    Statistical Analyses

     Comments on specific aspects of the statistical analyses employed in the Needleman et al.
(1979) study include the following:

          (a)  The  statistical  analyses  for  teachers'  rating  scores for classroom  behavior
               were based on  classification  of the children's Pb exposure  levels  in terms of
               first dentine lead values obtained for the first tooth submitted by each of the
               2146 subjects  included  in  the  analyses  (vide supra).  Six lead exposure cate-
               gories were  defined  as  indicated  in Figure 1 (i.e.,  <5.1, 5.1-8.1,  8.2-11.8,
               11.9-17.1, 17.2-27.0, and  >27.0  ppm dentine Pb), with  group  boundaries chosen
               to give symmetrical cell sizes around the median (i.e., 6.8% in Groups 1 and 6,
               17.6% in Groups 2 and 5, and 25.6% in Group 4, respectively).   However, no sta-
               tistical  analyses  that take into  account other potentially  confounding vari-
               ables were done  on the teachers' rating data  shown  in Figure 1 and,  thus, the
               dose-response  relationships  shown  in  that  figure  cannot  be  attributed to  Pb
               exposure alone.
                    In addition, questions arise  regarding the appropriateness or accuracy of
               classification of subjects in terms of the narrow dentine Pb ranges  employed in
               plotting the dose-response data  shown in Figure 1.   Given  the 15% variability
               noted for replicate  analyses of  teeth for those subjects  with the  most highly
               concordant dentine Pb values,  many subjects who were  included in one or another
               of the six exposure categories  based on first dentine Pb analyses could be more
               appropriately classified as belonging in a different  exposure category, accord-
               ing  to  later  replicate/duplicate  dentine  Pb  values.   This is  particularly
               likely if the  same or analogous concordance criteria  used  by Or.  Needleman to
               select low-Pb  and high-Pb subjects  for inclusion in statistical  analyses  of
               later psychometric test scores  were  used for the  teachers'   rating  analyses.
               Inspection of  raw dentine Pb  values for subjects provisionally  classified  as
               low-Pb or  high-Pb subjects for  psychometric testing,  but then excluded  from
               final statistical  analyses of the  psychometric test  results because  of  non-
               concordance of dentine  Pb  values,  revealed that shifts across the six exposure
               categories could  be  substantial if  replicate or duplicate  dentine Pb values
               beyond the first dentine Pb value were taken into account.
          (b)  In regard to the  statistical  analysis of results from the subsequent psychome-
               tric testing  phase  of  the study,  comparisions  were  made only between  those

                                              33

-------
     children reported  to be ranked  in  the highest 10th percent!le  for  dentine Pb
     concentrations and those in the lowest 10th percent!le.   This strategy certain-
     ly maximizes the  chances  for finding significant differences.  Reviewing Figure
     1 of the 1979 report, the Committee notes that a group with "moderate" exposure
     might serve  to provide evidence  for a  dose-response  relationship  (which,  if
     found, would argue more strongly in favor of a causal  connection than the polar
     low-Pb vs.  high-Pb group  comparision used).   An earlier  report  on the subject
     (Needleman, 1977) suggests an intention to use low, medium, and high  dentine Pb
     groups,  and a  very recent  report (Needleman, 1983; Table 6;  p.  237) does show
     some results on  behavior  ratings  for a group of 13 subjects with "middle" den-
     tine Pb levels.   Psychological  test scores have evidently  been  obtained for a
     middle group of  subjects  (Needleman, 1983, p. 242).  The Committee  recommends
     that analyses be undertaken to evaluate any available  psychometric testing data
     for "intermediate" lead-exposure subjects.
(c)  Many  questions  about  sampling procedures  arise from  the exclusion of  large
     numbers  of potential  participants  1n  the psychometric testing  phase of  this
     study.   From 542 provisionally eligible participants,  almost half were excluded
     from neuropsychological assessment,  and  41 percent of those tested  were  later
     excluded from  data  analyses.  Although  reasons  for the  exclusions  were  given
     (see Table 1 of  the  1979  article),  the distribution of  demographic and psycho-
     logical  outcome  measures  for  those excluded from the low- and  high-Pb  groups
     was provided  neither in the  published article  nor by the .investigator  to the
     Committee.   The Committee  could not fully evaluate sources of possible bias due
     to such exclusions  in the selection of  the  sample reported in  this  paper and
     other publications reviewed.
          Some  of the  criteria  used to define Pb  exposure  levels or to exclude sub-
     jects from statistical  analyses  seemed arbitrary, and different results  might
     have been obtained with application of equally  good or  better alternative cri-<
     teria for  classification  of  Pb exposure levels  and groups. For example,  some
     subjects provisionally classified as  low-Pb  or  high-Pb subjects  based on ini-
     tial  dentine Pb values  were  excluded from final  data  analyses  based  on discor-
     dances arising  from  later replicate  or  duplicate  Pb  values obtained for the
     same or different teeth, although certain key "discordant"  dentine  Pb  values
     were not meaningfully different from the cut-off criteria levels for inclusion
     as low-Pb or  high-Pb  subjects. Thus, exclusion of some  subjects  from  the  low-Pb
     group for statistical analyses hinged  on a single dentine Pb  value (e.g.,  10.1
                                    34

-------
     or  10.5 ppm)  barely exceeding  the 10  ppm  criterion ultimately  selected and
     rigidly enforced as  defining the low-Pb (or lowest decile) exposure group, al-
     though such dentine Pb values were as likely to have true readings below 10 ppm
     as were certain key values (e.g., 9.5 or 9.8 ppm) for some subjects included in
     the low-Pb group likely to have true readings above 10 ppm.  Also, since inclu-
     sion or exclusion  of subjects in the statistical analyses was based on dentine
     Pb  values  for all  teeth submitted  by  a given subject over  the  course of the
     study, some subjects may have been classified as high-Pb or low-Pb children (or
     excluded from analysis)  based on replicate or duplicate  dentine  Pb values ob-
     tained  for  teeth shed  up  to 1-3  years after their  psychometric  testing.  The
     impact of this may not have been symmetrically exerted on the high-Pb and low-
     Pb groups.  That is,  it is not  likely  that  "actual"  high-Pb exposure children
     at the  time of  psychometric testing would have  distinctly lower  later dentine
     Pb values; but  low-Pb  children  with initial values <10  ppm could have experi-
     enced  lead  exposures after  psychometric testing that  substantially increased
     their later dentine  Pb  values and resulted in their  exclusion from the low-Pb
     group.
(d)  Normalized outcomes  for which  age-normed  scores were not  available  were con-
     structed by regressing on age before analysis of covariance.  Assuming that age
     effects were accounted  for,  five covariates (namely,  mother's age at subject's
     birth,   mother's  educational  level,  father's  socioeconomic status,  number  of
     pregnancies,  and parental  IQ) were considered.   Only five covariates were used
     because that is  the  limit  dictated by a widely  used  computer software package
     (SPSS).   However, the number of covariates considered should not be arbitrarily
     dictated by the constraints of a packaged program but should be determined with
     the goal  of  properly  controlling  confounding  variables.   Father's  education
     (grade) level  was  not  included  separately, although  (as  Or.  Needleman argued)
     it is part of  father's  socioeconomic status.   It would seem to be  better,  based
     on the  results shown  in Table 5 of  the  1979  report,  to use father's education
     directly rather  than as part of a  diluted   "socioeconomic  status"  variable.
(e)  The Committee  reviewed  computer printouts from numerous SPSS analysis of covar-
     iance runs on psychometric  testing data indicated by  Dr.  Needleman as forming
     the basis  for the  results and  conclusions  presented in  the 1979  report  and
     noticed many missing data points among the analyses.  In fact, the  actual number
     of data points used in  certain regression analyses was sometimes as much as 20%
                                    35

-------
     fewer  than  those  for  the 158  cases  claimed  in  the 1979  paper to have  been
     analyzed for the  low-Pb  and  high-Pb subjects.   For example, the analyses later
     reported in the Bellinger and Needleman (1983)  paper (on the same data set dis-
     cussed in the  1979 article)  are based on 17 fewer cases than the 158 stated to
     have  been  included  in the  final   statistical  analyses  of psychometric  test
     results appearing  in the  1979 article, because of missing parental  IQ data for
     the 17  cases.   Missing  data,  not  alluded  to  in  the  1979 report,  can  pose a
     serious validity problem if the missing observations are not randomly distribu-
     ted across  the  important variables.   The effects  of  such  missing data  are
     impossible to  assess  without detailed  analysis  of  the  available  data  set.
(f)  Based  upon  cursory  inspection  of  the  numerous  statistical analysis  computer
     runs  provided  by  Dr.  Needleman  (which  was  all  that  was possible  during  the
     limited time of  access to the printouts), the  Committee came  away with the im-
     pression that most runs  led  to non-sfgnificant findings.  In a  recent publica-
     tion (Needleman, 1983),  the  investigator notes that of the 66  outcomes evalua-
     ted, 15 were  significantly  different  between  the  low-  and high-lead  groups,
     given the  control  variables  included in  the analyses.   He notes that 1  in  20
     would  be  expected  by  chance, ifthe  outcome  variableswere uncorrelated.   Of
     course, most of  the psychological  assessments  in this study are moderately to
     highly correlated,  so  that  this  probability  does  not  apply.   In  addition,
     apparent group  differences  are affected by the method  of handling important
     covariates.
(g)  Of special  interest, printouts for several regression analyses  in which child's
     age was entered as a control  variable  showed reduced and  generally non-signifi-
     cant coefficients  for  Pb levels,  but such findings  are not presented  in  the
     1979 report or  later articles  by  Needleman and colleagues.  This is  in contrast
     to the earlier  reporting  (Needleman et al., 1978)  of statistically  significant
     Pb  effects  when  age  was included  as a  covariate in preliminary  statistical
     analyses  performed when the  collection of psychometric data for  the study was
     about one-half  completed.  The standardized psychometric measures with age norms
     provided do not perfectly correct  for age differences  in a specific  sample.
     Because there  are  significant age  differences between the high-Pb  and  low-Pb
     groups in this  study,  the regressions of  raw test scores on child's own  age
     would  have  been  the  more  desirable  analyses to  report. The  Committee  has
     reached this  conclusion despite the principal  investigator's (Or. Needleman's)
     argument that it is undesirable to  "correct for age twice."
                                    36

-------
4.   Committee Conclusions and Recommendations

     Estimation of  Pb exposures by dentine  Pb  measurements is more appropriate than blood Pb
as an index of cumulative exposure, and the  analytical determination of such dentine Pb levels
appears  to  have  generally been done competently in the study.  However, it is not possible to
estimate  the  variance of the dentine  Pb  measurements in replicate/duplicate analyses (beyond
the  15% estimate arrived at  for replicate  analyses  for  the  most concordant samples) without
full access  to  the coded, raw data  of all  children who participated in the study.  The blood
Pb  measurements,  obtained  earlier  for  some  of  the children,  are  of  unknown reliability.
Because  the  blood data appear to have been  obtained  at varying ages for the children sampled,
the  reported  blood  Pb data probably do not  uniformly assess peak exposure levels for them and
cannot  be accepted  as providing quantitive  estimates of  Pb levels associated with any neuro-
behavioral deficits demonstrated to exist  among the children studied.
     Teacher  ratings  of children's classroom behaviors were collected on more than 2000 chil-
dren who also contributed shed deciduous  teeth for dentine  Pb analyses.  The failure to revise
the  lead  classification of  the children  based on discrepancies with later replicate/duplicate
dentine  Pb  values in the analysis of  teachers'  ratings  contrasts sharply with the demand for
concordance  of  dentine  Pb readings  in  the neuropsychological  testing  phase of  the  study.
Also, the failure to analyze for possible contributions  of confounding factors or covariates
to the  teachers'  rating results is disturbing.  (The covariance adjustment used for teachers'
ratings on  the  158 children included  in  the neuropsychological  testing phase of the study is
subject  to  the  criticisms  noted  for  other analyses of data  for  those groups.)   The  dose-
response  relationships  reported to  exist  between  dentine  Pb  levels  and teachers'  rating
scores, therefore, cannot be accepted as valid based on the published analyses.
     A  comprehensive  neuropsychological  test battery was administered to the children defined
as belonging  in  low-Pb or high-Pb subgroups.  Serious questions exist regarding the basis for
classification of subjects in these groups or exclusion of others from them.  Also, discrepan-
cies between  WISC-R  verbal  and performance  scores,  if as large or numerous as they seem upon
cursory inspection, may raise questions about the test administration or the sample selection.
Errors  in  the calculation  of some averaged parental IQ  scores,  evident  in  coded materials
provided to the Committee, introduced unknown errors into the regression analyses for the psy-
chometric testing results.  The  use of the  PPVT  for parental IQ was not ideal, but was  still
acceptable.    Exclusion  of  large  numbers of  eligible participants prior to data analysis  could
have resulted in systematic bias in the results. However, the Committee was unable to evaluate
this possibility  fully,  given the  limited information  made  available by  the  investigator.
                                              37

-------
The treatment of covariates in the statistical analysis of the psychometric testing
results was unsatisfactory. The failure to report statistical analyses showing generally
reduced or non-significant negative correlations between dentine Pb levels and performance on
the psychometric tests also lessens the credibility of those few statistically significant
effects attributed to Pb in the published version(s) of the Needleman et al. (1979) study.
In summary, at this time, based on the questionable Pb exposure categorization and sub-
ject exclusion methods, problems with missing data, and concerns regarding the statistical
analyses employed and selected for reporting, the Committee concludes that the study results,
as reported in the Needleman et al. (1979) paper, neither support nor refute the hypothesis
that low or moderate levels of Pb exposure lead to cognitive or other behavioral impairments
in children. The Committee strongly recommends that the subject data set be reanalyzed to
correct for errors in data calculation and entry noted above, that the reanalysis be based on
better exposure classification of subjects, and that all potentially confounding variables
(including age) be assessed using a backwards elimination approach analogous to that recom-
mended earlier for the reanalysis of Ernhart data.

C. Comments on the Burchfiel et al. (1980) Study

The Committee carried out only a very preliminary review of the Burchfiel et al. (1980)
study, focusing mainly on consideration of Pb exposure, neuropsychologic testing, and statis-
tical analysis aspects of the study. Review of the electrophysiological recording aspects of
the study would require additional committee members or a separate review committee with
recognized expertise in electrophysiology and, in particular, electroencephalography.
In view of the fact that the Pb exposure and psychometric measurement data utilized in
the Burchfiel study are subsets of the data underlying the Needleman et al. (1979) article
discussed above, most of the preceding comments regarding those aspects apply here as well.
Only a few additional remarks are, therefore, felt to be necessary here. Specifically,
no definite dentine Pb or blood Pb values were reported for the specific children from the
Needleman et al. (1979) study who underwent the EEG evaluations reported by Burchfiel et al.
(1980). It is therefore impossible to determine with any confidence the specific Pb exposure
levels (including either blood Pb or dentine Pb values) that may have been associated with the
reported EEG effects. Nor is it possible to accept with much confidence any reported rela-
tionships between the observed brain wave alterations, the psychometric testing scores, and Pb
exposure classification as low-Pb or high-Pb, especially in view of the various problems noted
above regarding exposure classification, psychometric testing, and statistical treatment of
covariates or confounders that preclude acceptance of the findings reported in the 1979 publi-
cation.
38

-------
0. Comments on the Needleman (1982) Report

The 1982 report published by Needleman represents mainly a summary or restatement of
findings already reported in the earlier Needleman et al. (1979), Burchfiel et al. (1980), and
Needleman et al. (1982) publications. The comments provided above on the first two of the
earlier publications, obviously, also apply here.
One additional point worthy of discussion concerns the plot of cumulative frequency dis-
tributions of verbal IQ scores for low-Pb and high-Pb subjects shown in Figure 2 of the
Needleman (1982) report, as reprinted from the Needleman et al. (1982) article. Given the
serious reservations expressed earlier by the Committee regarding the Pb-exposure classifica-
tion procedures, aspects of the psychometric testing, and statistical treatment of covariates
or confounding factors as employed in the analyses reported in the 1979 article, the particu-
lar cumulative distribution curves shown in the figure for verbal IQ scores among the low-Pb
and high-Pb subjects cannot be accepted at this time as being either qualitatively valid
(i.e., as demonstrating lower IQs for high-Pb subjects than for low-Pb subjects) or quantita-
tively accurate (i.e., in terms of absolute decreases in IQ implied to be associated with Pb
exposure). Similarly, the Committee finds certain statements in the discussion (page 731 of
the 1982 Needleman paper) of the cumulative distribution curves to be somewhat misleading in
noting that none of the included high-Pb subjects had an ,IQ over 125 (while 5% of the low-Pb
subjects did) but failing to mention that at least one subject excluded because of overt plum-
bism had a full-scale WISC-R IQ over 125.

E. Comments on the Bellinger & Needleman (1983) Study

This paper reports two kinds of reanalyses of the data from the previous (Needleman et
al., 1979) report and, again, most of the comments made earlier on aspects of that study also
apply here. Certain additional comments are warranted, however. First, child IQ is regressed
on mother's IQ separately for the low-Pb and high-Pb groups. The results are that mother-
child IQ correlations do not differ for the two Pb exposure groups and the high-Pb group has
lower than predicted IQ scores (controlling for maternal IQ).
Second, the residuals of child's IQ regressed on mother's IQ from the first analysis were
regressed on dentine Pb levels, arranged by individual values. Four ranges of lead values
were used to estimate regression slopes of residual IQ on lead. The sample size for the low-
Pb group in this report was N=94; for high-Pb, N=47; and for two subsamples of the high-Pb
group, i.e., dentine Pb levels of 20.0-29.9 ppm, N=24, and for 30.0-39.9 ppm, N=17. The
39

-------
latter two groups are far too small to be used to estimate slopes that can be credibly gerp
eralized to other samples. The results for the regression of child's IQ residuals on lead in
the low-Pb group had, not surprisingly, a slope of zero because of extreme restriction of the
range of lead values. The slope for the high-Pb group was -0.36, significantly different from
zero. One serious problem in interpreting these results is that only maternal IQ was used as
a "covariate" for child IQ. No other background factors, as reported in the earlier paper,
were included as adjustments for the residualized IQ scores in this study.
To control, in part, for additional covariates that could affect the relationships
between residual IQ and lead level, a stepwise regression was done. Surprisingly, and con-
trary to all of their other analyses of these data, lead level was allowed to enter the equa-
tion second, before two of the three control variables. Table III of the Bellinger and
Needleman (1983) article reported results in the form of unstandardized regression coeffi-
cients without accompanying standard errors or significance levels. The F ratios reported
seemed to be those of the equations, not of the individual coefficients, except of course for
the first variable in the first step. Thus, it is not clear that the Pb coefficient is
actually reliably different from zero.
Given the above problems and concerns, the reanalyses of the Needleman data set presented
in the Bellinger and Needleman (1983) paper cannot be accepted as providing credible or reli-
able estimates of quantitative relationships between Pb exposure and neuropsychologic deficits
in children. Nor can the reported results be taken as either qualitatively supporting or
refuting the hypothesis of associations between low-level lead exposure and cognitive deficits
in children.

F. Comments on the Needleman (1981) Report

Shed teeth and teacher ratings were collected in 1977-1978 from a new sample of about
1300 first-grade children in Lowell, MA. Children were classified into five groups according
to their dentine Pb levels: Group 1, <6.4 ppm; Group 2, 6.5 to 8.7 ppm; Group 3, 8.71 to 12
ppm; Group 4, 12.01 to 18.1 ppm; Group 5, £18.2 ppm. The association of teacher ratings on 11
behavior scales with Pb levels is displayed separately for males and females in Figures 4 and
5. No effort was made to control for confounding variables in this overall set of results.
Essentially complete follow-up data and Pb levels were obtained on 130 children of the
447 males selected for follow-up. Given the design of this study, the expected analysis would
investigate the relationship between teacher ratings and Pb level following adjustment for
confounding variables, collected on the follow-up sample. Such an analysis was not reported.
40

-------
In  the Committee's view,  these data  should be  reanalyzed  to show  clearly the  form  of the
relationship between  Pb level and teacher  ratings,  with appropriate controls for the follow-
up subjects.

                                     POSTSCRIPT

     In addition to evaluation of the studies of  Ernhart and Needleman, the Committee reviewed
available  reports  (some published  and others  as yet unpublished) of  other studies from the
United  States  and  Europe.   These studies  included, for  example,  those by:   Winneke  et al.
                    .*•
(1982,  1983),  Winneke  (1983),  Yule  et al.  (1981),  Lansdown et al.  (1983),  Smith (1983), and
Harvey  et  al.  (1983).   Although an exhaustive, in-depth evaluation of the world  literature on
low-level  Pb exposure  was  beyond the current charge to  the Committee, we  note  that new stu-
dies  reported  in the  spring and summer of  1983, with only a few  exceptions,  failed to find
significant association between low-level Pb exposure and neuropsychologic deficits, once con-
trol variables were taken into account.
     From  its  review  of the recent research  literature  covered  in this report, the Committee
concludes  that:   (1)  in the  absence  of control  for other variables,  a  negative association
between Pb  exposure and neuropsychologic functioning has  been established;  (2) the extent of
this negative  association  is reduced or eliminated when confounding factors are appropriately
controlled; and  (3) the Committee  knows of no studies that, to date, have validly established
(after  proper  control  for  confounding variables) a relationship between low-level Pb exposure
and neuropsychologic deficits in children.
     Research addressing possible dose-response relationships between lead and cognitive func-
tioning in children is a worthy effort, and the Committee hopes that future studies can gather
data that  speak  more  adequately to this issue.   In  the  view of the Committee, it is unlikely
that continued use of cross-sectional  epidemiological analyses will produce much credible evi-
dence for  or against  the hypothesis that  low to  moderate levels of lead exposure are respon-
sible  for  neurobehavioral   deficits  in  apparently  asymptomatic  children.   The  study  design
generally does not  allow for unambiguous disentangling of possible contributions of such lead
exposures  to  observed  cognitive or  behavioral  deficits  versus the  contributions of numerous
other potentially confounding factors.   There is a great need for longitudinal and time-series
analyses, which  include detailed prospective  measurements of Pb exposure  indices from early
childhood onward and repeated sampling of neurobehavioral endpoints, both during preschool and
school-age years.
                                              41

-------
   60
   50
   40
O  30
GC
IU
Q.
   20
                           MALES LOWELL
                           DENTINE LEAD
                           1. < 6.4 ppm
                           2.  6.5 - 8.7
                           3.  8.71 - 12
                           4. 12.01 - 18.1
                           5. 2* 18.2
                     rtl
                          /l
                                                  rl
    /  *
  /
 / t
0   /
                 £   £    &££&£&  £&   v
                 ff   &   ^    <£    ^    ^   /r^ JR*  ^T
'/
                                                HAS
                                             DIFFICULTY
     Figure 4. The relationship between negative teachers' ratings and
     dentine lead level in males. Each sample was classified into 5 groups
     according to dentine lead level. Each item was then scored. Within
     each item. Group 1, lowest lead level, is at the left; Group 5, highest
     lead level, is at the right.

     Source: Needleman (1981).
                                 42

-------
60


50


40



30


20


10
                    FEMALES-LOWELL
                    DENTINE LEAD
                    1. < 6.4 ppm
                    2. 6.5 - 8.7
                    3. 8.71 - 12
                    4.12.01 - 18.1
                    5. > 18.2
        nnrLmlfL




                              .  HAS	.
                              1 DIFFICULTY

Figure 5. The relationship between negative teachers' ratings and
dentine lead level in females. Each sample was classified into 5 groups
according to dentine lead level. Each item was then scored. Within
each item. Group 1, lowest lead level, is at the left; Group 5, highest
lead level, is at the right.

Source: Needleman (1981).
                         43

-------
                                     REFERENCES


Ammons,  R.  B. ;  Ammons, C.  H.  (1962) The quick test  (QT):  provisional manual.  Psycho!.  Rep.
     11: 111-161 (monograph suppl. i-viii).

Bellinger, D. C.; Needleman, H. L. (1983) Lead and the  relationship  between maternal  and  child
     intelligence.  J. Pediatr. (St. Louis) 102: 523-527.

Boone,  J.;  Hearn,  T.; Lewis, S.  (1979) A comparison  of interlaboratory  results  for  blood lead
     with  results  from a  definitive method.   Clin.  Chem.  (Winston  Salem, N.C.) 25:  389-393.

Burchfiel, J. L.; Duffy, F. H.; Bartels, P. H.; Needleman,  H.  L.  (1980)  The combined discrimi-
     nating  power  of quantitative  electroencephalography  and neuropsychologic  measures  in
     evaluating central nervous system effects of  lead at low levels.  In:   Needleman, H.  L.,
     ed.   Low level  lead  exposure:  the clinical  implications  of current  research.   New  York,
     NY:  Raven Press; pp. 75-90.

Carter, G.  F.  (1978)  The paper punched disc technique  for  lead in blood samples with abnormal
     haemoglobin values.  Br. J.  Ind. Med. 35: 235-240.

Ernhart,  C.  B.  (February  3,  1983)  [Letter  to D.  Weil].   Account of  conversation with  I.  M.
     Shapiro. Available for inspection at:   U.S. Environmental Protection Agency, Environmen-
     tal Criteria and Assessment Office, Research Triangle  Park, NC.

Ernhart, C.  B.  (March 11, 1983)  Summary  of  conversation between N. B.  Schell and C.  Ernhart.
     Submitted  to  Committee at March 17-18,  1983 meeting.  Available for  inspection  at:   U.S.
     Environmental  Protection  Agency,  Environmental  Criteria  and Assessment Office,  Research
     Triangle Park, NC.

Ernhart, C.  B. ;  Landa,  B. ; Callahan,  R.  (1980)  The McCarthy  scales:  predictive validity and
     stability of scores for urban black children.  Educ. Psychol. Meas. 40:  1183-1188.

Ernhart, C.  B. ;  Landa,  B. ; Schell,  N.  B.  (1981)  Subclinical  levels of  lead  and developmental
     deficit - a multivariate follow-up reassessment.   Pediatrics 67:  911-919.

Harvey,  P.; Hamlin,  M.;   Kumar,  R.  (1983)  The Birmingham blood lead  study.   Presented  at:
     annual  conference  of  the  British Psychological  Society, symposium  on  lead and health:
     some  psychological  data;   April;  University of  York,  United  Kingdom.  Available for  in-
     spection at:   U.S. Environmental Protection Agency, Environmental Criteria  and  Assessment
     Office, Research Triangle Park, NC.

Hessel,  D.  W.   (1968)  A  simple  and  rapid quantitative determination of  lead in blood.   At.
     Absorpt. Newsl. 7: 55-56.

Kleinbaum,  D. G. ;  Kupper,  L.  L. ;  Morgenstern,  H.  (1982) Epidemiological  research:  principles
     and quantitative methods.   Belmont, CA:  Lifetime Learning Publishers.

Lansdown,  R. ;  Yule,  W. ;   Urbanowicz,  M-A.;   Millar,  I. B. (1983)  Blood  lead,  intelligence,
     attainment and behavior in school  children:  overview  of  a pilot  study.   In:  Rutter,  M.;
     Russell Jones,  R. ,  eds.   Lead  versus health.   New York, NY:  John  Wiley  & Sons,  Ltd.;
     pp.267-296.
                                              44

-------
McCarthy,  D.  (1972)  McCarthy scales  of children's  abilities.   New  York,  NY:  Psychological
     Corporation.

Needleman, H. L.  (1977) Lead in the child's world, a model for action.   In:  Hemphill, D.  D.,
     ed.  Trace   substances   in   environmental  health-XI:    [proceedings  of  University  of
     Missouri's  llth  annual  conference  on trace  substances   in  environmental  health]; June;
     Columbia, MO.  Columbia, MO:  University of Missouri-Columbia; pp. 229-235.

Needleman,  H.  L.  (1981)  Studies  in  children exposed  to low  levels of  lead.   Boston,   MA
     Children's  Hospital  Medical  Center;  U.S.  Environmental Protection  Agency  report   no
     EPA-600/1-81-066.  Available from:  NTIS, Springfield, VA; PB82-108432.

Needleman, H. L.  (November 22, 1982) [Letter to L. Grant]. Available  for inspection at:  U.S
     Environmental  Protection  Agency, Environmental  Criteria  and Assessment Office,  Researc i
     Triangle Park, NC.

Needleman,  H.  L. (1982) The  neurobehavioral  consequences of  low lead exposure in childhood
     Neurobehav.  Toxicol. Teratol.  4: 729-732.

Needleman,  H.  L.  (1983)  Low  level   lead exposure  and  neuropsychological  performance.    In:
     Rutter, M.;  Russell Jones, R., eds.  Lead  versus health.   New  York,  NY:   John Wiley &
     Sons, Ltd.;  pp. 229-248.

Needleman, H. L. ; Gunnoe,  C. ;  Leviton, A.; Peresle,  H.  (1978) Neuropsychological dysfunction
     in children with "silent" lead exposure.   Pediatr. Res. 12:374.

Needleman,  H.  L. ;  Gunnoe,  C. ;  Leviton,  A.;  Reed,  R. ;   Peresie,  H. ;  Maher, C.;  Barrett,  P.
     (1979) Deficits in psychological and classroom performance of children with elevated den-
     tine lead levels. N. Engl. J.  Med. 300:  689-695.

Needleman,  H.  L. ;  Leviton,  A.;  Bellinger,  D.  (1982)  Lead-associated  intellectual  deficit
     [Letter].  N. Engl. J.  Med. 306:  367.

Perino, J. (1973) The relation of subclinical  lead levels to cognitive and sensorimotor impair-
     ment in preschool black children.  Hempstead, NY:  Hofstra University.  Ph.D. Dissertation.

Perino, J.; Ernhart, C. B.  (1974) The relation of subclinical  lead level  to cognitive  and sen-
     sorimotor impairment in black preschoolers. J. Learn. Dis. 7: 616-620.

Sattler,  J.  M.  (1982)  Assessment  of  children's  intelligence   and  special abilities.    2nd
     ed. Boston,  MA: Allyn and Bacon.

Scarr,  S.; Yee,  D.  (1980)  Heritability and educational  policy:  genetic  and environmental  ef-
     fects on IQ, aptitude and achievement. Educ.  Psychol. 15:1-22.

Shapiro,  I. M.;  Dobkin,  B.;  Tuncay,  0. C.; Needleman, H. L. (1973) Lead  levels in dentine  ar ;
     circumpulpal dentine  of  deciduous  teeth  of  normal  and  lead poisoned  children,  Clir
     Chim. Acta 46:  119-123.

Shapiro,  I. M.;  Burke,  A.;  Mitchell, G.; Bloch, P. (1978) X-ray fluorescence analysis of lea I
     in teeth of  urban children  iji situ: correlation between  the tooth lead level and concer
     tration of blood lead and free erythroporphyrins.  Environ.  Res.  17: 46-52.
                                              45

-------
Smith, M.;  Delves,  T.;  Lansdown, R.; Clayton, B.; Graham,  P.  (1983)  The effects  of  lead expo-
     sure  on  urban children:   the Institute  of  Child  Health/Southampton  study.   London,
     United Kingdom:  Department of  the Environment.

Steenhout,  A.;  Pourtois,  M.  (1981)  Lead  accumulation in teeth as  a  function  of  age with dif-
     ferent exposures.  Br. J. Ind. Med. 38: 297-303.

U.S.  Environmental  Protection Agency,  Health Effects  Research Laboratory (1977) Air  quality
     criteria  for  lead.   Research  Triangle  Park, NC:   U.S.  Environmental Protection -Agency,
     Criteria  and  Special Studies  Office; EPA  report  no.  EPA-600/8-77-017.   Available  from:
     NTIS, Springfield, VA; PB-280411.

U.S. Environmental Protection Agency. (August, 1983)  Air  quality criteria for  lead.   First Ex-
     ternal Review  Draft.   Research Triangle Park, NC:   U.S.  Environmental Protection  Agency,
     Environmental Criteria and Assessment Office.

Winneke, G. (1983)  Neurobehavioral  and neuropsychological  effects  of lead.    In:  Rutter, M.;
     Russell Jones, R., eds.  Lead versus  health.  New York, NY:  John Wiley & Sons,  Ltd.; pp.
     249-265.

Winneke,  G.;  Hrdina,  K-G.;  Brockhaus,  A. (1982) Neuropsychological  studies  in children  with
     elevated  tooth-lead  concentrations.   Part  I:    Pilot  study.   Int.  Arch.  Occup. Environ.
     Health 51: 169-183.

Winneke, G.;  Kramer, U.;  Brockhaus, A.; Ewers, U.; Kujanek, G.; Lechner,  H.;  Janke,  W.  (1983)
     Neuropsychological studies in children with elevated tooth lead  concentrations.  Part II:
     Extended study.  Int. Arch.  Occup. Environ. Health 51: 231-252.

Yamins, J.  (1976)  The  relationship of subclinical lead intoxication  to  cognitive and language
     functioning and  preschool  children.   Hempstead, NY:   Hofstra University.  Dissertation.

Yule, W. ;  Lansdown, R. ;  Millar,  I.  B. ; Urbanowicz, M-A.  (1981) The relationship between  blood
     lead concentrations,  intelligence  and attainment in a school population:  a pilot  study.
     Dev.  Med. Child Neurol. 23:  567-576.
                                              46

-------
                                         ATTACHMENT 1


   Additional materials considered  in review of  studies  by Dr.  Claire Ernhart and colleagues

 1.  Grant, L. D.  (March  7, 1983) [Letter to C. Ernhart].  Available for inspection at:  U.S.
          Environmental  Protection  Agency,  Environmental  Criteria and  Assessment  Office,
          Research Triangle Park, NC.

 2.  Ernhart, C.  B. Coded raw data entries of psychometric test scores and background variabls
          values (sex, parental  IQ,  etc.) for subjects evaluated in Perino and Ernhart (1974)
          and  Ernhart et  al.  (1981)  studies.   Submitted  to Committee at  March 17-18,  198J
          meeting.    Available  for  inspection  at: U.S.  Environmental Protection  Agency,  En-
          vironmental Criteria and Assessment Office, Research Triangle Park, NC.

 3.  Ernhart, C.  B.;  Landa,  B.  (1980) "Cumulative deficit,"  a longitudinal analysis of score;
          on McCarthy scales.  Psychol. Rep. 47: 283-286.

 4.  Ernhart, C.  B.; Landa, B.;  Schell,  N.  B.  (1981) Lead  levels  and  intelligence [Letter].
          Pediatrics 68:  903-904.

 f.  Spector, S.;  Brown,  K.  E.  (1982) Lead study questioned  [Letter]. Pediatrics 69:  134-135.

 6.  Ernhart, C.  B.; Landa, B.; Schell, N. B. (1982) Lead study questioned: in reply [Letter].
          Pediatrics 69:  135.

 7.  Ernhart, C.  B.  (1982)  Lead results:  no  justification  [Letter].  Sci.  News (Washington,
          O.C.) 122: 3.

 8.  Ernhart, C.  B. Scatter plots and associated coded data listings of residualized cognitive
          test performance  scores  vs.  blood  lead  levels for  subjects  used  in  Perino  and
          Ernhart  (1979)  study.   Submitted  to  Committee at March 17-18, 1983 meeting.   Avail-
          able for inspection at: U.S. Environmental  Protection Agency,  Environmental  Criteria
          and Assessment Office,  Research Triangle Park, NC.

 9.  Ernhart, C.  B.  Scatter diagram of parents'  IQ  vs.  child's IQ for  low  and moderate lead
          groups in Perino and Ernhart (1979) study (based on Perino dissertation).  Submitted
          to Committee  at March 17-18,  1983 meeting.   Available for inspection  at:  U.S.  En-
          vironmental  Protection  Agency,   Environmental   Criteria  and  Assessment  Office,
          Research Triangle Park, NC.

10.  Ernhart, C.  B. ;  Landa,  B.;  Schell, N.  B.  (1983) Lead and intelligence - the effect of an
          outlier.   Pediatrics (Submitted for publication).

11.  Ernhart, C.  B.  Summary of conversation between N.  B.  Schell  and C. Ernhart on March 11,
          1983.   Submitted to  Committee  at March 17-18, 1983 meeting.   Available for inspec-
          tion  at:  U.S.  Environmental  Protection  Agency,  Environmental  and  Criteria  and
          Assessment Office, Research Triangle Park,  NC.

12.  Ernhart, C. B.  (1983)  Summary report:   reanalysis of data of three studies:  the effects
          of lead  on children.   Available  for inspection  at:   U.S.  Environmental Protection
          Agency,   Environmental  Criteria and Assessment Office, Research Triangle Park,  NC.
                                              47

-------
13.   Ernhart,  C.  B.  (October 7,  1983)  Response to  [preliminary draft  of] Appendix  12-C.
          Available  for inspection  at:   U.S. Environmental  Protection Agency,  Environmental
          Criteria and Assessment Office,  Research Triangle Park,  NC.

14.   Ernhart,  C. B.  (October 25,  1983) [Letter  to  L.  Grant].   Available for inspection  at:
          U.S.  Environmental  Protection Agency,  Environmental Criteria and Assessment  Office,
          Research Triangle Park, NC.

 Additional materials  considered  In review of studies by Dr.  Herbert  Needleman and  colleagues

 1.  Grant, L.  D.  (October 25, 1982)  [Letter to H.  Needleman].   Available for inspection  at:
          U.S.  Environmental  Protection Agency,  Environmental Criteria and Assessment  Office,
          Research Triangle Park, NC.

 2.  Needleman,  H.  L.  (November 22,  1982) [Reply to L. D.  Grant].   Available for  inspection
          at:  U.S.  Environmental  Protection Agency,  Environmental and Criteria and  Assessment
          Office, Research Triangle Park,  NC.

 3.  Needleman,  H.  L.   (November  22,  1982)  [List of 250 excluded subjects for Chelsea  run].
          Available  for inspection  at:   U.S. Environmental  Protection Agency,  Environmental
          Criteria  and Assessment Office,  Research Triangle  Park,  NC.  Attachment to Item 2
          above.

 4.  Needleman,  H.  L.  (November 22, 1982)  [Frequency distribution of  psychometric  scores  for
          high-lead  subjects.]   Available for  inspection at:  U.S.   Environmental  Protection
          Agency, Environmental  Criteria  and Assessment Office,  Research Triangle Park,  NC.
          Attachment to Item 2 above.

 5.  Needleman,  H.  L.  (November 22, 1982)  [Frequency distribution of  psychometric  scores  for
          low-lead  subjects.]   Available for  inspection  at:  U.S.  Environmental  Protection
          Agency, Environmental  Criteria  and Assessment Office,  Research Triangle Park,  NC.
          Attachment to Item 2 above.

 6.  Needleman,  H.  L.  (1977)  Lead in  the  child's  world, a model for action.   In:  Hemphill, D.
          D.,  ed. Trace substances in environmental  health-XI:  [proceedings  of University of
          Missouri's  llth annual  conference  on  trace  substances  in environmental health];
          June;  Columbia,  MO.  Columbia,  MO:  University  of  Missouri-Columbia;  pp. 229-235.

 7.   Needleman,   H.  L. ;   Leviton,  A.  (1979) Lead  and neurobehavioral  deficit  in children
          [Letter].  Lancet 2(8133): 104.

 8.  Needleman,  H.  L.   (1980)  Lead and  neuropsychological  deficit: finding  a threshold.  In:
          Needleman, H.  L.,  ed.  Low level lead  exposure: the  clinical  implications  of  current
          research.  New York, NY:  Raven  Press; pp. 43-51.

 9.  Needleman,  H.  L.; Landrigan,  P.  J.  (1981) The health effects of low level exposure to
          lead.  Ann. Rev. Public Health  2: 277-298.

10.  Needleman,  H.  L.;  Bellinger,  D.; Leviton,  A.  (1981)  Does lead at low dose  affect intel-
          ligence in children? Pediatrics  68: 894-896.

11.  Needleman,  H.  L.   (1982) The lead  debate:  a response. Environ.  Sci.  Technol.  16:   208A.

12.  Needleman,  H.  L.  (1983)  The  prevention of mental  retardation  and learning  disabilities
          due  to  lead  exposure.   In:  Jahiel, R.  I.,  ed. The  handbook of prevention of mental
          retardation and learning disabilities.    (In preparation)
                                              48

-------
13.  Grant,  L.  D. (March  14,  1983)  [Letter to H.  Needleman].   Available for inspection  at:
          U.S. Environmental Protection Agency, Environmental Criteria and Assessment Office,
          Research Triangle Park, NC.

14.  Jones,  L. V.  (March 23,  1983)  [Letter  to  L.  Kupper].  Available for inspection at: U.S.
          Environmental  Protection  Agency,  Environmental  Criteria  and Assessment  Office,
          Research Triangle Park, NC.

15.  Grant,  L. D. (April 8, 1983) [Letter to H. Needleman].  Available for inspection at: U.S.
          Environmental  Protection  Agency,  Environmental  Criteria  and Assessment  Office,
          Research Triangle Park, NC.

16.  Needleman, H. L. (April 13, 1983)  [Reply to L.  Grant].  Available for inspection at: U.S.
          Environmental  Protection  Agency,  Environmental  Criteria  and Assessment  Office,
          Research Triangle Park, NC.

17.  Needleman, H. L.  (May 26,  1983)  [Lists  of all  dentine lead  levels  for  included children
          and for the  30 excluded from neuropsychological  testing,  and the  unadjusted means,
          SD's and   t  tests].   Available  for  inspection at:    U.S.  Environmental Protection
          Agency, Environmental  Criteria and  Assessment Office,  Research Triangle Park,  NC.

18.  Needleman, H. L.;  Rabinowitz,  M.; Leviton, A.  (1983) The risk of minor  congenital anoma-
          lies in relation  to umbilical  cord  blood lead  levels.   Pediatr.  Res.  17:  300A.

19.  Needleman, H. L.;  Bellinger,  D.;  Leviton, A.;  Rabinowitz, M.; Nichols,  M. (1983) Umbili-
          cal cord blood lead  levels and neuropsychological performance  at  12 months of age.
          Pediatr. Res. 17: 179A.

20.  Needleman,  H.   L.  (June  14,  1983) [Letter  to  L.  Grant].   Final complete  dentine lead
          levels on  those  subjects  excluded because of  discordant values.   Available for  in-
          spection  at:  U.S.   Environmental  Protection  Agency,  Environmental   Criteria   and
          Assessment Office, Research Triangle Park, NC.

21.  Needelman,  H.   L.  (1983)   Listings of  coded raw  data for  dentine   lead  levels  of both
          "included"  and "excluded"  subjects who  underwent  psychometric  testing  in Needleman
          et al.  (1979) study.   Attachment to Item 20 above.

22.  Needleman, H.  L.  (1983)  Listings of coded  raw data  for psychometric  test results  and
          background variables  (e.g.,  sex,  age, father's education,  parental I.Q., etc.)  for
          "included"  and  "excluded"  subjects in Needleman et al. (1979)  study.  Attachment to
          Item 20 above.

23.  Needleman, H.  L.  (1983) Computer printouts  of frequency  distribution  of IQ scores  and
          other psychometric test results  for high  and  low lead subjects in Needleman et  al.
          (1979)  study.   Inspected  by  Committee at H. L. Needleman1s facilities at University
          of Pittsburgh (Children's  Hospital), Pittsburgh, PA.

24.  Needleman, H.  L.  (1983) Computer printouts  of results of  SPSS version  of analysis of
          covariance  for psychometric  test  scores of high  and  low lead  subjects of Needleman
          et al.   (1979)  study,  taking  into account  up  to five covariates (including age as a
          covariate  in some runs).   Inspected by Committee at H. L. Needleman1s facilities at
          University of Pittsburgh (Children's Hospital), Pittsburgh, PA.

25.  Needleman,  H.  L.  (October 4, 1983)  [Letter  to  L. Grant].   Available for inspection  at:
          U.S.  Environmental Protection Agency,  Environmental  Criteria and Assessment Office,
          Research Triangle Park,  NC.

                                              49

-------
26.   Needleman,  H.  L.  (October 4, 1983)  [Letter to B.  Goldstein].   Available for  Inspection
          at:   U.S.   Environmental Protection Agency,  Environmental  Criteria and  Assessment
          Office, Research Triangle Park,  NC.

27.   Grant,  L.  D.  (October  7,  1983)  [Reply to H. L. Needleman].  Available  for inspection at:
          U.S.  Environmental Protection Agency,  Environmental Criteria and  Assessment Office,
          Research Triangle  Park,  NC.

28.   Needleman, H.  L.  (October 7, 1983)  [Letter to L.  Grant].   Available for inspection at:
          U.S.  Environmental Protection Agency,  Environmental Criteria and  Assessment Office,
          Research Triangle  Park,  NC.

                                                                      4ULS. GOVERNMENT HlllfflMS OFFICE: 1WJ-WM17/HW
                                               50

-------
&EFA
United States       Environmental Criteria and
Environmental Protection Assessment Office
Agency          Research Triangle Park NC 27711
                                                   EPA-600/8-83-028A
                                                   October 1983
                                                   External Review Draft
                 Research and Development
Air Quality
Criteria for Lead

Volume I of IV
Review
Draft
                                                    (Do Not
                                                    Cite or Quote)
                                             FBI COPY
                                NOTICE
                 This document is a preliminary draft. It has not been formally
                 released by EPA and should not at this stage be construed to
                 represent Agency policy. It is being circulated for comment on its
                 technical accuracy and policy implications.

-------
                           Review Comment Guidelines
     Reviewers are  encouraged  to submit comments  regarding  the material pre-
sented In this document.   To facilitate the revision process, we request that
certain guidelines be followed.  Both general and specific comments on each of
the chapters  will be useful to  us.   However, we  cannot  emphasize enough the
importance  of making your  comments  specific,  to the  extent  of  indicating
section  and page  numbers of  the material  in  question.  This  will  greatly
increase the likelihood that the editors and authors will be able to adequate-
ly  address  your   comments  within a reasonable  period of time.   We also hope
your comments  will  include  your concurrence with the information  as  well  as
any disagreements.

     Please carefully review the following guidelines prior to providing your
written comments.  Since these comments may be included,  as provided, possibly
in a separate volume at a later date,  please have your comments neatly presen-
ted; typing is preferred.
Guidelines to Reviewers-
     1.   Chapters  should be  reviewed using a  separate form  for each.

     2.   General  comments,  impressions,  or opinions   of  the  chapter
          should be presented on the comment form, separate from specific
          comments  referencing  sections,   paragraphs,   or  statements.

     3.   Comments regarding concepts  or discussions presented should be
          addressed by noting the section at the beginning of the review-
          er's  comments.   Specific comments on  the  text within sections
          should be  identified with page number,  paragraph  (P) and line
          (L), if necessary.
     Example-
     Secti on/Page/Paragraph/Line

     2.2.3   15      2      2
     Specific Comments

Discussion  on ...  does  not
consider  studies  by  ...  in
making conclusions.
     2.2.3   20
"Susceptible to change"
would  be  more  appropriate
than "sensitive."
     4.   When commenting on specific sections/discussion and paragraphs/
          statements, some  suggested  reviewer responses that may be use-
          ful include (with additional  appropriate details by reviewer):

-------
               information/statement inaccurate
          -    information/statement   Incorrectly   interprets  available
               data and/or literature
               information/statement is based on insufficient data and/or
               literature


Commonly,  reviewers'  comments  note "this  information  is  not  correct!"  You
should be  aware  that  unless  specific evidence  is  provided  demonstrating that
the information is, in fact,  inaccurate, no action can be taken.

     During your review  of each chapter, it will be necessary to keep in mind
the intended purpose  and scope of this  document.   Comments  will be addressed
by the authors as  deemed appropriate.   These authors are charged with produc-
ing a scientifically  sound and accurate document and are not expected to make
recommendations for the  use  of the material in future decision-making proces-
ses.   They will therefore address comments on technical accuracy only.

     Only one copy of the review form follows.   This form should be copied and
used for all written comments.

-------
                         DOCUMENT REVIEW FORM
             Title/Draft
  Air Quality Criteria  for  Lead
   (External Review Draft #1)
                             Return to:
                 Project Officer for  Lead
                 Environmental  Criteria and
                  Assessment Office,  U.S.  EPA
                 MD-52
                 Research Triangle Park, NC 27711
    Commenter/Organizatlon/Address  *
                             Chapter No./Title
           RECOMMENDATIONS
       O (1) Acceptable a* is
       O (2) Acceptable after minor revision
       O (3) Acceptable after major rtvision
       O (4) Not acceptable
        H you have checked either 3 or 4,
        please specifically itaier«ason(«) in
        the comments space below.
  Reviewer's Signature
Date
             Additional Contributors to Document
                Review (Name and Affiliation)
     General Comments  (use extra pages  If  needed,  but please be concise)
*If comments being submitted on behalf of another organization,  include that
 Information.

-------
                                       EPA-600/8-83-028A
                                       October 1983
Draft                                  External Review Draft

Do Not Quote or Cite
             Air Quality  Criteria
                      for Lead

                      Volume  I
                            NOTICE

This document is a preliminary draft. It has not been formally released by EPA and should not at this stage
be construed to represent Agency policy. It is being circulated for comment on its technical accuracy and
policy implications.
           Environmental Criteria and Assessment Office
           Office of Health and Environmental Assessment
               Office of Research and Development
               U.S. Environmental Protection Agency
               Research Triangle Park, N.C. 27711

-------
                               NOTICE

Mention of trade names or commercial products does not constitute
endorsement or recommendation for use.
                                  ii

-------
                                   ABSTRACT

     The document evaluates and  assesses  scientific information on the  health
and welfare effects associated with exposure to various concentrations of lead
in ambient air.  The  literature  through 1983 has been reviewed thoroughly for
information relevant  to air  quality  criteria, although  the  document  is  not
intended as  a complete  and  detailed review  of all  literature  pertaining to
lead.  An  attempt has  been  made to  identify the major  discrepancies  in our
current knowledge and understanding of the effects of these pollutants.
     Although  this document   is  principally  concerned  with  the  health  and
welfare effects of  lead,  other scientific data are presented and evaluated in
order to provide a  better understanding of this pollutant in the environment.
To this  end,  the  document  includes chapters  that discuss the  chemistry and
physics  of  the  pollutant;   analytical  techniques;   sources,   and  types  of
emissions;   environmental  concentrations  and  exposure  levels;  atmospheric
chemistry  and dispersion  modeling; effects  on vegetation;  and respiratory,
physiological, toxicological,  clinical, and  epidemiological  aspects of human
exposure.
                                     iii

-------
                                       PRELIMINARY DRAFT
                                           CONTENTS
                                                                                         Page
VOLUME I
  Chapter 1.

VOLUME II
  Chapter 2.
  Chapter 3.
  Chapter 4.
  Chapter 5.
  Chapter 6.
  Chapter 7.
  Chapter 8.

VOLUME III
  Chapter 9.

  Chapter 10.
  Chapter 11.

Volume IV
  Chapter 12.
  Chapter 13.
Executive Summary and Conclusions
 Introduction	
 Chemical and Physical Properties 	
 Sampling and Analytical Methods for Environmental Lead 	
 Sources and Emissions 	
 Transport and Transformation	,	
 Environmental Concentrations and Potential Pathways to Human Exposure
 Effects of Lead on Ecosystems 	
 Quantitative Evaluation of Lead and Biochemical Indices of Lead
 Exposure in Physiological Media 	
 Metabol i sm of Lead	
 Assessment of Lead Exposures and Absorption in Human Populations
 Biological Effects of Lead Exposure 	
 Evaluation of Human Health Risk Associated with Exposure to Lead
 and Its Compounds 	
 1-1
 2-1
 3-1
 4-1
 5-1
 6-1
 7-1
 8-1
 9-1
10-1
11-1
12-1

13-1
                                              iv

-------
                                       PRELIMINARY DRAFT
                                       TABLE OF CONTENTS
                                           CHAPTER'1
                               EXECUTIVE SUMMARY AND CONCLUSIONS


                                                                                           Page

LIST OF FIGURES 	       v
LIST OF TABLES 	      vi


 1.  EXECUTIVE SUMMARY AND CONCLUSIONS 	     1-1
     1.1  INTRODUCTION	     1-1
     1.2  ORGANIZATION OF DOCUMENT	     1-3
     1.3  CHEMICAL AND PHYSICAL PROPERTIES OF LEAD 	     1-4
     1.4  SAMPLING AND ANALYTICAL METHODS FOR ENVIRONMENTAL LEAD 	     1-6
          1.4.1  Sampling Techniques	     1-7
          1.4.2  Analytical Procedures	     1-10
     1.5  SOURCES AND EMISSIONS	     1-13
     1.6  TRANSPORT AND TRANSFORMATION 	     1-22
     1:7  ENVIRONMENTAL CONCENTRATIONS AND POTENTIAL PATHWAYS
          TO HUMAN EXPOSURE ,	     1-34
          1.7.1  Lead in Air	     1-34
          1.7.2  Lead in Soil and Dust	     1-37
          1.7.3  Lead in Food	•	     1-38
          1.7.4  Lead in Water	     1-39
          1.7.5  Basel ine Exposures to Lead	     1-40
          1.7.6  Additional Exposures 	     1-45
                 Urban atmospheres 	     1-45
                 Houses with interior lead paint	     1-47
                 Fami ly gardens 	     1-47
                 Houses with lead plumbing 	<	     1-47
                 Residences near smelters and refineries	     1-48
                 Occupational exposures	     1-48
                 Secondary occupational exposure	     1-49
                 Special habits or activities 	     1-49
     1.8  EFFECTS OF LEAD ON ECOSYSTEMS	     1-52
          1.8.1  Effects on Plants	     1-57
          1.8.2  Effects of Animals	     1-61
          1.8.3  Effects on Microorganisms	     1-63
          1.8.4  Effects on Ecosystems	     1-64
          1.8.5  Summary	     1-66
     1.9  QUANTITATIVE EVALUATION OF LEAD AND BIOCHEMICAL INDICES OF LEAD
          EXPOSURE IN PHYSIOLOGICAL MEDIA	     1-67
          1.9.1  Determinations of Lead in Biological Media 		     1-67
                 Measurements of 1 ead i n bl ood	     1-68
                 Lead in plasma	     1-69
                 Lead in teeth 	     1-69
                 Lead in hair	:	     1-69
                 Lead in urine 	     1-70
                 Lead i n other ti ssues	     1-70
                 Quality assurance procedures in lead analyses	     1-70
          1.9.2  Determination of Erythrocyte Porphyrin (Free Erythrocyte
                 Protoporphyrin,  Zinc Protoporphyrin) 	     1-71
          1.9.3  Measurement of Uri nary Coproporphyrin 	     1-72


CHP1D/B                                                                                9/30/83

-------
                                       PRELIMINARY DRAFT



                                TABLE OF CONTENTS (continued).

                                                                                          Page

          1.9.4  Measurement of Delta-Aminolevulinic Acid Dehydrase Activity 	    1-72
          1.9.5  Measurement of Delta-Aminolevulim'c Acid in Urine and
                 Other Media	    1-73
          1.9.6  Measurement of Pyrimidine-5'-Nucleotidase Activity	    1-74
     1.10 METABOLISM OF LEAD	    1-74
          1.10.1 Lead Absorption in Humans and Animals	    1-75
                 Respiratory absorption of lead	    1-75
                 Gastrointestinal absorption of lead 	    1-75
                 Percutaneous absorption of lead	    1-76
                 Transplacental transfer of lead	,	    1-76
          1.10.2 Distribution of Lead in Humans and Animals 	    1-77
                 1.10.2.1 Lead in Blood	    1-77
                 1.10.2.2 Lead Levels in Tissues 	    1-77
                          Soft tissues 	    1-78
                          Mineralizing tissue 	    1-78
                          Chelatable lead	    1-79
                          Animal studies	    1-79
          1.10.3 Lead Excretion and Retention in Humans and Animals 	    1-80
                 Human studies 	    1-80
                 Animal studies 	    1-81
          1.10.4 Interactions of Lead with Essential Metals and Other Factors 	    1-81
                 Human studies	    1-81
                 Animal studies	    1-82
          1.10.5 Interrelationships of Lead Exposure with Exposure Indicators
                 and Tissue Lead Burdens	    1-82
                 Temporal characters!"tics of internal indicators
                 of 1 ead exposure 	    1-83
                 Biological aspects of external  exposure-
                 internal indicator relationships 	    1-83
                 Internal indicator-tissue lead relationships 	    1-83
          1.10.6 Metabolism of Lead Alkyls 	    1-84
                 Absorption of lead alkyIs in humans and animals 	    1-84
                 Biotransformation and tissue distribution of lead alkyls	    1-85
                 Excretion of lead alkyls	    1-85
     1.11 ASSESSMENT OF LEAD EXPOSURES AND ABSORPTION IN HUMAN POPULATIONS 	    1-85
          1.11.1 Levels of Lead and Demographic Covariates
                 in U.S.  Populations 	    1-86
          1.11.2 Blood Lead vs. Inhaled Air Lead Relationships 	    1-92
          1.11.3 Dietary Lead Exposures Including Water 	    1-96
          1.11.4 Studies Relating Lead in Soil and Dust to Blood Lead	    1-97
          1.11.5 Paint Lead Exposures 	    1-98
          1.11.6 Specific Source Studies 	    1-99
          1.11.7 Primary Smelters Populations 		    1-102
          1.11.8 Secondary Exposure of Children 	    1-105
     1.12 BIOLOGICAL EFFECTS OF LEAD EXPOSURE 	    1-106
          1.12.1 Introduction	    1-106
          1.12.2 Subcellular Effects of Lead	    1-lOf
          1.12.3 Effects of Lead on Heme Biosynthesis,  Erythropoiesis, and
                 Erythrocyte Physiology in Humans and Animals 	    1-lOf
          1.12.4 Neurotoxic Effects of Lead 	    1-115
          1.12.5 Effects of Lead on the Kidney	,	    1-119

                                              vi
CHP1D/B                                                                                9/30/S3

-------
                                       PRELIMINARY DRAFT



                                TABLE OF CONTENTS (continued).

                                                                                          Page

          1.12.6 Effects of Lead on Reproduction and Development 	    1-121
          1.12.7 Genotoxic and Carcinogenic Effects of Lead 	    1-122
          1.12.8 Effects of Lead on the Immune System 	    1-123
          1.12.9 Effects of Lead on Other Organ Systems 	    1-123

     1.13 EVALUATION OF HUMAN HEALTH RISKS ASSOCIATED WITH EXPOSURE TO LEAD AND
          ITS COMPOUNDS	    1-123
          1.13.1 Introduction		    1-123
          1.13.2 Exposure Aspects 	    1-124
          1.13.3 Lead Metabolism:  Key Issues for Human Health Risk Evaluation 	    1-130
          1.13.4 Biological Effects of Lead Relevant to the General Human Population  .    1-136
          1.13.5 Dose-Response Relationships for Lead Effects in Human Populations ...    1-145
          1.13.5 Populations at Risk	    1-148
          1.13.7 Summary and Conclusions 		    1-151
                                              vii
CHP1D/B                                                                                9/30/83

-------
                                       PRELIMINARY DRAFT



                                        LIST OF TABLES

Table                                                                                     Page

 1-1   Estimated atmospheric lead emissions for the United States, 1981 and
         the world	      1-17
 1-2   Summary of surrogate and vegetation surface deposition of lead	      1-30
 1-3   Estimated global deposition of atmospheric lead 	      1-31
 1-4   Atmospheric lead in urban, rural, and remote areas of the world 	      1-35
 1-5   Background lead in basic food crops and meats 	      1-39
 1-6   Summary of environmental concentrations of lead 	      1-41
 1-7   Summary by age and sex of estimated average levels of lead ingested
         from mi 1 k and foods	.	      1-43
 1-8   Summary of basel ine human exposures to lead	      1-46
 1-9   Weighted geometric mean blood lead levels from NHANES II survey by
         degree of urbanization of place of residence in the U.S. by age
         and race, United States 1976-80	      1-89
 1-10  Summary of pooled geometric standard deviations and estimated
         analytic errors 	      1-93
 1-11  Summary of blood inhalation slopes (jg/dl per vg/m* 	      1-94
 1-12  Estimated contribution of leaded gasoline to blood lead by inhalation
          and non-inhalation pathways	     1-101
 1-13  Summary of basel i ne human exposures to 1 ead	     1-126
 1-14  Relative baseline human lead exposures expressed per kilogram body weight 	     1-127
 1-15  Summary of potential additive exposures to lead 	     1-128
 1-16  Direct contributions of air lead to blood lead (PbB) in adults at fixed
       inputs of water and food lead 	     1-135
 1-17  Direct contributions of air tead to blood lead in children at fixed inputs
       of water and food lead	     1-135
 1-18  Contributions of dust/soil lead to blood lead in children at fixed inputs
       of air, food, and water 1 ead	     1-135
 1-19  Summary of lowest observed effect levels for key lead-induced health effects
       in adults	     1-139
 1-20  Summary of lowest observed effect levels for key lead-induced health effects
       in children	     1-141
 1-21  EPA-estimated percentage of subjects with ala-u exceeding limits for
       various blood lead levels	     1-147
 1-22  Provisional estimate of the number of individuals in urban and rural
       population segments at greatest potential risk to lead exposure 	     1-151
                                               viii
CHP1D/B                                                                                9/30/83

-------
                                       PRELIMINARY DRAFT



                                        LIST OF FIGURES

                                                                                          Page

 1-1   Pathways of lead exposure from the environment to man 	     1-2
 1-2   Metal complexes of lead	     1-5
 1-3   Softness parameters of metals	     1-6
 1-4   Chronological record of the relative increase of lead in snow strata,
         pond and lake sediments, marine sediments, and tree rings.   The data
         are expressed as a ratio of the latest year of the record and should
         not be interpreted to extend back in time to natural or uncontaminated
         levels of lead concentration 	      1-14
 1-5   The global lead production has changed historically in response to
         major economic and political events.  Increases in lead production
         (note log scale) correspond approximately to historical increases
         in lead emissions shown in Figure 1-4		     1-15
 1-6   Locations of major 1 ead operations in the United States		     1-18
 1-7   Trend in lead content of U.S.  gasolines, 1975-1982 	     1-20
 1-8   Relationship between lead consumed in gasoline and composite
         maximum quarterly average lead levels, 1975-1980 	     1-21
 1-9   Profile of lead concentrations in the central northeast Pacific.   Values
         below 1000 m are an order magnitude lower than reported by Tatsumoto and
         Patterson (1963) and Chow and Patterson (1966) 	     1-26
 1-10  Lead concentration profile in snow strata of northern Greenland 	     1-27
 1-11  Variation of lead saturation capacity with cation exchange capacity in
         soi 1 at sel ected pH val ues 	     1-32
 1-12  This figure depicts cycling process within major components of a terrestrial
         ecosystem, i.e.  primary producers, grazers, and decomposers.   Nutrient and
         non-nutrient elements are stored in reservoirs within these components.
         Processes that take place within reservoirs regulate the flow of elements
         between reservoirs along established pathways.  The rate of flow is in
         part a function of the concentration in the preceding reservoir.  Lead
         accumulates in decomposer reservoirs which have a high binding capacity
         for this metal.   It is likely that the rate of flow away from these
         reservoirs has increased in past decades and will continue to increase
         for some time until the decomposer reservoirs are in equilibrium with the
         entire ecosystem.  Inputs to and outputs from the ecosystems as a whole
         are not shown 	     1-54
 1-13  Geometric mean blood lead levels by race and age for younger children in the
         NHANES II study, and the Kellogg/Silver Valley and New York childhood
         screening studies	     1-87
 1-14  Average blood lead levels of U.S.  population 6 months - 74 years.  United
         States, February 1976 - February 1980, based on dates of examination of
         NHANES II examinees with blood lead determinations 	     1-90
 1-15  Time dependence of blood lead for blacks, aged 24 to 35 months, in New York
         City and Chicago	     1-91
 1-16  Change in ^Pb/^lpb ratios in petrol, airborne particulate and
         blood from 1974 to 1981	     1-100
 1-17  Geometric mean blood lead levels of New York City children (aged 25-36
         months) by ethnic group, and ambient air lead concentration vs.
         quarterly sampling period, 1970-1976 	     1-103
 1-18  Geometric mean blood lead levels of New York City children (aged 25-36
         months) by ethnic group, and estimated amount of lead present in gasoline
         sold in New York, New Jersey, and Connecticut vs. quarterly sampling
         period, 1970-1976	     1-104

                                               ix
CHP1D/B                                                                                9/30/83

-------
                                       PRELIMINARY DRAFT



                                 LIST OF FIGURES (continued).

Figure                                                                                    Page

 1-19  Dose-response for elevation of EP as a function of blood lead level using
       probit analysis 	    1-146
 1-20  Dose-response curve for FEP as a function of blood lead level:  in sub-
       populations	    1-146
 1-21  EPA calculated dose-response for ALA-U	    1-147
CHP1D/B                                                                                9/30/83

-------
                            AUTHORS AND CONTRIBUTORS
Chapter 1:  Executive Summary

Principal Author

Dr. Lester D. Grant
Director
Environmental Criteria and Assessment Office
Environmental Protection Agency
MD-52
Research Triangle Park, NC  27711

Contributing Authors:

Dr. J. Michael Davis
Environmental Criteria and
  Assessment Office
MD-52
Research Triangle Park, NC  27711

Dr. Vic Hasselblad
Biometry Division
Health Effects Research Laboratory
MD-55
Research Triangle Park, NC  27711

Dr. Paul Mushak
Department of Pathology
School of Medicine
University of North Carolina
Chapel Hill, NC  27514
Dr. Robert W. Ellas
Environmental Criteria and
  Assessment Office
MD-52
Research Triangle Park, NC  27711

Dr. Dennis J. Kotchmar
Environmental Criteria and
  Assessment Office
MD-52
Research Triangle Park, NC  27711

Dr. David E. Weil
Environmental Criteria and
  and Assessment Office
MD-52
Research Triangle Park, NC  27711
                                       xi

-------
                                       PRELIMINARY DRAFT
                                     LIST OF ABBREVIATIONS
AAS
Ach
ACTH
AOCC
ADP/0 ratio
AIDS
AIHA
All
ALA
ALA-D
ALA-S
ALA-U
APDC
APHA
ASTM
ASV
ATP
B-cells
Ba
BAL
BAP
BSA
BUN
BW
C.V.
CaBP
CaEDTA
CBD
Cd
CDC
CEC
CEH
CFR
CMP
CNS
CO
COHb
CP-U

c8ah
D.F.
DA
DCMU
DDP
DNA
OTH
EEC
EEG
EMC
EP
EPA
Atomic absorption spectrometry
Acetyleneline
Adrenocoticotrophic hormone
Antibody-dependent cell-mediated cytotoxicity
Adenosine diphosphate/oxygen ratio
Acquired immune deficiency syndrome
American Industrial Hygiene Association
Angiotensin II
Aminolevulinic acid
Aminolevulinic acid dehydrase
Aminolevulinic acid synthetase
Aminolevulinic acid in urine
Ammoniurn pyrrolidine-di thi ocarbamate
American Public Health Association
Amercian Society for Testing and Materials
Anodic stripping voltammetry
Adenosine triphosphate
Bone marrow-derived lymphocytes
Barium
British anti-Lewisite (AKA dimercaprol)
benzo(a)pyrene
Bovine serum albumin
Blood urea nitrogen
Body weight
Coefficient of variation
Calcium binding protein
Calcium ethylenediaminetetraacetate
Central business district
Cadmium
Centers for Disease Control
Cation exchange capacity
Center for Environmental Health
reference method
Cytidine monophosphate
Central nervous system
Carbon monoxide
Carboxyhemoglobin
Urinary coproporphyrin
plasma clearance of p-aminohippuric acid
Copper
Degrees of freedom
Dopamine
[3-(3,4-dichlorophenyl)-l,l-dimetnylurea
Differential pulse polarography
Deoxyribonucleic acid
Delayed-type hypersensitivity
European Economic Community
Electroencephalogram
Encephalomyocarditis
Erythrocyte protoporphyri n
U.S. Environmental Protection Agency
TCPBA/D
                                           Xlt
                                                                9/20/83

-------
                                       PRELIMINARY DRAFT
                              LIST OF ABBREVIATIONS (continued).


FA                       Fulvic acid
FDA                      Food and Drug Administration
Fe                       Iron
FEP                      Free erythrocyte protoporphyrin
FY                       Fiscal year
G.M.                     Grand mean
G-6-PD                   Glucose-6-phosphate dehydrogenase
GABA     •                Gamma-aminobutyric acid
GALT                     Gut-associated lymphoid tissue
GC                       Gas chromatography
GFR                      Glomerular filtration rate
HA                       Humic acid
Hg                       Mercury
hi-vol                   High-volume air sampler
HPLC                     High-performance liquid chromatography
i.m.                     Intramuscular (method of injection)
i.p.                     Intraperitoneally (method of injection)
i.v.                     Intravenously (method of injection)
IAA                      Indol-3-ylacetic acid
IARC                     International Agency for Research on Cancer
ICO                      International classification of diseases
ICP                      Inductively coupled plasma
IDMS                     Isotope dilution mass spectrometry
IF                       Interferon
ILE                      Isotopic Lead Experiment (Italy)
IRPC                     International Radiological Protection Commission
K                        Potassium
LAI                      Leaf area index
LDH-X                    Lactate dehydrogenase isoenzyme x
LC50                     Lethyl concentration (50 percent)
LDgQ                     Lethal dose (50 percent)
LH                       Luteinizing hormone
LIPO                     Laboratory Improvement Program Office
In                       National logarithm
LPS                      Lipopolysaccharide
LRT                      Long range transport
mRNA                     Messenger ribonucleic acid
ME                       Mercaptoethanol
MEPP                     Miniature end-plate potential
MES                      Maximal electroshock seizure
MeV                      Mega-electron volts
MLC                      Mixed lymphocyte culture
MMD                      Mass median diameter
MMED                     Mass median equivalent diameter
Mn                       Manganese
MND                      Motor neuron disease
MSV                      Moloney sarcoma virus
MTD                      Maximum tolerated dose
n                        Number of subjects
N/A                      Not Available


                                          xiii

TCPBA/D                                                                                  9/20/83

-------
                                       PRELIMINARY DRAFT
                                     LIST OF ABBREVIATIONS
NA
NAAQS
NADB
NAMS
HAS
NASN
NBS
NE
NFAN
NFR-82
NHANES II
N1
OSHA
P
R
PAH
Pb
PBA
Pb(Ac)2
PbB   *
PbBrCl
PBG
PFC
PH
PHA
PHZ
PIXE
PMN
PND
PNS
ppra
PRA
PRS
PWM
Py-5-N
RBC
RBF
RCR
redox
RES
RLV
RNA
S-HT
SA-7
SCfll
S.D.
SDS
S.E.M.
SES
SCOT
Not Applicable
National ambient air quality standards
National Aerometric Data Bank
National Air Monitoring Station
National Academy of Sciences
National Air Surveillance Network
National Bureau of Standards
Norepinephrine
National Filter Analysis Network
Nutrition Foundation Report of 1982
National Health Assessment and Nutritional Evaluation Survey II
Nickel
Occupational Safety and Health Administration
Potassium
Significance symbol
Para-aminohippuric acid
Lead
Air lead
Lead acetate
concentration of lead in blood
Lead (II) bromochloride
Porphobilinogen
Plaque-forming cells
Measure of acidity
Phytohemaggluti ni n
Polyacrylamide-hydrous-zirconia
Proton-induced X-ray emissions
Polymorphonuclear leukocytes
Post-natal day
Peripheral nervous system
Parts per million
Plasma renin activity
Plasma renin substrate
Pokeweed mitogen
Pyrimide-5'-nucleotidase
Red blood cell; erythrocyte
Renal blood flow
Respiratory control ratios/rates
Oxidation-reduction potential
Reticuloendothelial system
Rauscher leukemia virus
Ribonucleic acid
Serotonin
Simian adenovirus
Standard cubic meter
Standard deviation
Sodium dodecyl sulfate
Standard error of the mean
Socioeconomic status
Serum glutamic oxaloacetic transaminase
                                             xiv
TCPBA/D
                                                                9/20/83

-------
                                       PRELIMINARY DRAFT
                              LIST OF ABBREVIATIONS (continued).
sig
SLAMS
SMR
Sr
SRBC
SRMs
STEL
SW voltage
T-cells
t- tests
TBL
TEA
TEL
TIBC
TML
TMLC
TSH
TSP
U.K.
UMP
USPHS
VA

V^R
WHO
Zn
ZPP
Surface immunoglobulin
State and local air monitoring stations
Standardized mortality ratio
Stronti urn
Sheep red blood cells
Standard reference materials
Short-term exposure limit
Slow-wave voltage
Thymus-derived lymphocytes
Tests of significance
tri-n-butyl lead
Tetraethy1-ammoni urn
Tetraethyllead
Total iron binding capacity
Tetrame thy Head
Tetramethyllead chloride
Thyroid-stimulating hormone
Total suspended particulate
United Kingdom
Uridine monophosphate
U.S. Public Health Service
Veterans Administration
Deposition velocity
Visual evoked response
World Health Organization
X-Ray fluorescence
Chi squared
Zinc
Erythrocyte zinc protoporphyrin
                                   MEASUREMENT ABBREVIATIONS
dl
ft
g
g/gal
g/ha-mo
km/hr
1/min
mg/km
mm
ng/cm2
nm
nM
sec
deciliter
feet
gram
gram/gallon
gram/hectare•month
kilometer/hour
liter/minute
mi 11igram/ki1ometer
microgram/cubic meter
millimeter
micrometer
nanograms/square centimeter
namometer
nanomole
second
                                             xv
TCPBA/D
                                                                9/20/83

-------

-------
PRELIMINARY DRAFT
1. EXECUTIVE SUMMARY AND CONCLUSIONS

1.1 INTRODUCTION
This criteria document evaluates and assesses scientific information on the health and
welfare effects associated with exposure to various concentrations of lead in ambient air.
According to Section 108 of the Clean Air Act of 1970, as amended in June 1974, a cri-
teria document for a specific pollutant or class of pollutants shall:
. . . accurately reflect the latest scientific knowledge useful in indicating
the kind and extent of all identifiable effects on public health or welfare which
may be expected from the presence of such pollutant in the ambient air, in varying
quantities.
Air quality criteria are of necessity based on presently available scientific data, which
in turn reflect the sophistication of the technology used in obtaining those data as well as
the magnitude of the experimental efforts expended. Thus air quality criteria for atmospheric
pollutants are a scientific expression of current knowledge and uncertainties. Specifically
air quality criteria are expressions of the scientific knowledge of the relationships between
various concentrations—averaged over a suitable time period—of pollutants in the same atmos-
phere and their adverse effects upon public health and the environment. Criteria are issued
as a basis for making decisions about the need for control of a pollutant and as a basis for
development of air quality standards governing the pollutant. Air quality criteria are
descriptive; that is, they describe the effects that have been observed to occur as a result
of external exposure at specific levels of a pollutant. In contrast, air quality standards
are prescriptive; that is, they prescribe what a political jurisdiction has determined to be
the maximum permissible exposure for a given time in a specified geographic area.
This criteria document is a revision of the previous Air Quality Criteria Document for
Lead (EPA-600/8-77-017) published in December, 1977. This revision is mandated by the Clean.
Air Act (Sect. 108 and 109), as amended UVS.C. §§7408 and 7409. The criteria document sets
forth what is known about the effects of lead contamination in the environment on human health
and welfare. This requires that the relationship between levels of exposure to lead, via all
routes and averaged over a suitable time period, and the biological responses to those levels
be carefully assessed. Assessment of exposure must take into consideration the temporal and
spatial distribution of lead and its various forms in the environment. Thus, the literature
through June, 1983, has been reviewed thoroughly for information relevant to air quality cri-
teria, for lead, but the document is not Intended as a complete and detailed review of all
literature pertaining to lead. Also, efforts are made to identify major discrepancies in "our
current knowledge and understanding of the effects of lead compounds.
SUMPB/D 1-1 9/30/83

-------
                                        PRELIMINARY DRAFT
     Lead  is  a naturally occurring  element that may be  found in the earth's crust and in all
components  of the biosphere.   It may  be found  in  water, soil, plants,  animals, and humans.
Because lead  also  occurs in ore  bodies that have been  mined for centuries by man, this metal
has also  been distributed  throughout the biosphere  by  the industrial activities of man.   Of
particular  importance  to the human  environment are  emissions  of  lead to'the atmosphere.  The
sources of  these emissions and the  pathways of  lead through the environment to man are shown
in  Figure  1-1.   This  figure   shows   natural  inputs   to  soil   by crustal  weathering  and
anthropogenic  inputs  to the  atmosphere  from automobile  emissions and  stationary industrial
sources.  Natural emissions of lead  to  the  atmosphere from volcanoes and windblown soil are of
minor importance.
SUMPB/D
                                          FECES  URINE
Figure 1-1. Pathways of lead exposure from the environment to man.

                  1-2
9/30/83

-------
PRELIMINARY DRAFT
From these emission sources, lead moves through the atmosphere to various components of
the human environment. Lead is deposited on soil and plants and in animals, becoming incor-
porated into the food chain of man. Atmospheric lead is a major component of household and
street dust; lead is also inhaled directly from the atmosphere.
1.2 ORGANIZATION OF DOCUMENT
This document focuses primarily on lead as found in its various forms in the ambient
atmosphere; in order to assess its effects on human health, however, the distribution and bio-
logical availability of lead in other environmental media have been considered. The rationale
for structuring the document was based primarily on the two major questions of exposure and
response. The first portion of the document is devoted to lead in the environment—its physi-
cal and chemical properties; the monitoring of lead in various media; sources, emissions, and
concentrations of lead; and the transport and transformation of lead within environmental
media. The latter portion is devoted to biological responses and effects on human health and
ecosystems.
In order to facilitate printing, distribution, and review of the present draft materials,
this First External Review Draft of the revised EPA Air Quality Criteria Document for Lead is
being released in four volumes. The first volume (Volume I) contains this executive summary
and conclusions chapter (Chapter 1) for the entire document. Volume II contains Chapters 2-8,
which include: the introduction for the document (Chapter 2); discussions of the above listed
topics concerning lead in the environment (Chapters 3-7); and evaluation of lead effects on
ecosystems (Chapter 8). The remaining two volumes contain Chapters 9-13, which deal with the
extensive available literature relevant to assessment of health effects associated with lead
exposure.
An effort has been made to limit the document to a highly critical assessment of the sci-
entific data base. The scientific literature has been reviewed through June 1983. The refer-
ences cited do not constitute an exhaustive bibliography of all available lead-related litera-
ture but they are thought to be sufficient to reflect the current state of knowledge on those
issues most relevant to the review of the air quality standard for lead.
The status of control technology for lead is not discussed in this document. For informa-
tion on the subject, the reader is referred to appropriate control technology documentation
published by the Office of Air Quality Planning and Standards (OAQPS), EPA. The subject of
"adequate margin of safety" stipulated in Section 108 of the Clean Air Act also is not expli-
citly addressed here; this topic will be considered in depth by EPA's Office of Air Quality
Planning and Standards in documentation prepared as a part of the process of revising the Na-
tional Ambient Air Quality Standard (NAAQS) for Lead.
SUMPB/D 1-3 9/30/83

-------
PRELIMINARY DRAFT
1.3 CHEMICAL AND PHYSICAL PROPERTIES OF LEAD
Lead is a gray-white metal of bright luster that, because of its easy isolation and low
melting point, was among the first of the metals to be extensively utilized by man. Lead was
used as early as 2000 B.C. by the Phoenicians. The most abundant ore is galena, from which
metallic lead is readily smelted. The metal is soft, malleable, and ductile, a poor
electrical conductor, and highly impervious to corrosion. This unique combination of physical
properties has led to its use in piping and roofing, and in containers for corrosive liquids.
The metal and the dioxide are used in storage batteries, and organolead compounds are used in
gasoline additives to boost octane levels. Since lead occurs in highly concentrated ores from
which it is readily separated, the availability of lead is far greater than its natural abun-
dance would suggest. The great environmental significance of lead is the result both of its
utility and of its availability.
The properties of organolead compounds (i.e., compounds containing bonds between lead and
carbon) are entirely different from those of the inorganic compounds of lead. Because of their
use as antiknock agents in gasoline and other fuels, the most important organolead compounds
have been the tetraalkyl compounds tetraethyllead (TEL) and tetramethyllead (TML). These lead
compounds are removed from internal combustion engines by a process called lead scavenging, in
which they react in the combustion chamber with halogenated hydrocarbon additives (notably
ethylene dibromide and ethylene dichloride) to form lead halides, usually bromochlorolead(II).
The donor atoms in a metal complex could be almost any basic atom or molecule; the pnly
requirement is that a donor, usually called a ligand, must have a pair of electrons available
for bond formation. In general, the metal atom occupies a central position in the complex, as
exemplified by the lead atom in tetramethyllead (Figure l-2a) which is tetrahedrally sur-
rounded by four methyl groups. In these simple organolead compounds, the lead is usually pre-
sent as Pb(IV), and the complexes are relatively inert. These simple ligands, which bind to
metal at only a single site, are called monodentate ligands. Some ligands, however, can bind
to the metal atom by more than one donor atom, so as to form a heterocyclic ring structure.
Rings of this general type are called chelate rings, and the donor molecules which form then
are called polydentate ligands or chelating agents. In the chemistry of lead, chelation nor-
mally involves Pb(II). A wide variety of biologically significant chelates with ligands
such as amino acids, peptides, and nucleotides are known. The simplest structure of this
type occurs with the ami no acid glycine, as represented in Figure l-2b for a 1:2
(metal:ligand) complex. The importance of chelating agents in the present context is their
widespread use in the treatment of lead and other metal poisoning.
SUMPB/D 1-4 9/30/83

-------
PRELIMINARY DRAFT
^

H3C
(a)
c

CH2
Pb
NH2
H2O

(b)
Figure 1-2. Metal complexes of lead.
^
Metals are often classified according to some combination of their electronegativity,
ionic radius, and formal charge. These parameters are used to construct empirical classifi-
cation schemes of relative hardness or softness. In these schemes, "hard" metals form strong
bonds with "hard" anlons and, likewise, "soft" metals bond with "soft" am'ons. Some metals
are borderline, having both soft and hard character. Pb(II), although borderline, demon-
strates primarily soft character (Figure 1-3). The term Class A may also be used to refer to
hard metals, and Class B to soft metals. Since Pb(II) is a relatively soft (or class B) metal
ion, it forms strong bonds to soft donor atoms like the sulfur atoms in the cysteine residues
of proteins and enzymes. In living systems, lead atoms bind to these peptide residues in pro-
teins, thereby changing the tertiary structure of the protein or blocking a substrate's
approach to the active site of an enzyme. This prevents the proteins from carrying out their
functions. As has been demonstrated in several studies (Jones and Vaughn, 1978; Williams and
Turner, 1981; Williams et al., 1982), there is an inverse correlation between the LD50 values
of metal complexes and the chemical softness parameter. Lead(II) has a higher softness para-
meter than either cadmium(II) or mercury(II), so lead(II) compounds would not be expected to
be as toxic as their cadmium or mercury analogues.
The role of the chelating agents is to compete with the peptides for the metal by forming
stable chelate complexes that can be transported from the protein and eventually be excreted
by the body. For simple thermodynamic reasons, chelate complexes are much more stable than
nonodentate metal complexes, and it is this enhanced stability that is the basis for their
ability to compete favorably with proteins and other ligands for the metal ions.
SUMPB/D
1-5
9/30/83

-------
                                       PRELIMINARY DRAFT
CLASS B OR COVALENT INDEX, X*mr
a.u
i
4.5
4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
0
• I I I I I ! I I " I " I
t AiT
t t
• AS- w .
.••Pi" •Bi" •
_ • Pb(IV) 	
^Ti- Hg"
• Ti"
_0Cu' CLASS B _-
*Pb" •Sbllll)
~ cT* *cu" A««»" "
,.,ji>co" |n" * •
— * •«!" • • F«" SndVI 	
Cr"
— Mn"» v" Ga' * BORDERLINE ~~
_ ^ Qd" LU" —
C»' Ba" • • y?. Al"
AK- •••Ca" L'" -
\N* Sr" •
• Ba"
~ L' CLASS A
1 1 1 1 1 1 1 1 ,,! .,1
                     0    2    4     6    8    10   12   14    16    20    23
                                    CLASS A OR IONIC INDEX. Z*/r

                             Figure 1-3. Softness parameters of metals.
                             Source:  Nieboer and Richardson (1980).
     It should  be noted that both  the stoichiometry  and structures of metal  chelates  depend
upon pH, and that structures  different from those manifest in solution may occur in crystals.
It will suffice  to  state,  however,  that several ligands can be found that are capable of suf-
ficiently strong  chelation with  lead  present  in  the body under physiological  conditions  to
permit their use in the effective treatment of lead poisoning.
1.4  SAMPLING AND'ANALYTICAL METHODS FOR ENVIRONMENTAL LEAD
     Lead, like all criteria  pollutants,  has a designated Reference Method for monitoring and
analysis  as  required  in State  Implementation  Plans for determining compliance with  the  lead
National Ambient Air Quality  Standard.   The Reference Method  uses  a high volume  sampler  (hi-
vol) for sample collection and atomic absorption spectrometry (AAS) for analysis.
     For  a  rigorous  quality  assurance  program,  it is essential that  investigators  recognize
all sources of contamination and use every precaution to eliminate  them.   Contamination occurs
SUMPB/D
1-6
9/30/83

-------
PRELIMINARY DRAFT

on the surfaces of collection containers and devices, on the hands and clothing of the inves-
tigator, in the chemical reagents, in the laboratory atmosphere, and on the labware and tools
used to prepare the sample for analysis.

1.4.1 Sampling Techniques
Sampling strategy encompasses site selection, choice of instrument used to obtain repre-
sentative samples, and choice of method used to preserve sample integrity. In the United States,
some sampling stations for air pollutants have been operated since the early I9601s. These
early stations were a part of the National Air Surveillance Network (NASN), which has now
become the National Filter Analysis Network (NFAN). Two other types of networks have been
established to meet specific data requirements. State and Local Air Monitoring Stations
(SLAMS) provide data from specific areas where pollutant concentrations and population densi-
ties are the greatest and where monitoring of compliance to standards is critical. The Na-
tional Air Monitoring Station (NAMS) network is designed to serve national monitoring needs,
including assessment of national ambient trends. SLAMS and NAMS stations are maintained by
state and local agencies and the air samples are analyzed in their laboratories. Stations in
the NFAN network are maintained by state and local agencies, but the samples are analyzed by
laboratories in the U.S. Environmental Protection Agency, where quality control procedures are
rigorously maintained.
Data from all three networks are combined into one data base, the National Aerometric
Data Bank (NADB). These data may be individual chemical analyses of a 24-hour sampling period
arithmetically averaged over a calendar period, or chemical composites of several filters used
to determine a quarterly composite. Data are occasionally not available for a given location
because they do not conform to strict statistical requirements.
In September, 1981, EPA promulgated regulations establishing ambient air monitoring and
data reporting requirements for lead comparable to those already established in May of 1979
for the other criteria pollutants. Whereas sampling for lead is accomplished when -sampling
for total suspended particulate (TSP), the designs of lead and TSP monitoring stations must be
complimentary to insure compliance with the NAMS criteria for each pollutant.
There must be at least two SLAMS sites for lead in any area that has a population greater
than 500,000 and any area where lead concentration currently exceeds the ambient lead standard
(1.5 ug/m3) or has exceeded it since January 1, 1974.
To clarify the relationship between monitoring objectives and the actual siting of a mon-
itor, the concept of a spatial scale of representativeness was developed. The spatial scales
are discribed in terms of the physical dimensions of the air space surrounding the monitor
throughout which pollutant concentrations are fairly similar.

SUMPB/D 1-7 9/30/83

-------
PRELIMINARY DRAFT
The time scale may also be an important factor. Siting criteria must include sampling
times sufficiently long to include average windspeed and direction, or a sufficient number of
samples must be collected over short sampling periods to provide an average value consistent
with a 24-hour exposure.
Airborne lead is primarily inorganic particulate matter but may occur in the form of or-
ganic gases. Devices used for collecting samples of ambient atmospheric lead include the
standard hi-vol sampler and a Variety of other collectors employing filters, impactors,
impingegers, or scrubbers, either separately or in combination, that measure lead in ug/m3.
2
Some samplers measure lead deposition expressed in ug/cm ; some instruments separate parti-
cles by size. As a general rule, particles smaller in aerodynamic diameter than 2.5 urn are
classified as "fine", and those larger than 2.5 urn as "coarse."
The present SLAMS and NAMS employ the standard hi-vol sampler (U.S. Environmental Protec-
tion Agency, 1971) as part of their sampling networks. As a Federal Reference Method Sampler,
the hi-vol operates with a specific flow rate of 1600 to 2500 m3 of air per day
When sampling ambient lead with systems employing filters, it is likely that vapor-phase
organolead compounds will pass through the filter media. The use of bubblers downstream from
the filter containing a suitable reagent or absorber for collection of these compounds has
been shown to be effective. Organolead may be collected on iodine crystals, adsorbed on acti-
vated charcoal, or absorbed in an iodine monochloride solution. In one experiment, Purdue et
al. (1973) operated two bubblers in series containing iodine monochloride solution. One hun-
dred percent of the lead was recovered in the first bubbler.
Sampling of stationary sources for lead requires the use of a sequence of samplers in the
smokestack. Since lead in stack emissions may be present in a variety of physical and chemical
forms, source sampling trains must be designed to trap and retain both gaseous and particulate
lead.
Three principal procedures have been used to obtain samples of auto exhaust aerosols for
subsequent analysis for lead compounds: a horizontal dilution tunnel, plastic sample collec-
tion bags, and a low residence time proportional sampler. In each procedure, samples are air
diluted to simulate roadside exposure conditions. In the most commonly used procedure, the
air dilution tube segregates fine combustion-derived particles from larger lead particles.
Such tunnels of varying lengths have been limited by exhaust temperatures to total flows above
approximately 11 mVmin. Similar tunnels have a centrifugal fan located upstream, rather than
a positive displacement pump located downstream. This geometry produces a slight positive
pressure in the tunnel and expedites transfer of the aerosol to holding chambers for studies
of aerosol growth. However, turbulence from the fan may affect the sampling efficiency.
Since the total exhaust plus dilution airflow is not held constant in this system, potential
errors can be reduced by maintaining a very high dilution air/exhaust flow ratio.
SUMPB/D 1-8 9/30/83

-------
PRELIMINARY DRAFT

In the bag technique, auto emissions produced during simulated driving cycles are air-
diluted and collected in a large plastic bag. The aerosol sample is passed through a filtra-
tion or impact!on sampler prior to lead analysis. This technique may result in errors of
aerosol size analysis because of condensation of low vapor pressure organic substances onto
the lead particles.
To minimize condensation problems, a third technique, a low residence time proportional
sampling system, has been used. It is based on proportional sampling of raw exhaust, again
diluted with ambient air followed by filtration or impaction. Since the sample flow must be a
constant proportion of the total exhaust flow, this technique may be limited by the response
time of the equipment to operating cycle phases that cause relatively small transients in the
exhaust flow rate.
Other primary environmental media that may be affected by airborne lead include precipi-
tation, surface water, soil, vegetation, and foodstuffs. The sampling plans and the sampling
methodologies used in dealing with these media depend on the purpose of the experiments, the
types of measurements to be carried out, and the analytical technique to be used.
Lead at the start of a rain event is higher in concentration than at the end, and rain
striking the canopy of a forest may rinse dry deposition particles from the leaf surfaces.
Rain collection systems should be designed to collect precipitation on an event basis and to
collect sequential samples during the event.
Two automated systems have recently been used. The Sangamo Precipitation Collector,
Type A, collects rain in a single bucket exposed at the beginning of the rain event (Samant
and Vaidya, 1982). A second sampler, described by Coscio et al. (1982), also remains covered
between rain events; it can collect a sequence of eight samples during the period of rain and
may be fitted with a refrigeration unit for sample cooling.
Because the physicochemical form of lead often influences environmental effects, there is
a need to differentiate among the various chemical forms. Complete differentiation among all
such forms is a complex task that has not yet been fully accomplished. The most commonly used
approach is to distinguish between dissolved and suspended forms of lead. All lead passing
through a 0.45 UN membrane filter is operationally defined as dissolved, while that retained
on the filter is defined as suspended (Kopp and McKee, 1979).
Containers used for sample collection and storage should be fabricated from essentially
lead-free plastic or glass, e.g., conventional polyethylene, Teflon , or quartz. These con-
tainers must be leached with hot acid for several days to ensure minimum lead contamination
(Patterson and Settle, 1976).
The distance from emission sources and depth gradients associated with lead in soil must
be considered in designing the sampling plan. Vegetation, litter, and large objects such as

SUMPB/D 1-9 9/30/83

-------
PRELIMINARY DRAFT
stones should not be included in the sample. Depth samples should be collected at not greater
than 2 cm intervals to preserve vertical integrity.
Because most soil lead is in chemical forms unavailable to plants, and because lead is
not easily transported by plants, roots typically contain very little lead and shoots even
less. Before analysis, a decision must be made as to whether or not the plant leaf material
should be washed to remove surface contamination from dry deposition and soil particles. If
the plants are sampled for total lead content (e.g., if they serve as animal food sources),
they cannot be washed; if the effect of lead on internal plant processes is being studied, the
plant samples should be washed. In either case, the decision must be made at the time of sam-
pling, as washing cannot be effective after the plant materials have dried.
In sampling for airborne lead, air is drawn through filter materials such as glass fiber,
cellulose acetate, or porous'plastic. These materials often include contaminant lead that can
interfere with the subsequent analysis. Procedures for cleaning filters to reduce the lead
blank rely on washing with acids or complexing agents. The type of filter and the analytical
method to be used often determines the ashing technique. In some methods, e.g., X-ray fluo-
rescence, analysis can be performed directly on the filter if the filter material is suitable.
Skogerboe (1974) provided a general review of filter materials.
The main advantages of glass fiber filters are low pressure drop and high particle col-
lection efficiency at high flow rates. The main disadvantage is variability in the lead blank,
which makes their use inadvisable in many cases. This has placed a high priority on the stan-
dardization of a suitable filter for hi-vol samples. Other investigations have indicated,
however, that glass fiber filters are now available that do not present a lead interference
®
problem (Scott et al., 1976b). Teflon filters have been used since 1975 by Dzubay et al.
(1982) and Stevens et al. (1978), who have shown these filters to have very low lead blanks
(<2 ng/cm2). The collection efficiencies of filters, and also of impactors, have been shown
to be dominant factors in the quality of the derived data.

1.4.2 Analytical Procedures
The choice of analytical method depends on the nature of the data required, the type of
sample being analyzed, the skill of the analyst, and the equipment available. For general
determination of elemental lead, atomic absorption spectroscopy (AAS) is widely used and re-
commended (C.F.R., 1982 40: § 50). Optical emission spectrotoetry and X-ray fluorescence
(XRF) are rapid and inexpensive methods for multielemental analyses. X-ray fluorescence can
measure lead concentrations reliably to 1 ng/m3 using samples collected with commercial
dichotomous samplers. Other analytical methods have specific advantages appropriate for
special studies.

SUMPB/D 1-10 9/30/83

-------
PRELIMINARY DRAFT

With respect to measuring lead without contamination during sampling or from the labora-
tory, several investigators have shown that the magnitude of the problem is quite large. It
appears that the problem may be caused by failure to control the blank or by failure to stan-
dardize instrument operation (Patterson, 1983; Skogerboe, 1982). The laboratory atmosphere,
collecting containers, and the labware used may be primary contributors to the lead blank
problem (Patterson, 1983; Skogerboe, 1982). Failure to recognize these and other sources of
contamination such as reagents and hand contact is very likely to result in the generation of
artificially high analytical results. Samples with less than 100 ng lead should be analyzed
in a clean laboratory especially designed for the elimination of lead contamination. Moody
(1982) has described the construction and application of such a laboratory at the National
Bureau of Standards.
For AAS, the lead atoms in the sample must be vaporized either in a precisely controlled
flame or in a furnace. Furnace systems in AAS offer high sensitivity as well as the ability
to analyze small samples. These enhanced capabilities are offset in part by greater dif-
ficulty in analytical calibration and by loss of analytical precision.
Particles may also be collected on cellulose acetate filters. Disks (0.5 cm2) are
punched from these filters and analyzed by Insertion of nichrome cups containing the disks
into a flame. Another application involves the use of graphite cups as particle filters with
the subsequent analysis of the cups directly in the furnace system. These two procedures
offer the ability to determine particulate lead directly with minimal sample handling.
In an analysis using AAS and hi-vol samplers, atmospheric concentrations of lead were
found to be 0.076 ng/m3 at the South Pole (Maenhaut et al., 1979). Lead analyses of 995 par-
ticulate samples from the NASN were accomplished by AAS with an indicated precision of 11 per-
cent (Scott et al., 1976a). More specialized AAS methods for the determination of tetraalkyl
lead compounds in water and fish tissue have been described by Chau et al. (1979) and in air
by Birnie and Noden (1980) and Rohbock et al. (1980).
Techniques for AAS are still evolving. An alternative to the graphite furnace, evaluated
by Jin and Taga (1982), uses a heated quartz tube through which the metal ion in gaseous
hydride form flows continuously. Sensitivities were 1 to 3 ng/g for lead. The technique is
similar to the hydride generators used for mercury, arsenic, and selenium. Other nonflane
atomiration systems, electrodeless discharge lamps, and other equipment refinements and tech-
nique developments have been reported (Horlick, 1982).
Optical emission spectroscopy is based on the measurement of the light emitted by
elements when they are excited in an appropriate energy medium. The technique has been used
to determine the lead content of soils, rocks, and minerals at the 5 to 10 M9/9 level with a
relative standard deviation of 5 to 10 percent; this method has also been applied to the ana-
lysis of a large number of air samples (Sugimae and Skogerboe, 1978). The primary advantage
SUMP8/D 1-11 9/30/83

-------
PRELIMINARY DRAFT
of this method is that it allows simultaneous measurement of a large number of elements in a
small sample. In a study of environmental contamination by automotive lead, sampling times
were shortened by using a sampling technique in which lead-free porous graphite was used both
as the filter medium and as the electrode in the spectrometer. Lead concentrations of 1 to
10 n9/ro3 were detected after a half-hour flow at 800 to 1200 ml/min through the filter.
More recent activities have focused attention on the inductively coupled plasma (ICP)
system as a valuable means of excitation and analysis (Garbarino and Taylor, 1979). The ICP
system offers a higher degree of sensitivity with less analytical interference than is typical
of many of the other emission spectroscopic systems. Optical emission methods are inefficient
when used for analysis of a single element, since the equipment is expensive and a high level
of operator training is required. This problem is largely offset when analysis for several
elements is required, as is often the case for atmospheric aerosols. X-ray fluorescence (XF)
allows simultaneous identification of several elements, including lead, using a high-energy
irradiation source. With the X-ray tubes coupled with fluorescers, very little energy is
transmitted to the sample; thus sample degradation is kept to a minimum. Electron beams and
radioactive isotope sources have been used extensively as energy sources for XRF analysis.
X-ray emission induced by charged-particle excitation (proton-induced X-ray emission or
PIXE) offers an attractive alternative to the more common techniques. The excellent capabi-
lity of accelerator beams for X-ray emission analysis is partially due to the relatively low
background radiation associated with the excitation.
X-radiation is the basis of the electron microprobe method of analysis. When an intense
electron beam is incident on a sample, it produces several forms of radiation, including
X-rays, whose wavelengths depend on the elements present in the material and whose intensities
depend on the relative quantities of these elements. The method is unique in providing com-
positional information on individual lead particles, thus permitting the study of dynamic che-
mical changes and perhaps allowing improved source identification.
Isotope dilution mass spectrometry (IDMS) is the most accurate measurement technique
known at the present time. No other techniques serve more reliably as a comparative refer-
ence; it has been used for analyses of subnanogram concentrations of lead in a variety of sam-
ple types (Chow et al., 1969, 1974; Facchetti and Geiss, 1982; Hirao and Patterson, 1974;
Murozumi et al., 1969; Patterson et al., 1976; Rabinowitz et al., 1973). The isotopic compo-
sition of lead peculiar to various ore bodies and crustal sources may also be used as a means
of tracing the origin of anthropogenic lead.
Colorimetric or spectrophotometric analysis for lead using dithizone (diphenylthiocarba-
zone) as the reagent has been used for many years. It was the primary method recommended by a
National Academy of Sciences (1972) report on lead, and the basis for the tentative method of

SUMPB/D 1-12 9/30/83

-------
PRELIMINARY DRAFT
testing for lead in the atmosphere by the American Society for Testing Materials (1975b).
Prior to the development of the IDMS method, colorimetHc analysis served as the reference by
which other methods were tested.
Analytical methods based on electrochemical phenomena are found in a variety of forms.
They are characterized by a high degree of sensitivity, selectivity, and accuracy derived from
the relationship between current, charge, potential, and time for electrolytic reactions in
solutions. Anodic stripping voltammetry (ASV) is a two step process in which the lead is pre-
concentrated onto a mercury electrode by an extended but selected period of reduction. After
the reduction step, the potential is scanned either linearly or by differential pulse to oxi-
dize the lead and allow measurement of the oxidation (stripping) current.
The majority of analytical methods are restricted to measurement of total lead and cannot
directly identify the various compounds of lead. Gas chromatography (GC) using the electron
capture detector has been demonstrated to be useful for organolead compounds. The use of
atomic absorption as the GC detector for organolead compounds has been described by De Jonghe
et al. (1981), while a plasma emission detector has been used by Estes et al. (1981). In ad-
dition, Messman and Rains (1981) have used liquid chromatography with an atomic absorption
detector to measure organolead compounds. Mass spectrometry may also be used with gas chroma-
tography (Mykytiuk et al., 1980).
1.5 SOURCES AND EMISSIONS
The history of global lead emissions has been assembled from chronological records of de-
position in polar snow strata, marine and freshwater sediments, and the annual rings of trees.
These records aid in establishing natural background levels of lead in air, soils, plants,
animals, and humans, and they document the sudden increase in atmospheric lead at the time of
the industrial revolution, With a later burst during the 1920's when lead-alkyls were first
added to gasoline. Pond sediment analyses have shown a 20-fold increase in lead deposition
during the last 150 years (Figure 1-4), documenting not only the increasing use of lead since
the beginning of the industrial revolution in western United States, but also the relative
fraction of natural vs. anthropogenic lead inputs. Other studies have shown the same magni-
tude of increasing deposition in freshwater marine sediments. The pond and marine sediments
also document the shift in isotopic composition of atmospheric caused by increased commercial
use of the New Lead Belt in Missouri, where the ore body has an isotopic composition substan-
tially different from other ore bodies of the world.
Perhaps the best chronological record is that of the polar ice strata of Murozumi et al.
(1969), which extends nearly three thousand years back in time (Figure 1-4). At the South

SUMPB/D 1-13 9/30/83

-------
                                       PRELIMINARY DRAFT
        1.0

        0.9

        0.8

        0.7

        o.e

        0.5

        0.4

        0.3

        0.2

        0.1
         1750     1775     1800      1825     1850     1875
                                               YEAR
                 1900
1925
1950
1976
          Figure 1-4. Chronological record of the relative increase of lead in snow strata, pond
         and lake sediments, marine sediments, and tree rings. The data are expressed as a
         ratio of the latest year of the record and should not be interpreted to extend back in
         time to natural or uncontaminated levels of lead concentration.
         Source: Adapted from Murozumi et al. (1969) (O), Shirahata et al. (1980) (D), Edgington
         and Bobbins (1976) (A), Ng and Patterson (1982) ( A), and Rolfe (1974) ( • ).
Pole, Boutron  (1982)  observed a 4-fold Increase of lead in snow from  1957 to 1977 but saw no
increase during  the  period 1927 to 1957.   The author suggested the extensive atmospheric lead
pollution which  began in the 1920's did not  reach the  South Pole until the nrid-1950's.  This
interpretation  agrees with that  of Haenhaut  et al. (1979), who  found  atmospheric concentra-
tions of  lead of  0.000076 pg/m3 at  the  same  location.   This concentration  is about 3-fold
higher  than  the 0.000024  pg/m3  estimated by  Patterson  (1980) and Servant (1982)  to be the
natural lead concentration  in the atmosphere.  In summary, it is likely that atmospheric lead
emissions have  increased 2000-fold  since  the pre-Roman  era,  that even  at this early time the
atmosphere may  have  been contaminated by a factor of three over natural  levels (Murozumi  et
al.  1969),  and that  global  atmospheric  concentrations  have  increased  dramatically  since the
1920's.
SUMPB/D
1-14
                  9/30/83

-------
                                        PRELIMINARY DRAFT
     The  history  of global  emissions may  also be  inferred from  total production  of lead.
The historical picture of  lead production has been  pieced together from many sources by Settle
and  Patterson (1980)  (Figure 1-5).    Until  the industrial  revolution, lead  production  was
determined  largely by the ability  or desire to  mine  lead for its silver content.  Since that
time,  lead, has  been used as an  industrial  product in  its own right,  ana  efforts to improve
smelter  efficiency,  including control  of stack  emissions and fugitive  dusts,  have made lead
production  more  economical.  This  improved  efficiency  is not reflected in the chronological
record  because  of  atmospheric   emissions  of  lead   from many  other  anthropogenic  sources,
especially  gasoline  combustion (see  Section  5.3.3).   From this knowledge of the chronological
record,  it  is possible to sort out contemporary anthropogenic  emissions from natural sources
of atmospheric lead.
                                                                   SPANISH PRODUCTION
                                                                       OF SILVER
                                                                     IN NEW WORLD
                                                                           INDUSTRIAL
                                                                           REVOLUTION
                                                        EXHAUSTION
                                                         OF ROMAN
                                                        LEAD MINES
            SILVER
          PRODUCTION
          IN GERMANY
                                             INTRODUCTION
                                              OF COINAGE
               DISCOVERY OF
               CUPELLATION
ROMAN REPUBLIC
  AND EMPIRE
                                            RISE AND FALL
                                             OF ATHENS
        10°
            5500   5000   4500   4000   3500   3000  2500   2000   1500   1000   500     0

                                      YEARS BEFORE PRESENT

          Figure 1-5. The global lead production has changed historically in response to
         major economic and political events. Increases in  lead production (note log
         scale) correspond approximately to historical increases in lead emissions shown
         in Figure 5-1.

         Source:  Adapted from Settle and Patterson (1980).

SUHPB/D                                     1-15
                               9/30/83

-------
PRELIMINARY DRAFT
Lead enters the biosphere from lead-bearing minerals in the lithosphere through both
natural and man-made processes. Measurements of soil materials taken at 20-cm depths in the
continental United States show a median lead concentration of IS to 16 ug Pb/g soil. In
natural processes, lead is first incorporated in soil in the active root zone, from which it
may be absorbed by plants, leached into surface waters, or eroded into windborne dusts.
Calculations of natural contributions using geochemical information indicate that natural
sources contribute a relatively small amount of lead to the atmosphere. It has been estimated
from geochemical evidence that the natural particulate lead level is less than 0.0005 ug/m3
(National Academy of Sciences, 1980), and probably lower than the 0.000076 ug/m3 measured at
the South Pole (Maenhaut et al., 1979). In contrast, average lead concentrations in urban
suspended particulate matter range as high as 6 M9/m3 (U.S. Environmental Protection Agency,
1979, 1978). Evidently, most of this urban particulate lead originates from man-made sources.
Lead occupies an important position in the U.S. economy, ranking fifth among all metals
in tonnage used. Approximately 85 percent of the primary lead produced in this country is
from native mines, although often associated with minor amounts of zinc, cadmium, copper,
bismuth, gold, silver, and other minerals (U.S. Bureau of Mines, 1972-1982). Missouri lead
ore deposits account for approximately 80 to 90 percent of the domestic production. Total
utilization averaged approximately 1.36x10 t/yr over the 10-year period, with storage bat-
teries and gasoline additives accounting for ~70 percent of total use. Certain products,
especially batteries, cables, plumbing, weights, and ballast, contain lead that is
economically recoverable as secondary lead. Lead in pigments, gasoline additives, ammunition,
foil, solder, and steel products is widely dispersed and therefore is largely unrecoverable.
Approximately 40-50 percent of annual lead production is recovered and eventually recycled.
Lead or its compounds may enter the environment at any point during mining, smelting,
processing, use, recycling, or disposal. Estimates of the dispersal of lead emissions into
the environment by principal sources indicate that the atmosphere is the major initial
recipient. Estimated lead emissions to the atmosphere are shown in Table 1-1. Mobile and
stationary sources of lead emissions, although found throughout the nation, tend to be con-
centrated in areas of high population density, and near smelters. Figure 1-6 shows the ap-
proximate locations of major lead mines, primary and secondary smelters and refineries, and
alkyl lead paints (International Lead Zinc Research Organization, 1982).
The majority of lead compounds found in the atmosphere result from leaded gasoline com-
bustion. Several reports indicate that transportation sources contribute over 80 percent of
the total atmospheric lead. Other mobile sources, including aviation use of leaded gasoline
and diesel and jet fuel combustion, contribute insignificant lead emissions to the atmosphere.
Automotive lead emissions occur as PbBrCl in fresh exhaust particles. The fate of emit-
ted lead particles depends upon particle size. Particles initially formed by condensation of
SUMPB/D 1-16 9/30/83

-------
                                       PRELIMINARY DRAFT
                   TABLE 1-1.  ESTIMATED ATMOSPHERIC LEAD EMISSIONS FOR THE
                               UNITED STATES, 1981 AND THE WORLD
Source Category
Gasoline combustion
Waste oil combustion
Solid waste disposal
Coal combustion
Oil combustion
Wood combustion

Gray iron production
Iron and steel production
Secondary lead smelting
Primary copper smelting
Ore crushing and grinding
Primary lead smelting
Other metallurgical
Zn smelting
Ni smelting
Lead alkyl manufacture
Type metal
Portland cement production
Miscellaneous
Total
Annual
U.S.
Emissions
(t/yr)
35,000
830
319
950
226
—

295
533
631
30
326
921
54


245
85
71
233
40,739a
Percentage of
U.S. Total
Emissions
85.9
2.0
0.8
2.3
0.6
--
\
0.7
1.3
1.5
0.1
0.8
2.3
0.1


0.6
0.2
0.2
0.5
100%
Annual
Global
Emissions
(t/yr)
273,000
8,900

14,000
6,000
4,500

50,000

770
27,000
8,200
31,000

16,000
2,500

7,400

5,900
449,170
 Inventory does not include emissions from exhausting workroom air, burning of lead-painted
 surfaces, welding of lead-painted steel structures, or weathering of painted surfaces.
Source:  For U.S. emissions, Battye (1983); for global emissions, Nriagu (1979).

lead compounds in the combustion gases are quite small (well under 0.1 um in diameter).  Parti-
cles in this size category are subject to growth by coagulation and, when airborne, can remain
suspended in the atmosphere for 7 to 30 days and travel thousands of miles from their original
source.   Larger  particles are formed  as  the  result of agglomeration  of smaller condensation
particles and have limited atmospheric lifetimes.
     During the lifetime of the vehicle, approximately 35 percent of the lead contained in the
gasoline  burned  by  the  vehicle  will  be emitted  as small particles  [<0.25 um  mass  median
equivalent diameter (MMED)],  and  approximately 40 percent will be emitted as larger particles
SUMPB/D
1-17
9/30/83

-------
                                  • MINES (IS)
                                  A SMELTERS AND REFINERIES (7) '
                                  O SECONDARY SMELTERS AND REFINERIES (56)
                                  • LEAD ALKYL PLANTS (4)
Figure 1-6.  Locations of major lead operations in the United States.

Source: International Lead Zinc Research Organization (1982).

-------
PRELIMINARY DRAFT

(>10 |jm MMED) (Ter Haar et al., 1972). The remainder of the lead consumed in gasoline combus-
tion is deposited in the engine and exhaust system.
Although the majority (>90 percent on a mass basis) of vehicular lead compounds are emit-
ted as inorganic particles (e.g., PbBrCl), some organolead vapors (e.g., lead alkyls) are also
emitted. The largest volume of organolead vapors arises from the manufacture, transport, and
handling of leaded gasoline. Such vapors are photoreactive, and their presence in local atmo-
spheres is transitory. Organolead vapors are most likely to occur in occupational settings
and have been found to contribute less than 10 percent of the total lead present in the atmo-
sphere.
The use of lead additives in gasoline, which increased in volume for many years, is now
decreasing as automobiles designed to use unleaded fuel constitute the major portion of the
automotive population. The decline in the use of leaded fuel is the result of two regulations
promulgated by the U.S. Environmental Protection Agency (F.R., 1973 Decembers). The first
required the availability of unleaded fuel for use in automobiles designed to meet federal
emission standards with lead-sensitive emission control devices (e.g., catalytic converters);
the second required a reduction or phase-down of the lead content in leaded gasoline. Compli-
ance with the phase-down of lead in gasoline has recently been the subject of proposed rule-
makings. The final action (F.R., 1982 October 29) replaced the present 0.5 g/gal standard for
the average lead content of all gasoline with a two-tiered standard for the lead content of
leaded gasoline. Under this proposed rule, refineries would be required to meet a standard of
1.10 g/gal for leaded gasoline while maintaining an average 0.5 g/gal for all gasoline.
The trend in lead content for U.S. gasolines is shown in Figure 1-7. Of the total gas-
oline pool, which includes both leaded and unleaded fuels, the average lead content has
decreased 63 percent, from an average of 1.62 g/gal in 1975 to 0.60 g/gal in 1981.
Data describing the lead consumed in gasoline and average ambient lead levels (composite
of maximum quarterly values) versus calendar year are plotted in Figure 1-8. The linear cor-
relation between lead consumed in gasoline and the composite maximum average quarterly ambient
average lead level is very good. Between 1975 and 1980, the lead consumed in gasoline
decreased 52 percent (from 165,577 metric tons to 78,679 metric tons) while the corresponding
composite maximum quarterly average of ambient lead decreased 51 percent (from 1.23 |jg/m3 to
0.60 HQ/in3)- This indicates that control of lead in gasoline over the past several years has
effected a direct decrease in peak ambient lead concentrations.
Furthermore, the equation in Figure 1-8 implies that the complete elimination of lead
from gasoline might reduce the composite average of the maximum quarterly lead concentrations
at these stations to 0.05 pg/m3, a level typical of concentrations reported for nonurban sta-
tions in the U.S.

SUMPB/D 1-19 9/30/83

-------
                                          PRELIMINARY DRAFT
               2.40
               2.00
           i
               1.50
           (9
               1.00

               0.60
               0.00
                               1111(1
                                                    LEADED FUEL
                       SALES-WEIGHTED TOTAL
                       GASOLINE POOL
                       (LEADED AND UNLEADED
                       "AVERAGE")
                             UNLEADED FUEL
                      *       t       f      t       t
                     1976     197V     1977     1978     1979     1980     1981    1982*

                                             CALENDAR YEAR

          Figure 1-7.  Trend in lead content of U.S. gasolines, 1975-1982. (DuPont, 1982).

          •1982 DATA ARE FORECASTS.
SUNPB/D
1-20
9/30/83

-------
                                       PRELIMINARY  DRAFT
           180
           160
           140
       1   12°

       &
       u*
       5   100
       (9
       Z
       £   80

       to
       8

       a   "
           40
           20
                                 I
                   AVERAGE Pfo - 6.93 x 10* (Pb CONSUMED) + 0.05
                            r2 = 0.99
                                                                    1978
                                                     19791
1981
                                                                         1975
            •1980
                                    1982
                       I
  I
I
                     0.20       0.40      0.60      0.80       1.00      1.20

                  COMPOSITE MAXIMUM QUARTERLY AVERAGE LEAD LEVELS,
     Figure 1-8. Relationship between lead consumed in gasoline and composite maximum
     quarterly average lead levels, 1975-1980.
     •1981 AND 1982 DATA ARE ESTIMATES.
SUMPB/D
             1-21
                          9/30/83

-------
PRELIMINARY DRAFT
Solid waste incineration and combustion of waste oil are principal contributors of lead
emissions from stationary sources. The manufacture of consumer products such as lead glass,
storage batteries, and lead additives for gasoline also contributes significantly to station-
ary source lead emissions. Since 1970, the quantity of lead emitted from the metallurgical
industry has decreased somewhat because of the application of control equipment and the clos-
ing of several plants, particularly in the zinc and pyrometallurgical industries.
A new locus for lead emissions emerged in the mid-1960s with the opening of the "Viburnum
Trend" or "New Lead Belt" in southeastern Missouri. The presence of ten mines and three ac-
companying lead smelters in this area makes it the largest lead-producing district in the
world.
There is no doubt that atmospheric lead has been a component of the human environment
since the earliest written record of civilization. Atmospheric emissions are recorded in
glacial ice strata and pond and lake sediments. The history of global emissions seems
closely tied to production of lead by industrially oriented civilizations. Although the
amount of lead to the atmosphere emitted from natural sources is a subject of controversy,
even the most liberal estimate (25 x 10? t/year) is dwarfed by the global emissions from
anthropogenic sources (450 X 10? t/year). The contribution of gasoline lead to total atmo-
spheric emissions has remained high, at 85 percent, as emissions from stationary sources have
decreased at the same pace as from mobile sources. The decrease in stationary source emis-
sions is due primarily to control of stack emissions, whereas the decrease in mobile source
emissions is a result of switchover to unleaded gasolines. Production of lead in the
United States has remained steady at about 1.2 X 10? t/year for the past decade. The gasoline
additive share of this market has dropped from 18 to 9.5 percent during the period 1971 to
1981. The decreasing use of lead in gasoline is projected to continue through 1990.
1.6 TRANSPORT AND TRANSFORMATION
At any particular location and time, the concentration of lead found in the atmosphere
depends on the proximity to the source, the amount of lead emitted from sources, and the de-
gree of mixing provided by the motion of the atmosphere. At the source, lead emissions are
typically around'10,000 ug/m?, while lead values in city air are usually between 0.1 and 10
ug/m3. These reduced concentrations are the result of dilution of effluent gas with clean air
and the removal of particles by wet or dry deposition. Characteristically, lead concentra-
tions are highest in confined areas close to sources and are progressively reduced by dilution
or deposition in districts more removed from sources. In parking garages or tunnels, atmo-
spheric lead concentrations can be ten to a thousand times greater than values measured near
roadways or in urban areas. In turn, atmospheric lead concentrations are usually about 2*j
SUMPB/D 1-22 9/30/83

-------
PRELIMINARY DRAFT
times greater in the central city than in residential suburbs. Rural areas have even lower
concentrations. Particle size distribution stabilizes within a few hundred kilometers of the
sources, although atmospheric concentration continues to decrease with distance. Ambient
organolead concentrations decrease more rapidly than inorganic lead, suggesting conversion
from the organic to the inorganic phase during transport. Inorganic lead appears to convert
from lead halides and oxides to lead sulfates.
Lead is removed from the atmosphere by wet or dry deposition. The mechanisms of dry de-
position have been incorporated into models that estimate the flux of atmospheric lead to the
earth's surface. Of particular interest is deposition on vegetation surfaces, since this lead
may be incorporated into food chains. Between wet and dry deposition, it is possible to cal-
culate an atmospheric lead budget that balances the emission inputs with deposition outputs.
Particles in air streams are subject to the same principles of fluid mechanics as par-
ticles in flowing water. The first principle is that of diffusion along a concentration gra-
dient. If the airflow is steady and free of turbulence, the rate of mixing is determined by
the diffusivity of the pollutant. By making generalizations of windspeed, stability, and sur-
face roughness, it is possible to construct models using a variable transport factor called
eddy diffusivity (K), in which K varies in each direction, including vertically. There is a
family of K-theory models that describe the dispersion of participate pollutants. The sim-
plest K-theory model produces a Gaussian plume, called such because the concentration of the
pollutant decreases according to a normal or Gaussian distribution in both the vertical and
horizontal directions. These models have some utility and are the basis for most of the air
quality simulations performed to date (Benarie, 1980). Another family of models is based on
the conservative volume element approach, where volumes of air are seen as discrete parcels
having conservative meteorological properties, (Benarie, 1980). The effect of pollutants on
these parcels is expressed as a mixing ratio. These parcels of air may be considered to move
along a trajectory that follows the advective wind direction. None of the models have been
tested for lead. All of the models require sampling periods of two hours or less in order for
the sample to conform to a well-defined set of meteorological conditions. In most cases, such
a sample would be below the detection limits for lead. The common pollutant used to test
models is SO, which can be measured over very short, nearly instantaneous, time periods. The
question of whether gaseous SO- can be used as a surrogate for particulate lead in these
models remains to be answered. v
Dispersion not influenced by complex terrain features depends on emission rates and the
volume of clean air available for mixing. These factors are relatively easy to estimate and
some effort has been made to describe ambient lead concentrations which can result under
selected conditions. On an urban scale, the routes of transport can be inferred from an iso-
pleth, i.e., a plot connecting points of identical ambient concentrations. These plots always
show that lead concentrations are maximum where traffic density is highest.
SUMPB/D 1-23 9/30/83

-------
PRELIMINARY DRAFT
Dispersion beyond cities to regional and remote locations is complicated by the fact that
there are no monitoring network data from which to construct isopleths, that removal by
deposition plays a more important role with time and distance, and that emissions from many
different geographic locations sources converge. Dispersion from point sources such as
smelters and refineries is described with isopleths in the manner of urban dispersion,
although the available data are notably less abundant.
Trijonis et al. (1980) reported lead concentrations for seven sites in St. Louis,
Missouri. Values around the CBD are typically two to three times greater than those found in
the outlying suburbs in St. Louis County to the west of the city. The general picture is one
of peak concentrations within congested commercial districts which gradually decline in out-
lying areas. However, concentration gradients are not steep, and the whole urban area has
levels of lead above 0.5 jjg/m?. Lead in the air decreases 2!s-fold from maximum values in
center city areas to well populated suburbs, with a further 2-fold decrease in the outlying
areas. These modeling estimates are generally confirmed by measurement.
The 15 mines and 7 primary smelters and refineries shown in Figure 1-6 are not located in
urban areas. Most of the 56 secondary smelters and refineries are likewise non-urban. Con-
sequently, dispersion from these point sources should be considered separately, but in a man-
ner similar to the treatment of urban regions. In addition to lead concentrations in air,
concentrations in soil and on vegetation surfaces are often used to determine the extent of
dispersion away from smelters and refineries.
Beyond the immediate vicinity of urban areas and smelter sites, lead in air declines
rapidly to concentrations of 0.1 to 0.5 nfl/m?- Two mechanisms responsible for this change are
dilution with clean air and removal by deposition.
Source reconciliation is based on the concept that each type of natural or anthropogenic
emission has a unique combination of elemental concentrations. Measurements of ambient air,,
properly weighted during multivariate regression analysis, should reflect the relative amount
of pollutant derived from each of several sources (Stolzenberg et al., 1982). Sievering et
al. (1980) used the method of Stolzenberg et al. (1982) to analyze the transport of urban air
from Chicago over Lake Michigan. They found that 95 percent of the lead in Lake Michigan air
could be attributed to various anthropogenic sources, namely coal fly ash, cement manufacture,
iron and steel manufacture, agricultural soil dust, construction soil dust, and incineration
emissions. Cass and McRae (1983) used source reco/iciliation in the Los Angeles Basin to
interpret 1976 NFAN data based on emission profiles from several sources. Their chemical ele-
ment balance model showed that 20 to 22 percent of the total suspended particle mass could be
attributed to highway sources.
Harrison and Williams (1982) determined air concentrations, particle size distributions,
and total deposition flux at one urban and two rural sites in England. The urban site, which
SUMPB/D 1-24 9/30/83

-------
PRELIMINARY DRAFT

had no apparent industrial, commercial or municipal emission sources, had an air lead concen-
tration of 3.8 ug/m?, whereas the two rural sites were about 0.15 ug/m?. The average particle
size became smaller toward the rural sites, as the MMED shifted downward from 0.5 urn to 0.1
Mm.
Knowledge of lead concentrations in the oceans and glaciers provides some insight into
the degrees of atmospheric mixing and long range transport. Patterson and co-workers have
measured dissolved lead concentrations in sea water off the coast of California, in the
Central North Atlantic (near Bermuda), and in the Mediterranean. The profile obtained by
Schaule and Patterson (1980) is shown in Figure 1-9. Surface concentrations in the Pacific
(14 ng/kg) were found to be higher than those of the Mediterranean or the Atlantic, decreasing
abruptly with depth to a relatively constant level of 1 to 2 ng/kg. The vertical gradient was
found to be much less in the Atlantic. Below the mixing layer, there appears to be no differ-
ence between lead concentrations in the Atlantic and Pacific. These investigators calculated
that industrial lead currently is being added to the oceans at about 10 times the rate of in-
troduction by natural weathering, with significant amounts being removed from the atmosphere
by wet and dry deposition directly into the ocean. Their data suggest considerable contamina-
tion of surface waters near shore, diminishing toward the open ocean.
Investigations of trace metal concentrations (including lead) in the atmosphere in remote
northern and southern hemispheric sites have revealed that the natural sources for such atmos-
pheric trace metals include the oceans and the weathering of the earth's crust, while the
major anthropogenic source is particulate air pollution. Enrichment factors for concentra-
tions relative to standard values for the oceans and the crust were calculated; ninety percent
of the particulate pollutants in the global troposphere are injected in the northern hemi-
sphere (Robinson and Robbins, 1971). Since the residence times for particles in the tropo-
sphere are much less than the interhemispheric mixing time, it is unlikely that significant
amounts of particulate pollutants can migrate from the northern to the southern hemisphere via
the troposphere.
Murozumi et al. (1969) have shown that long range transport of lead particles emitted
from automobiles has significantly polluted the polar glaciers. They collected samples of
snow and ice from Greenland and the Antarctic (Figure 1-10). The authors attribute the gra-
dient increase after 1750 to the Industrial Revolution and the accelerated increase after 1940
to the increased use of lead alky Is in gasoline. The most recent levels found in the
Antarctic snows were, however, less than those found in Greenland by a factor of 10 or more.
Evidence from remote areas of the world suggests that lead and other fine particle com-
ponents are transported substantial distances, up to thousands of kilometers, by general
weather systems. The degree of surface contamination of remote areas with lead depends both
on weather influences and on the degree of air contamination. However, even in remote areas,
man's primitive activities can play an important role in atmospheric lead levels.
SUMPB/0 1-25 9/30/83

-------
                                   PRELIMINARY DRAFT
                        1000  -
                     g  2000
                     V
                     I
                        3000
                        4000
                        5000
                            I   I   I    I   I   I
                                              • DISSOLVED Pb

                                              D PARTICULATE Pb
                                    4    6   8  10  12   14  16   0

                                    CONCENTRATION, ng Pb/kg
                      Figure 1-9.  Profile of lead concentrations in the
                     central northeast Pacific. Values below 1000 m are
                     an order of magnitude tower than reported by
                     Tatsumoto and Patterson (1963) and Chow and
                     Patterson (1966).

                     Source: Schaule and Patterson (1980).
SUMPB/D
1-26
9/30/83

-------
PRELIMINARY DRAFT
AGE OF SAMPLES

Figure 1-10. Lead concentration profile in snow
strata of Northern Greenland.
Source: Murozumi et al. (1969).

Whitby et al. (1975) placed atmospheric particles into three different size regimes: the
nuclei mode (<0.1 urn), the accumulation mode (0.1 to 2 urn), and the large particle mode (>2
Urn). At the source, lead particles are generally in the nuclei and large particle modes.
Large particles are removed by deposition close to the source and particles in the nuclei mode
diffuse to surfaces or agglomerate while airborne to form larger particles of the accumulation
mode. Thus it is in the accumulation mode that particles are dispersed great distances.
A number of studies have used gas absorbers behind filters to trap vapor-phase lead com-
pounds. Because it is not clear that all the lead captured in the backup traps is, in fact,
in the vapor phase in the atmosphere, "organic" or "vapor phase" lead is an operational
definition in these studies. Purdue et al. (1973) measured both particulate and organic lead
in atmospheric samples. They found that the vapor phase lead was about 5 percent of the total
lead in most samples. It is noteworthy, however, that in an underground garage, total lead
concentrations were approximately five times those in ambient urban atmospheres, and the
organic lead increased to approximately 17 percent.
Lead is emitted into the air from automobiles as lead halides and as double salts with
ammonium halides (e.g., PbBrCl • 2NH4C1). From mines and smelters, PbS04> PbO-PbS04, and PbS
appear to be the dominant species. In the atmosphere, lead is present mainly as the sulfate
SUMPB/D
1-27
9/30/83

-------
PRELIMINARY DRAFT
with minor amounts of halides. It is not completely clear just how the chemical composition
changes in transport.
The ratio of Br to Pb is often cited as an indication of automotive emissions. From the
mixtures commonly used in gasoline additives, the mass Br/Pb ratio should be about 0.386 if
there has been no fractionation of either element (Harrison and Sturges, 1983). However,
several authors have reported loss of halide, preferentially bromine, from lead salts in atmo-
spheric transport. Both photochemical decomposition and acidic gas displacement have been
postulated as mechanisms. The Br/Pb ratios maybe only crude estimates of automobile emissions;
this ratio would decrease with distance from the highway from 0.39 to 0.35 at less proximate
sites and 0.25 in suburban residential areas. Habibi et al. (1970) studied the composition of
auto exhaust particles as a function of particle size. Their main conclusions follow:

1. Chemical composition of emitted exhaust particles is related to particle size.
2. There is considerably more soot and carbonaceous material associated with fine-
mode particles than with coarse-mode particles. Particulate matter emitted
under typical driving conditions is rich in carbonaceous material.
3. Only small quantities of 2PbBrCl-NH4Cl were found in samples collected at the
tailpipe from the hot exhaust gas. Lead-halogen molar ratios in particles of
less than 10 urn MMED indicate that much more halogen is associated with these
solids than the amount expected from the presence of 2PbBrCl*NH4Cl.

Lead sulfide is the main constituent of samples associated with ore handling and fugitive
dust from open mounds of ore concentrate. The major constituents from sintering and blast
furnace operations appeared to be PbSO. and PbO-PbSO., respectively.
Before atmospheric lead can have any effect on organisms or ecosystems, it must be trans-
ferred from the air to a surface. For natural ground surfaces and vegetation, this process
may be either dry or wet deposition. Transfer by dry deposition requires that the particle
move from the main airstream through the boundary layer to a surface. The boundary layer is
defined as the region of minimal air flow immediately adjacent to that surface. The thickness
of the boundary layer depends mostly on the windspeed and roughness of the surface. Airborne
particles do not follow a smooth, straight path in the airstream. On the contrary, the path
of a particle may be affected by micro-turbulent air currents, gravitation, or its own iner-
tia. There are several mechanisms which alter the particle path sufficient to cause transfer
to a surface. These mechanisms are a function of particle size, windspeed, and surface char-
acteristics. Transfer from the main airstream to the boundary layer is usually by sedimenta-
tion or wind eddy diffusion. From the boundary layer to the surface, transfer may be by any
of the six mechanisms, although those which are independent of windspeed (sedimentation, in-
terception, Brownian diffusion) are more likely.
SUMPB/D 1-28 9/30/83

-------
PRELIMINARY DRAFT
Particles transported to a surface by any mechanism are said to have an effective de-
position velocity (V.) which is measured not by rate of particle movement but by accumulation
on a surface as a function of air concentration. Several recent models of dry deposition have
evolved from the theoretical discussion of Fuchs (1964) and the wind tunnel experiments of
Chamberlain (1966). The models of SI inn (1982) and Davidson et al. (1982) are particularly
useful for lead deposition. SI inn's model considers a multitude of vegetation parameters to
find several approximate solutions for particles in the size range of 0.1 to 1.0 urn, estima-
ting deposition velocities of 0.01 to 0.1 cm/sec. The model of Davidson et al. (1982) is
based on detailed vegetation measurements and wind data to predict a V. of 0.05 to 1.0 cm/sec.
Deposition velocities are specific for each vegetation type. Both models show a decrease in
deposition velocity as particle size decrease down to about 0.1 to 0.2 urn; as diameter
decreases further from 0.1 to 0.001 urn, deposition velocity increases (see Figure 6-1).
Several investigators have used surrogate surface devices to measure dry deposition
rates. The few studies available on deposition to vegetation surfaces show deposition rates
comparable to those of surrogate surfaces and deposition velocities in the range predicted by
the models discussed above (Table 1-2). These data show that global emissions are in approxi-
mate balance with global deposition.
Andren et al. (1975) evaluated the contribution of wet and dry deposition of lead in a
study of the Walker Branch Watershed in Oak Ridge, Tennessee, during the period June, 1973 -
July, 1974. The mean precipitation in the area is approximately 130 cm/yr. Wet deposition
contributed approximately 67 percent of the total deposition for the period.
The geochemical mass balance of lead in the atmosphere may be determined from quantita-
tive estimates of inputs and outputs. Inputs amount to 450,000 - 475,000 metric tons an-
nually (Table 1-1). . The amount of lead removed by wet deposition is approximately 208,000
t/yr (Table 1-3).
The deposition flux for each vegetation type shown on Table 1-3 totals 202,000. The
combined wet and dry deposition is 410,000 metric tons, which compares favorably with the es-
timated 450,000 - 475,000 metric tons of emissions.
Soils have both a liquid and solid phase, and trace metals are normally distributed be-
tween these two phases. In the liquid phase, metals may exist as free ions or as soluble com-
plexes with organic or inorganic ligands. Organic ligands are typically humic substances such
as fulvic or humic acid, and the inorganic ligands may be iron or manganese hydrous oxides.
Since lead rarely occurs as a free ion in the liquid phase (Camerlynck and Kiekens, 1982), its
mobility in the soil solution depends on the availability of organic or inorganic ligands.
The liquid phase of soil often exists as a thin film of moisture in intimate contact with the
solid phase. The availability of metals to plants depends on the equilibrium between the
liquid and solid phase. In the solid phase, metals may be incorporated into crystalline
SUMPB/D 1-29 9/30/83

-------
PRELIMINARY DRAFT
TABLE 1-2. SUMMARY OF SURROGATE AND VEGETATION SURFACE DEPOSITION OF LEAD
Depositional Surface
Flux
ng Pb/cmf/day
Air Cone
ng/m?
Deposition Velocity
cm/sec Reference
Tree leaves (Paris)
0.38
Tree leaves (Tennessee) 0.29-1.2
0.02-0.08
Plastic disk (remote
California)
Plastic plates
(Tennessee)
0.29-1.5
13-31

110
0.086
0.05-0.4
0.05-0.06
1
2
3
Tree leaves (Tennessee) — 110 0.005
Snow (Greenland) 0.004 0.1-0.2 0.1
Grass (Pennsylvania) — 590 0.2-1.1
Coniferous forest (Sweden) 0.74 21 - 0.41
4
5
6
7
1. Servant, 1975
2. Lindberg et al., 1982
3. Elias and Davidson, 1980
4. Lindberg and Harriss, 1981
5. Davidson et al., 1981 c
6. Davidson et al., 1982
7. Lannefors et al., 1983
minerals of parent rock material and secondary clay minerals or precipitated as insoluble
organic or inorganic complexes. They may also be adsorbed onto the surfaces of any of these
solid forms. Of these categories, the most mobile form is in soil moisture, where lead can
move freely into plant roots or soil microorganisms with dissolved nutrients. The least
mobile is parent rock material, where lead may be bound within crystalline structures over
geologic periods of time; intermediate are the lead complexes and precipitates. Trans-
formation from one form to another depends on the chemical environment of the soil. The water
soluble and exchangeable forms of metals are generally considered available for plant uptake
(Camerlynck and Kiekens, 1982). These authors demonstrated that in normal soils, only a small
fraction of the total lead is in exchangeable form (about 1 ug/g) and none exists as free lead
ions. Of the exchangeable lead, 30 percent existed as stable complexes, 70 percent as labile
complexes.
SUMPB/D
1-30
9/30/P3

-------
PRELIMINARY DRAFT
TABLE 1-3. ESTIMATED GLOBAL DEPOSITION OF ATMOSPHERIC LEAD
Deposition from Atmosphere
Mass Concentration Deposition
1017 kg/yr 10-* g/kg 10? kg/yr
Wet
To oceans
To continents
Dry
4.1
1.1

Area
101? km?
To oceans, ice caps, deserts 405
Grassland, agricultural
0.4
0.4

Deposition rate
10-? g/m?/yr
0.2
164
44
Deposition
10? kg/yr
89
areas, and tundra 46 0.71
Forests 59 1.5
Total dry:
Total wet:
Global:
33
80
202
208
410
Source: This report.
Atmospheric lead may enter the soil system by wet or dry deposition mechanisms. Lead
could be immobilized by precipitation as less soluble compounds [PbCO,, Pb(PO^)2], by ion ex-
change with hydrous oxides or clays, or by chelation with humic and fulvic acids. Lead im-
mobilization is more strongly correlated with organic chelation than with iron and managanese
oxide formation (Zimdahl and Skogerboe, 1977). If organic chelation is the correct model of
lead immobilization in soil, then several features of this model merit further discussion.
First, the total capacity of soil to immobilize lead can be predicted from the linear rela-
tionship developed by Zimdahl and Skogerboe (1977) (Figure 1-11) based on the equation:
N = 2.8 x 10"6 (A)
1.1 x 10"5 (B) - 4.9 x 10"5
where N is the saturation capacity of the soil expressed in moles/g soil, A is the cation ex-
change capacity of the soil in meq/100 g soil, and B is the pH.
SUMPB/D
1-31
9/30/83

-------
                                       PRELIMINARY DRAFT
     The  soil  humus  model  also facilitates  the calculation of  lead in  soil  moisture using
values  available  in the literature for conditional  stability constants  (K) with fulvic acid.
The  values  reported for log K  are  linear in the pH range of 3 to 6 so that interpolations in
the  critical  range  of pH 4 to  5.5  are possible (Figure 1-11).  Thus, at pH 4.5, the ratio of
complexed lead  to ionic lead is expected to be 3.8 x 103.   For  soils of 100 ug/g, the ionic
lead in soil moisture solution would be 0.03 ug/g.
               5.0
                                   pH =8
                            	pH = 6
                            	pH = 4
                                          50           76
                                           CEC, meq/100 g
100
125
                Figure 1-11. Variation of lead saturation capacity with cation exchange
               capacity in soil at selected pH values.
               Source:  Data from Zimdahl and Skogerboe (1977).
     It is also  important to consider the stability constant of the Pb-FA complex relative to
other metals.    Schnitzer and  Hansen (1970)  showed that  at  pH  3,  Fe3+  is the  most  stable in
the  sequence   Fc?+  >  Alf* > Cu2"*"  > Ni 2* >  Co** > Pbz+ > Ca2+ > Zn2+ > Mn2+ > Mg2+.    At  pH
5,  this  sequence   becomes  Ni2+  =  Coa+ > Pb2* > Cu2* > Zn2+  =  Hn2+ > Ca2+ > Mg2+.   This
means that at  normal  soil  pH levels of  4.5  to 8, lead is bound to FA + HA  in preference to
many other metals that are known plant nutrients (Zn,  Mn,  Ca, and Mg).
SUMPB/0                                     1-32                                       9/30/83

-------
PRELIMINARY DRAFT
Lead does not pass easily to ground or surface water. Any lead dissolved from primary
lead sulfide ore tends to combine with carbonate or sulfate ions to form insoluble lead car-
bonate or lead sulfate, or be absorbed by ferric hydroxide. An outstanding characteristic of
lead is its tendency to form compounds of low solubility with the major anions of natural
water. The hydroxide, carbonate, sulfide, and more rarely the sulfate may act as solubility
controls in precipitating lead from water. The amount of lead that can remain in solution is
a function of the pH of the water and the dissolved salt content. A significant fraction of
the lead carried by river water may be in an undissolved state. This insoluble lead can con-
sist of colloidal particles in suspension or larger undissolved particles of lead carbonate,
-oxide, -hydroxide, or other lead compounds incorporated in other components of particulate
lead from runoff; it may occur either as sorbed ions or surface coatings on sediment mineral
particles or be carried as a part of suspended living or nonliving organic matter.
The bulk of organic compounds in surface waters originates from natural __ sources.
(Neubecker and Allen, 1983). The humic and fulvic acids that are primary complexing agents in
soils are also found in surface waters at concentrations from 1 to 5 mg/1, occasionally ex-
ceeding 10 mg/1. The presence of fulvic acid in water has been shown to increase the rate of
solution of lead sulfide 10 to 60 times over that of a water solution at the same pH that did
not contain fulvic acid. At pH values near 7, soluble lead-fulvic acid complexes are present
in solution.
The transformation of inorganic lead, especially in sediment, to tetramethyllead (TML)
has been observed and biomethylation has been postulated. However, Reisinger et al. (1981)
have reported extensive studies of the methylation of lead in the presence of numerous
bacterial species known to alkylate mercury and other heavy metals. In these experiments no
biological methylation of lead was found under any condition.
Lead occurs in riverine and estuarial waters and alluvial deposits. Concentrations of
lead in ground water appear to decrease logarithmically with distance from a roadway. Rain-
water runoff has been found to be an important transport mechanism in the removal of lead from
a roadway surface in a number of studies. Apparently, only a light rainfall, 2 to 3 mm, is
sufficient to remove 90 percent of the lead from the road surface to surrounding soil and to
waterways. The lead concentrations in off-shore sediments often show a marked increase
corresponding to anthropogenic activity in the region. An average anthropogenic flux of 72
mg/m?-yr, of which 27 mg/m?-yr could be attributed to direct atmospheric deposition. Prior to
1650, the total flux was 12 mg/m?«yr, so there has been a 6-fold increase since that time. Ng
and Patterson (1982) found prehistoric fluxes of 1 to 7 mg Pb/m?-yr to three offshore basins
in southern California, which have now increased 3 to 9-fold to 11 to 21 mg/m2-yr. Much of
this lead is deposited directly from sewage outfalls, although at least 25 percent probably
comes from the atmosphere.
SUMPB/D 1-33 9/30/83

-------
PRELIMINARY DRAFT
The deposition of lead on the leaf surfaces of plants where the particles are often re-
tained for a long time can be important. Several studies have shown that plants near roadways
exhibit considerably higher levels of lead than those farther away. Rainfall does not gene-
rally remove the deposited particles. Animals or humans consuming the leafy portions o'f such
plants can be exposed to higher than normal levels of lead. The particle deposition on leaves
has led some investigators to stipulate that lead may enter plants through the leaves. Arvik
and Zimdahl (1974) have shown that entry of ionic lead through plant leaves is of minimal
importance. Using 'the leaf cuticles of several types of plants essentially as dialysing mem-
branes, they found that even high concentrations of lead ions would not pass through the cuti-
cles into distilled water on the opposite side.
1.7 ENVIRONMENTAL CONCENTRATIONS AND POTENTIAL PATHWAYS TO HUMAN EXPOSURE
In general, typical levels of human lead exposure may be attributed to four components of
the human environment: inhaled air, dusts of various types, food and drinking water. A base-
line level of potential human exposure is determined for a normal adult eating a typical diet
and living in a non-urban community. This baseline exposure is deemed to be unavoidable by
any reasonable means. Beyond this level, additive exposure factors can be determined for
other environments (urban, occupational, smelter communities), for certain habits and activi-
ties (smoking, drinking, pica, and hobbies), and for variations due to age, sex, or socio-
economic status.

1.7.1 Lead in Air
Ambient airborne lead concentrations may influence human exposure through direct inhala-
tion of lead-containing particles and through ingestion of lead which has been deposited from
the air onto surfaces. Our understanding of the pathways to human exposure is far from com-
plete because most ambient measurements were not taken in conduction with studies of the con-
centrations of lead in man or in components of his food chain.
The most complete set of data on ambient air concentrations may be extracted from the
National Filter-Analysis Network (NFAN) and its predecessors. In remote regions of the world,
air concentrations are two or three orders of magnitude lower than in urban areas, lending
credence to estimates of the concentrations of natural lead in the atmosphere. In the context
of this data base, the conditions which modify ambient air, as measured by the monitoring net-
works, to air inhaled by humans cause changes in particle size distributions, changes with
vertical distance above ground-, and differences between indoor and outdoor concentrations.
The wide range of concentration is apparent from Table 1-4, which summarizes data ob-
tained from numerous independent measurements. Concentrations vary from 0.000076 ug/m3 in
SUMPB/D 1-34 9/30/83

-------
                                       PRELIMINARY DRAFT
          TABLE 1-4.  ATMOSPHERIC LEAD IN URBAN, RURAL, AND REMOTE AREAS OF THE WORLD
Location Sampling Period
Urban
Miami
New York
Boston
St. Louis
Houston
Chicago
Salt Lake City
Los Angeles
Ottowa
Toronto
Montreal
Berlin
Vienna
Zurich
Brussels
Turin
Rome
Paris
Rio de Janeiro
Rural
New York Bight
Pramingham, MA
Chadron, NE
United Kingdom
Italy
Belgium
Remote
White Mtn. , CA
High Sierra, CA
Olympic Nat. Park, WA
Antarctica
South Pole
Thule, Greenland
Thule, Greenland
Prins Christian-
sund, Greenland
Dye 3, Greenland
Eniwetok, Pacific Ocean
Kumjung, Nepal
Bermuda
Spitsbergen

1974
1978-79
1978-79
1973
1978-79
1979
1974
1978-79
1975
1975
1975
1966-67
1970
1970
1978
1974-79
1972-73
1964
1972-73

1974
1972
1973-74
1972
1976-80
1978

1969-70
1976-77
1980
1971
1974
1965
1978-79

1978-79
1979
1979
1979
1973-75
1973-74
Lead cone, (ug/m3) Reference

1.3
1.1
0.8
1.1
0.9
0.8
0.89
1.4
1.3
1.3
2.0
3.8
2.9
3.8
0.5
4.5
4.5
4.6
0.8

0.13
0.9
0.045
0.13
0.33
0.37

0.008
0.021
0.0022
0.0004
0.000076
0.0005
0.008

0.018
0.00015
0.00017
0.00086
0.0041
0.0058

HASL, 1975
see Table 7-3
see Table 7-3
see Table 7-3
see Table 7-3
see Table 7-3
HASL, 1975
see Table 7-3
NAPS, 1975
NAPS, 1975
NAPS, 1975
Blokker, 1972
Hartl and Resch, 1973
HBgger, 1973
Roel s et al . , 1980
Facchetti and Geiss, 1982
Colacino and Lavagnini, 1974
Blokker, 1972
Branquinho and Robinson, 1976

Duce et al . , 1975
O'Brien et al., 1975
Struempler, 1975
Cawse, 1974
Facchetti and Geiss, 1982
Roels et al. 1980

Chow et al . , 1972
Elias and Davidson, 1980
Davidson et al. , 1982
Duce, 1972
Maenhaut et al . , 1979
Murozumi et al., 1969
He i dam, 1981

Keidam, 1981
Davidson et al., 1981c
Settle and Patterson, 1982
Davidson et al., 1981b
Duce et al . , 1976
Larssen, 1977
 'All  references listed as  cited  in Nriagu  (1978b).
SUMPB/D
1-35
9/30/83

-------
PRELIMINARY DRAFT
remote areas to over 10 pg/m3 near sources such as smelters. Many of the remote areas are far
from human habitation and therefore do not reflect human exposure. However, a few of the
regions characterized by small lead concentrations are populated by individuals with primitive
lifestyles; these data provide baseline airborne lead data to which modern American lead expo-
sures can be compared.
The remote area concentrations reported in Table 1-4 do not necessarily reflect natural,
preindustrial lead. Murozumi et al. (1969) and Ng and Patterson (1981) have measured a 200-
fold increase in the lead content of Greenland snow over the past 3000 years. The authors
state that this lead originates in populated mid-latitude regions, and is transported over
thousands of kilometers through the atmosphere to the Arctic. All of the concentrations in
Table 1-4, including values for remote areas, have been influenced by anthropogenic lead emis-
sions.
The data from the Air Filter networks show both the maximum quarterly average to reflect
compliance of the station to the ambient airborne standard (1.5 M9/m3), and quarterly aver-
ages to show trends at a particular location. The number of stations complying with the stan-
dard has increased, the quarterly averages have decreased, and the maximum 24-hour values ap-
pear to be smaller since 1977.
It seems likely that the concentration of natural lead in the atmosphere is between
0.00002 and 0.00007 |jg/m3. A value of 0.00005 will be used for calculations regarding the
contribution of natural air lead to total human uptake.
The effect of the 1978 National Ambient Air Quality Standard for Lead has been to reduce
the air concentration of lead in major urban areas. Similar trends may also be seen in urban
areas of smaller population density. There are many factors which can cause differences
between the concentration of lead measured at a monitoring station and the actual inhalation
of air by humans. Air lead concentrations usually decrease with vertical and horizontal dis-
tance from emission sources, and are generally lower indoors than outdoors.
New guidelines for placing ambient air lead monitors went into effect in July, 1981
(F.R., 1981 September 3). "Microscale" sites, placed between 5 and 15 meters from thorough-
fares and 2 to 7 meters above the ground, are prescribed, but until now few monitors have been
located that close to heavily travelled roadways. Many of these microscale sites might be ex-
pected to show higher lead concentrations than measured at nearby middlescale urban sites, due
complex. Our understanding of the complex factors affecting the vertical distribution of air-
borne lead is extremely limited, but the data indicate that air lead concentrations are pri-
marily a function of distance from the source, whether vertical or horizontal.
Because people spend much of their time indoors, ambient air data may not accurately
indicate actual exposure to airborne lead. Some studies show smaller indoor/outdoor ratios

SUMPB/D 1-36 9/30/83

-------
PRELIMINARY DRAFT
during the winter, when windows and doors are tightly closed. Overall, the data suggest
indoor/outdoor ratios of 0.6-0.8 are typical for airborne lead in houses without air con-
ditioning. Ratios in air conditioned houses are expected to be in the range of 0.3-0.5
(Yocum, 1982). Even detailed knowledge of indoor and outdoor airborne lead concentrations at
fixed locations may still be insufficient to assess human exposure to airborne lead. The
study of Tosteson et al. (1982) included measurement of airborne lead concentrations using
personal exposure monitors, carried by individuals going about their day-to-day activities.
In contrast to the lead concentrations of 0.092 and 0.12 ug/m3 at fixed locations, the average
personal exposure was 0.16 ug/m3. The authors suggest the inadequacy of using fixed monitors
at either indoor or outdoor locations to assess exposure.
Much of the lead in the atmosphere is transferred to terrestrial surfaces where it is
eventually passed to the upper layer of the soil surface. Crustal lead concentrations in soil
range from less than 10 to greater than 70 pfl/9- The range of values probably represent
natural levels of lead in soil, although there may have been some contamination with anthro-
pogenic lead during collection and handling.

1.7.2 Lead in Soil and Dust
Studies have determined that atmospheric lead is retained in the upper two centimeters of
undisturbed soil, especially soils with at least 5 percent organic matter and a pH of 5 or
above. There has been no general survey of this upper 2 cm of the soil surface in the United
States, but several studies of lead in soil near roadsides and smelters and a few studies of
lead in soil near old houses with lead-based paint can provide the backgound information for
determining potential human exposures to lead from soil. Because lead is immobilized by the
organic component of soil, the concentration of anthropogenic lead in the upper 2 cm is deter*
mined by the flux of atmospheric lead to the soil surface. Near roadsides, this flux is
largely by dry deposition and the rate depends on particle size and concentration. In gen-
eral, deposition flux drops off abruptly with increasing distance from the roadway. This
effect is demonstrated in studies which show surface soil lead decreases exponentially up to
25 m from the edge of the road. Roadside soils may contain atmospheric lead from 30 to 2000
mg/g in excess of natural levels within 25 meters of the roadbed, all in the upper layer of
the soil profile.
Near primary and secondary smelters, lead in soil decreases exponentially within a 5-10
km zone around the smelter complex. Soil lead contamination varies with the smelter emission
rate, length of time the smelter has been in operation, prevailing windspeed and direction,
regional climatic conditions, and local topography.
Urban soils may be contaminated from a variety of atmospheric and non-atmospheric-
sources. The major sources of soil lead seem to be paint chips from older houses and deposi-
tion from nearby highways. Lead in soil adjacent to a house decreases with distance; this may
SUMPB/D 1-37 9/30/83

-------
PRELIMINARY DRAFT
be due to paint chips or to dust of atmospheric origin washing from the rooftop (Wheeler and
Rolfe, 1979).
A definitive study which describes the source of soil lead was reported by Gulson et al.
(1981) for soils in the vicinity of Adelaide, South Australia, In an urban to rural transect,
stable lead isotopes were measured in the top 10 cm of soils over a 50 km distance. By their
isotopic compositions, three sources of lead were identified: natural, non-automotive in-
dustrial lead from Australia, and tetraethyl lead manufactured in the United States. The re-
sults indicated most of the soil surface lead originated from leaded gasoline. Lead may be
found in inorganic primary minerals, on humic substances, complexed with Fe-Mn oxide films, on
secondary minerals or in soil moisture. All of the lead in primary minerals is natural and is
bound tightly within the crystalline structure of the minerals. The lead on the surface of
these minerals is leached slowly into the soil moisture. Atmospheric lead forms complexes
with humic substances or on oxide films, that are in equilibrium with soil moisture, although
the equilibrium strongly favors the cpmplexing agents. Except near roadsides and smelters,
only a few ug of atmospheric lead have been added to each gram of soil. Several studies in-
dicate that this lead is available to plants and that even with small amounts of atmospheric
lead, about 75 percent of the lead in soil moisture is of atmospheric origin.
Lead on the surfaces of vegetation may be of atmospheric origin. In internal tissues,
lead maybe a combination of atmospheric and soil origin. As with soils, lead on vegetation
surfaces decreases exponentially with distance away from roadsides and smelters.-This de-
posited lead is persistent. It is neither washed off by rain nor taken up through the leaf
surface. Lead on the surface of leaves and bark is proportional to air lead concentrations
and particle size distributions. Lead in internal plant tissues is directly related to lead
in soil.

1.7.3 Lead in Food
In a study to determine the background concentrations of lead and other metals in agri-
cultural crops, the Food and Drug Administration (Wolnik et al., 1983), in cooperation with
the U.S. Department of Agriculture and the U.S. Environmental Protection Agency, analyzed over
1500 samples of 'the most common crops taken from a cross section of geographic locations.
.Collection sites were remote from mobile or stationary sources of lead. Soil lead concentra-
tions were within the normal range (8-25 ug/g) of U.S. soils. The concentrations of lead in
crops are shown as "Total" concentrations on Table 1-5. The total concentration data should
probably be seen as representing the lowest concentrations of lead in food available to
Americans. The data on these te'n crops suggest that root vegetables have lead concentrations
between 0.0046 and 0.009 ug/g, all soil lead. Aboveground parts not exposed to significant
amounts of atmospheric deposition (sweet corn and tomatoes) have less lead internally. If it
SUMPB/D 1-38 9/30/83

-------
PRELIMINARY DRAFT
is assumed that this same concentration is the internal concentration for aboveground parts
for other plants, it is apparent that five crops have direct atmospheric deposition in pro-
portion to surface area and estimated duration of exposure. The deposition rate of 0.04
ng/cm2-day in rural environments could account for these amounts of direct atmospheric lead.

TABLE 1-5. BACKGROUND LEAD IN BASIC FOOD CROPS AND MEATS
(ug/g fresh weight)
Crop
Wheat
Potatoes
Field corn
Sweet corn
Soybeans
Peanuts
Onions
Rice
Carrots
Tomatoes
Spinach
Lettuce
Beef (muscle)
Pork (muscle)
Natural
Pb
0.0015
0.0045
0. 0015
0.0015
0.021'
0.050
0.0023
0.0015
0.0045
0.001
0.0015
0.0015
0. 0002
0.0002
Indirect
Atmospheric
0. 0015
0. 0045
0.0015
0.0015
0.021
0.050
0.0023
0.0015
0.0045
0.001
0.0015
0.0015
0.002
0.002
Direct
Atmospheric
0.034
—
0.019
—
—

—
0.004
—
—
0.042
0.010
0.02
0.06
Total1"
0.037
0.009
0.022*
0.003
0.042
0.100
0.0046*
0.007*
0.009*
0.002*
0.045*
0.013
0.02**
0.06**
'except as indicated, data are from Wolnick et al. (1983)
*pre1iminary data provided by the Elemental Analysis Research Center, Food and Drug
Administration, Cincinnati, OH
**data from Penumarthy et al. (1980)

Lead in food crops varies according to exposure to the atmosphere and in proportion to the ef-
fort taken to separate husks, chaff, and hulls from edible parts during processing for human
or animal consumption. Root parts and protected aboveground parts contain natural lead and
indirect atmospheric lead, both derived from the soil. For exposed aboveground parts, any
lead in excess of the average of unexposed aboveground parts is considered to have been
directly deposited from the atmosphere.

1.7.4 Lead in Water
Lead occurs in untreated water in either dissolved or particulate form. Dissolved lead is
operationally defined as that which passes through a 0.45 urn membrane filter. Because atmos-
pheric lead in rain or snow is retained by soil, there is little correlation between lead in

SUMPB/D 1-39 9/30/83

-------
PRELIMINARY DRAFT
precipitation and lead In streams that drain terrestrial watersheds. Rather, the important
factors seem to be the chemistry of the stream (pH and hardness) and the volume of the stream
flow. For groundwater, chemistry is also important, as is the geochemical composition of the
water-bearing bedrock.
Streams and lakes are influenced by their water chemistry and the lead content of their
sediments. At neutral pH, lead moves from the dissolved to particulate form and the particles
eventually pass to sediments. At low pH, the reverse pathway is generally the case. Hard-
ness, which is a combination of the Ca and Mg concentration, can also influence lead concen-
trations. At higher concentrations of Ca and Mg, the solubility of lead decreases. Municipal
and private wells typically have a neutral pH and somewhat higher than average hardness. Lead
concentrations are not influenced by acid rain, surface runoff or atmospheric deposition.
Rather, the primary determinant of lead concentration is the geochemical makeup of the bedrock
that is the source of the water supply. Ground water typically ranges from 1 to 100 ug Pb/1
(National Academy of Sciences, 1980).
Whether from surface or ground water supplies, municipal waters undergo extensive chem-
ical treatment prior to release to the distribution system. Although there is. no direct ef-
fort to remove lead from the water supply, some treatments, such as flocculation and sedimen-
tation, may inadvertently remove lead along with other undesirable substances. On the other
hand, chemical treatment to soften water increases the solubility of lead and enhances the
possibility that lead1 will be added to* water as it passes through the distribution system.
For samples taken at the household tap, lead concentrations are usually higher in the initial
volume (first daily flush) than after the tap has been running for some time. Water standing
in the pipes for several hours is intermediate between these two concentrations. (Sharrett et
al., 1982; Worth et al., 1981).

1.7.5 Baseline Exposures to Lead
Lead concentrations in environmental media that are in the pathway to human consumption
are summarized on Table 1-6. Because natural lead is generally three to four orders of magni-
tude lower than anthropogenic lead in ambient rural or urban air, all atmospheric contri-
butions of lead are considered to be of anthropogenic origin. Natural soil lead typically
ranges from 10 to 30 M9/9> but much of this is tightly bound within the crystalline matrix of
soil minerals at normal soil pHs of 4 to 8. Lead in the organic fraction of soil is part
natural and part atmospheric. The fraction derived from fertilizer is considered to be
minimal. In undisturbed rural and remote soils, the ratio of natural to atmospheric lead is
about 1:1, perhaps as high as 1:3. This ratio persists through soil moisture and into in-
ternal plant tissues.

SUMPB/D 1-40 9/30/83

-------
                                       PRELIMINARY DRAFT
                 TABLE 1-6.  SUMMARY OF ENVIRONMENTAL CONCENTRATIONS OF LEAD
Medium
Air urban (ug/m3)
rural (ug/m3)
Soil Total (ug/g)
Food Crops (ug/g)
Surface water (ug/g)
Ground water (ug/g)
Natural
Lead
0.00005
0.00005
8-25
0.0025
0.00002
0.003
Atmospheric
Lead
0.8
0.2
3.0
0.027
0.005
—
Total
Lead
0.8
0.2
15.0
0.03
0.005
0.003
     In tracking air  lead  through pathways to  human  exposure,  it is necessary to distinguish
between atmospheric lead that has passed through the  soil,  called indirect atmospheric here,
and atmospheric lead  that  has deposited directly on  crops  or water.   Because indirect atmos-
pheric lead will remain  in the soil for many decades, this source is insensitive to projected
changes in atmospheric lead concentrations.
     Initially, a  current  baseline  exposure  scenario  is  described for an  individual  with a
minimum amount of  daily  lead consumption.   This person would live and work in a nonurban en-
vironment, eat  a  normal diet of food taken from a  typical  grocery shelf,  and  would  have no
habits or activities that would tend to increase lead exposure.   Lead exposure at the baseline
level  is  considered unavoidable  without  further reductions  of lead in the atmosphere or in
canned foods.   Most of the  baseline lead is of anthropogenic origin.
     To arrive at a minimum or baseline exposure for humans, it is necessary to begin with the
environmental  components, air,  soil,  food  crops and water,  that are the major sources of lead
consumed  by  humans  (Table 1-6).   These  components are  measured frequently, even  monitored
routinely in the case  of air, so that much  data are available on  their concentrations.   But
there are several  factors which modify these components prior to actual  human exposure:   We do
not  breathe  air as monitored  at  an  atmospheric  sampling  station;  we may be  closer  to or
farther from  the  source of  lead than  is  the  monitor;  we  may  be inside a  building, with or
without filtered air;  water  we drink does  not come directly from a stream  or river,  but often
has passed through  a  chemical treatment plant  and  a distribution system.    A  similar type of
processing has modified the lead levels present in  our food.
     Besides  the atmospheric  lead in environmental  components,  there are two other industrial
components which  contribute  to  this  baseline  of  human exposure:   paint  pigments and lead

SUMPB/D                                     1-41                                       9/30/83

-------
PRELIMINARY DRAFT
solder. Solder contributes directly to the human diet through canned food and copper water
distribution systems. Paint and solder are also a source of lead-bearing dusts. The most
common dusts in the baseline human environment are street dusts and household dusts. They
originate as emissions from mobile or stationary sources, as the oxidation products of surface
exposure, or as products of frictional grinding processes. Dusts are different from soil in
that soil derives from crustal rock and typically has a lead concentration of 10 to 30 ug/g,
whereas dusts come from both natural and anthropogenic sources and vary from 1000 to 10,000
M9/g.
The route by which many people receive the largest portion of their daily lead intake is
via foods. Several studies have reported average dietary lead intakes in the range 100 to 500
(jg/day for adults, with individual diets covering a much greater range (Nutrition Foundation,
1982). The sources of lead in plants and animals are air, soil, and untreated waters (Figure
1-13). Food crops and livestock contain lead in varying proportions from the atmosphere and
natural sources. From the farm to the dinner table, lead is added to food as it is harvested,
transported, processed, packaged, and prepared. The sources of this lead are dusts of atmos-
pheric and industrial origin, metals used in grinding, crushing, and sieving, solder used in
packaging, and water used in cooking. Pennington (1983) has identified 234 typical food cate-
gories for Americans grouped into eight age/sex groups. These basic diets are the foundation
for the Food and Drug Administration's revised Total Diet Study, often called the "Market
Basket Study", beginning in April, 1982. The diets used for this discussion include food,
beverages, and drinking water for the 2-year-old child, the adult female 25 to 30 years of
age, and the adult male 25 to 30 years of age.
Milk and foods are treated separately from water and beverages because solder and atmos-
pheric lead contribute significantly to each of these later dietary components (Figure 1-1).
Between the field and the food processor, lead is added to food crops. It is assumed
that this lead is all of direct atmospheric origin. Direct atmospheric lead can be deposited
directly on food materials by dry deposition, or it can be lead on dust which has collected on
other surfaces, then transferred to foods. For the purposes of this discussion, it is not
necessary to distinguish between these two forms, as both are a function of air concentration.
For some of the food items, data are available on lead concentrations just prior to fil-
ling of cans. In the case where the food product has not undergone extensive modification
(e.g. cooking or added ingredients), the added lead was most likely derived from the atmos-
phere or from the machinery used to handle the product.
From the 'time a product is packaged in bottles, cans, or.plastic containers until it is
opened in the kitchen, it may be assumed that no food item receives atmospheric lead. Most of
the lead which is added during this stage comes from the solder used to seal some types of

SUMPB/D 1-42 9/30/83

-------
PRELIMINARY DRAFT
TABLE 1-7. SUMMARY BY AGE AND SEX OF ESTIMATED AVERAGE LEVELS
OF LEAD INGESTED FROM MILK AND FOODS
Dietary consumption
(g/day)
2-yr-oTd
Child
A.
B.
C.
D.
Dairy
Meat
Food crops
Canned food
Total
381
113
260
58
812
Adult
Female
237
169
350
68
824
Adult
Male
344
288
505
82
1219
ug
0.
0.
0.
0.

Pb/g*
013
036
022
24

Lead consumption
ua/day
2-yr-old Adult
Child Female
5.
4.
5.
13.
28.
0
1
7
9
7
3.1
6.1
7.7
16.3
33.2
Adult
Male
4.
10.
11.
19.
45.
5
4
1
7
6
"Weighted average lead concentration in foods from Table 7-15 in Chapter 7 of this document.

cans. Estimates by the Food and Drug Administration, prepared in cooperation with the
National Food Processors Association, suggest that lead in solder contributes more than 66
percent of the lead in canned foods where a lead solder side seam was used. This lead is
thought to represent a contribution of 20 percent to the total lead consumption in foods.
The contribution of the canning process to overall lead levels in albacore tuna has been
reported by Settle and Patterson (1980). The study showed that lead concentrations in canned
tuna are elevated above levels in fresh tuna by a factor of 4000. Nearly all of the increase
results from leaching of the lead from the soldered seam of the can; tuna from an unsoldered
can is elevated by a factor of only 20 compared with tuna fresh from the sea.
It is assumed that no further lead is added to food packaged in plastic or paper con-
tainers, although there are no data to support or reject this assumption.
Studies that reflect contributions of lead added during kitchen preparation showed that
lead in acidic foods stored refrigerated in open cans can increase by a factor of 2 to 8 in
five days if the cans have a lead-soldered side seam not protected by an interior lacquer
coating (Capar, 1978). Comparable products in cans with the lacquer coating or in glass jars
showed little or no increase.
As a part of its program to reduce the total lead intake by children (0-5 years) to less
than 100 ug/day by 1988, the Food and Drug Administration estimated lead intakes for individ-
ual children in a large-scale food consumption survey (Beloian and McDowell, 1981). Between
1973 and 1978, intensive efforts were made by the food industry to remove sources lead from
infant food items. By 1980, there had been a 47 percent reduction in the age group 0-5 months
and a 7 percent reduction for 6-23 months. Most of this reduction was accomplished by the
removal of soldered cans used for infant formula.
SUMPB/D 1-43 9/30/83

-------
PRELIMINARY DRAFT
Because the Food and Drug Administration is actively pursuing programs to remove lead
from adult foods, it is probable that there will be a decrease in total dietary lead consump-
tion over the next decade independent of projected decreases in atmospheric lead concentra-
tion. With both sources of lead minimized, the lowest reasonable estimated dietary lead con-
sumption would be 10-15 ug/day for adults and children. This estimate assumes about 90 per-
cent of the direct atmospheric, solder lead and lead of undetermined origin would be removed
from the diet, leaving 8 ug from these sources and 3 jjg of natural and indirect atmospheric
lead.
There have been several studies in North America and Europe of the sources of lead in
drinking water. The baseline concentration of water across the whole United States is taken
to be 10 ug/1, although 6-8 ug/1 are often cited in the literature for specific locations. A
recent study in Seattle, WA by Sharrett et al. (1982) showed that the age of the house and the
type of plumbing determined the lead concentration in tap water. Standing water from houses
newer than five years (copper pipes) averaged 31 ug/1, while houses less than 18 months old
averaged about 70 ug/1. Houses older than five years and houses with galvanized pipe averaged
less than 6 ug/1. The source of the water supply, the length of the pipe, and the use of
plastic pipes in the service line had little or no effect on the lead concentrations. It ap-
pears certain that the source of lead in new homes with copper pipes is the solder used to
join these pipes, and that this lead is eventually worn away with age.
Ingestion, rather than inhalation, of dust particles appears to be the greater problem in
the baseline environment, especially ingestion during meals and playtime activity by small
children. Although dusts are of complex origin, they may be conveniently placed into a few
categories relating to human exposure. Generally, the most convenient categories are house-
hold dusts, soil dust, street dusts, and occupational dusts. It is a characteristic of dust
particles that they accumulate on exposed surfaces and are trapped in the fibers of clothing
and carpets. Two other features of dusts are important. First, they must be described in
both concentration and amount; the concentration of lead in street dust may be the same in a
rural and urban environment, but the amount of dust may differ by a wide margin. Secondly,
each category represents some combination of sources. Household dusts contain some atmospher-
ic lead, some paint lead, and some soi'l lead; street dusts contain atmospheric, soil, and oc-
casionally paint lead. For the baseline human exposure, it is assumed that workers are not
exposed to occupational dusts, nor do they live in houses with interior leaded paints. Street
dust, soil dust, and some household dust are the primary sources for baseline potential human
exposure.
In considering the impact of street dust on the human environment, the obvious question
arises as to whether lead in street dust varies with traffic density. It appears that in non-

SUMPB/D 1-44 9/30/83

-------
PRELIMINARY DRAFT
urban environments, street dust ranges from 80 to 130 ug/g, whereas urban street dusts range
from 1,000 to 20.,000 ug/g. For the purpose of estimating potential human exposure, an average
value of 90 ug/g in street dust is assumed for baseline exposure and 1500 ug/g in the discuss-
ions of urban environments.
Household dust is also a normal component of the home environment. It accumulates on all
exposed surfaces, especially furniture, rugs, and windowsills. In some households of workers
exposed occupationally to lead dusts, the worker may carry dust home in amounts too small for
efficient removal but containing lead concentrations much higher than normal baseline values.
Most of the dust values for nonurban household environments fall in the range of 50 to
500 ug/g. A value of 300 ug/g is assumed. The only natural lead in dust would be some frac-
tion of that derived from soil lead. A value of 10 ug/g seems reasonable, since some of the
soil lead is of atmospheric origin. Children ingest about 5 times as much dust as adults,
most of the excess being street dusts from sidewalks and playgrounds. Exposure to occupation-
al lead by children would be through clothing brought home by parents.
The values derived or assumed in the proceeding sections are summarized on Table 1-8.
These values represent only consumption, not absorption of lead by the human body.

1.7.6 Additional Exposures
There are many conditions, even in nonurban environments, where an individual may in-
crease his lead exposure by choice, habit, or unavoidable circumstance. These conditions are
discussed as separate exposures to be added as appropriate to the baseline of human exposure
described above. Most of these additive effects clearly derive from air or dust, few from
water or food. Ambient air lead concentrations are typically higher in an urban than a rural
environment. This factor alone can contribute significantly to the potential lead exposure of
Americans, through increases in inhaled air and consumed dust. Produce from urban gardens may
also increase the daily consumption of lead. Some environments may not be related only to
urban living, such as houses with interior lead paint or lead plumbing, residences near
smelters or refineries, or family gardens grown on high-lead soils. Occupational exposures
may also be in an urban or rural setting. These exposures, whether primarily in the occupa-
tional environment or secondarily in the home of the worker, would be in addition to other ex-
posures in an urban location or from the special cases of lead-based paint or plumbing.
Urban atmospheres. The fact that urban atmospheres have more airborne lead than nonurban
contributes not only to lead consumed fay inhalation but also to increased amounts of lead in
dust. Typical urban atmospheres contain 0.5-1.0 ug Pb/m3. Other variable are the amount of
indoor filtered air breathed by urban residents, the amount of time spent indoors, and the
amount of time spent on freeways. Dusts vary from 500 to 3000 ug/g in urban environments.

SUMPB/D 1-45 9/30/83

-------
                           TABLE 1-8.  SUMMARY OF  BASELINE  HUMAN  EXPOSURES TO  LEAD
                                             Units are  in ng/day
Soil
Source
Child-2 yr old
Inhaled Air
Food
Water & beverages
Dust
Total
Percent
Adult female
Inhaled air
Food
Water & beverages
Dust
Total
Percent
Adult male
Inhaled air
Food
Water & beverages
Dust
Total
Percent
Total
Lead
Consumed

0.5
28.7
11.5
21.0
61.4
100%

1.0
33.2
17.9
4.5
56.6
100%

1.0
45.7
25.1
4.5
76.3
100%
Natural
Lead
Consumed

0.001
0.9
0.01
0.6
1.5
2.4%

0.002
1.0
0.01
0.2
1.2
2. IX

0.002
1.4
0.1
0.2
1.7
2.2%
Indirect
Atmospheric
Lead*

-
0.9
2.1
~
3.0
4.9%

-
1.0
3.4
-
4.4
7.8%

-
1.4
4.7
-
6.1
8.0%
Direct
Atmospheric
Lead*

0.5
10.9
1.2
19.0
31.6
51.5%

1.0
12.6
2.0
2.9
18.5
32.7%

1.0
17.4
2.8
2.9
24.1
31.6%
Lead from
Solder or
Other Metals

-
10.3
7.8
— !_
18.1
29.5%

-
11.9
12.5
-
24.4
43.1%

-
16.4
17.5
-
33.9
44.4%
Lead Of
Undetermined
Origin

-
17.6
-
1.4
19.0
22.6%

-
21.6
-
1.4
23.0
26.8%

-
31.5
-
1.4
32.9
27.1%
"Indirect atmospheric lead has been previously incorporated into soil, and will  probably remain in the soil
 for decades or longer.   Direct atmospheric lead has been deposited on the surfaces of vegetation and living
 areas or incorporated during food processing shortly before human consumption.

-------
PRELIMINARY DRAFT
Houses with Interior lead paint. In 1974, the Consumer Product Safety Commission col-
lected household paint samples and analyzed them for lead content (National Academy of
Sciences, National Research Council, 1976).
Flaking paint can cause elevated lead concentrations in nearby soil. For example, Hardy
et al. (1971) measured soil lead levels of 2000 ug/g next to a barn in rural Massachusetts. A
steady decrease in lead level with Increasing distance from the barn was shown, reaching 60
ug/g at fifty feet from the barn. Ter Haar and Arnow (1974) reported elevated soil lead
levels, in Detroit near eighteen old wood frame houses painted with lead-based paint. The
average soil lead level within two feet of a house was just over 2000 vg/g; the average con-
centration at ten feet was slightly more than 400 ug/g. The same author reported smaller soil
lead elevations in the vicinity of eighteen brick veneer houses in Detroit. Soil lead levels
near painted barns located in rural areas were similar to urban soil lead concentrations near
painted houses, suggesting the importance of leaded paint at both urban and rural locations.
The baseline lead concentration for household dust of 300 u^/fl was increased to 2000 uQ/9 for
houses with interior lead based paints. The additional 1700 ug/g would add 85 ug Pb/day to
the potential exposure of a child. This increase would occur in an urban or nonurban environ-
ment and would be in addition to the urban residential increase if the lead-based painted
house were in an urban environment.
Family gardens. Several studies have shown potentially higher lead exposure through the
consumption of home-grown produce from family gardens grown on high lead soils or near sources
of atmospheric lead. In family gardens, lead may reach the edible portions of vegetables by
deposition of atmospheric lead directly onto aboveground plant parts or onto soil, or by the
flaking of lead-containing paint chips from houses. Air concentrations and particle size dis-
tributions are the important determinants of deposition to soil or vegetation surfaces. Even
at relatively high air concentrations (1.5 ug/m3) and deposition velocity (0.5 cm/sec), it is
unlikely that surface deposition alone can account for more than 2-5 ug/g lead on the surface
of lettuce during a 21-day growing period. It appears that a significant fraction of the lead
in both leafy and root vegetables derives from the soil.
Houses with lead plumbing. The Glasgow Duplicate Diet Study (United Kingdom Directorate
on Environmental Pollution, 1982) reports that children approximately 13 weeks old living in
lead-plumbed houses consume 6-480 M9 Pb/day. Water lead levels in the 131 homes studied
ranged from less than 50 to over 500 ug/1. Those children and mothers living in the homes
containing high water lead levels generally had greater total lead consumption and higher
blood lead levels, according to the study. Breast-fed infants were exposed to much less lead
than bottle-fed infants. The results of the study suggest that infants living in lead-plumbed
homes may have exposure to considerable amounts of lead. This conclusion was also demonstrat-
ed by Sherlock et al. (1982) in * duplicate diet study in Ayr, Scotland.
SUMPB/D 1-47 9/30/83

-------
PRELIMINARY DRAFT
Residences near smelters and refineries. Air concentrations within 2 km of lead smelters
and refineries average 5-15 ng/m3. Between inhaled air and dust, a child in this circumstance
would be exposed to 1300 |jg Pb/day above background levels. Exposures to adults would be much
less, since they consume only 20 percent of the dusts children consume.
Occupational exposures. The highest and most prolonged exposures to lead are found among
workers in the lead smelting, refining, and manufacturing industries (World Health Organiza-
tion, 1977). In all work areas, the major route of lead exposure is by inhalation and inges-
tion of lead-bearing dusts and fumes. Airborne dusts settle out of the air onto food, water,
the workers' clothing, and other objects, and may be subsequently transferred to the mouth.
Therefore, good housekeeping and good ventilation have a major impact on exposure. Even tiny
amounts (10 mg) of 100,000 ug/g dust can account for 1,000 ug/day exposure.
The greatest potential for high-level occupational exposure exists in the process of lead
smelting and refining. The most hazardous operations are those in which molten lead and lead
alloys are brought to high temperatures, resulting in the vaporization of lead, because con-
densed lead vapor or fume has, to a substantial degree, a small (respirable) particle size
range.
When metals that contain lead or are protected with a lead-containing coating are heated
in the process of welding or cutting, copious quantities of lead in the respirable size range
may be emitted. Under conditions of poor ventilation, electric arc welding of zinc silicate-
coated steel (containing 29 mg Pb/in2 of coating) produces breathing-zone concentrations of
lead reaching 15,000 ug/m3, far in excess of 450 pg/m3, the current occupational short-term
exposure limit in the United States. In a study of salvage workers using oxy-acetylene cut-
ting torches on lead-painted structural steel under conditions of good ventilation, breathing-
zone concentrations of lead averaged 1200 ug/m3 and ranged as high as 2400 ug/m3.
At all stages in battery manufacture except for final assembly and finishing, workers are
exposed to high air lead concentrations, particularly lead oxide dust. Excessive concentra-
tions, as great as 5400 ug/m3, have been quoted by the World Health Organization (1977). The
hazard in plate casting, which is a molten-metal operation, is from the spillage of molten
waste products, resulting in dusty floors.
Workers involved in the manufacture of both tetraethyl lead and tetramethyl lead, two
alkyl lead compounds, are exposed to both inorganic and alkyl lead. The major potential
hazard in the manufacture of tetraethyl lead and tetramethyl lead is from skin absorption, but
this is guarded against by the use of protective clothing.
In both the rubber products industry and the plastics industry there are potentially high
exposures to lead. The potential hazard of the use of lead stearate as a stabilizer in the
manufacture of polyvinyl chloride was noted in the 1971 United Kingdom Department of Employ-
ment, Chief Inspector of Factories (1972). The source of this problem is the dust that is
SUMPB/D 1-48 9/30/83

-------
PRELIMINARY DRAFT
generated when the lead stearate is milled and mixed with the polyvinyl chloride and the plas-
ticizer. An encapsulated stabilizer that greatly reduces the occupational hazard is reported
by Fischbein et al. (1982). Sakurai et al. (1974), in a study of biolndicators of lead expo-
sure, found ambient air concentrations averaging 58 M9/m3 in the lead-covering department of a
rubber hose manufacturing plant.
The manufacture of cans with leaded seams may expose workers to elevated environmental
lead levels. Bishop (1980) reports airborne lead concentrations of 25 to 800 M9/m3 in several
can manufacturing plants in the United Kingdom. Between 23 percent and 54 percent of the air-
borne lead was associated with respirable particles. Firing ranges may be characterized by
high airborne lead concentrations, hence instructors who spend considerable amounts of time in
such areas may be exposed to lead. Anderson et al. (1977) discuss plumbism in a 17-year-old
male employee of a New York City firing range, where airborne lead concentrations as great as
1000 Mg/ro3 were measured during sweeping operations. Removal of leaded paint from walls and
other surfaces in old houses may pose a health hazard. Feldman (1978) reports an airborne
lead concentration of 510 \*g/m3, after 22 minutes of sanding an outdoor post coated with paint
containing 2.5 mg Pb/cm2. After only five minutes of sanding an indoor window sill containing
0.8-0.9 mg Pb/cm2, the air contained 550 ug/m3. Garage mechanics may be exposed to excessive
lead concentrations. Clausen and Rastogi (1977) report airborne lead levels of 0.2-35.5 ug/m3
in ten garages in Denmark; the greatest concentration was measured in a paint workshop. Used
motor oils were found to contain 1500-3500 M9 Pb/g, while one brand of gear oil, unused, con-
tained 9280 \tg Pb/g. The authors state that absorption through damaged skin could be an
important exposure pathway. Other occupations involving risk of lead exposure include stained
glass manufacturing and repair, arts and crafts, and soldering and splicing.
Secondary occupational exposure. The amount of lead contained in pieces of cloth 1 in2
cut from bottoms of trousers worn by lead workers ranged from 700 to 19,000 ug, with a median
of 2,640 ug. In all cases, the trousers were worn under coveralls. Dust samples from 25
households of smelter workers ranged from 120 to 26,000 H9/9. W1'th a median of 2,400 ug/g.
Special habits or activities. The quantity of food consumed per body weight varies
greatly with age and somewhat with sex. A two-year-old child weighing 14 kg eats and drinks
1.5 kg food and water per day. This is 110 g/kg, or 3 times the consumption of an 80 kg adult
male, who eats 39 g/kg.
Children place their mouths on dust collecting surfaces and lick non-food items with
their tongues. This fingersucking and mouthing activity are natural forms of behavior for
young children which expose them to some of the highest concentrations of lead in their envi-
ronment. A single gram of dust may contain ten times more lead than the total diet of the
child.

SUMPB/D 1-49 9/30/83

-------
PRELIMINARY DRAFT
Lead is also present in tobacco. The World Health Association (1977) estimates a lead
content of 2.5-12.2 ug per cigarette; roughly two to six percent of this lead may be inhaled
by the smoker. The National Academy of Sciences (1980) has used these data to conclude that a
typical urban resident who smokes 30 cigarettes per day may inhale roughly equal amounts of
lead from smoking and from breathing urban air. The average adult consumption of table wine
in the U.S. is about 12 g. Even at 0.1 |jg/g, which is ten times higher than drinking water,
wine does not appear to represent a significant potential exposure. At one liter/day,
however, lead consumption would be greater than the total baseline consumption. McDonald
(1981) points out that^older wines with lead foil caps may represent a hazard, especially if
they have been damaged or corroded. Wai et al. (1979) found the lead content of wine rose
from 200 to 1200 pg/liter when the wine was allowed to pass over the thin ring of residue left
by the corroded lead foil cap. Newer wines (1971 and later) use other means of sealing.
Pica is the compulsive, habitual consumption of non-food items. In the case of paint
chips and soil, this habit can present a significant lead exposure to the afflicted person.
There are very little data on the amounts of paint or soil eaten by children with varying de-
grees of pica. Exposure can only be expressed on a unit basis. Billick and Gray (1978) re-
port lead concentrations of 1000-5000 ug/cm2 in lead-based paint pigments. A single chip of
paint can represent greater exposure than any other source of lead. A gram of urban soil may
have 150-2000 yg lead.
Beyond the baseline level of human exposure, additional amounts of lead consumption are
largely a matter of individual choice or circumstance. Most of these additional exposures a-
rise directly or indirectly from atmospheric lead, and in one or more ways probably affect 90
percent of the American population. In some cases, the additive exposure can be fully quan-
tified and the amount of lead consumed can be added to the baseline consumption. These may be
continuous (urban residence), or seasonal (family gardening) exposures. Some factors can be
quantified on a unit basis because of wide ranges in exposure duration or concentration. For
example, factors affecting occupational exposure are air lead concentrations (10-4000 pg/m3).
use and efficiency of respirators, length of time of exposure, dust control techniques, and
worker training in occupational hygiene.
Ambient airborne lead concentrations showed no marked trend from 1965 to 1977. Over the
past five years, however, distinct decreases occurred. Mean urban air concentration has
3 3
dropped from 0.91 ug/m 1977 to 0.32 pg/m in 1980. These decreases reflect the smaller lead
emissions from mobile sources in recent years. Airborne size distribution data indicate that
most of the airborne lead mass is found in submicron particles. Atmospheric lead is deposited
on vegetation and soil surfaces, entering the human food chain through contamination of grains
and leafy vegetables, of pasture lands, and of soil moisture taken up by all crops. Lead con-
tamination of drinking water supplies appears to originate mostly from within the distribution
system.
SUHPB/D 1-50 9/30/83

-------
PRELIMINARY DRAFT
Most people receive the largest portion of their lead intake through foods. Unprocessed
foods such as fresh fruits and vegetables receive lead by atmospheric deposition as well as
uptake from soil; crops grown near heavily traveled roads generally have greater lead levels
than those grown at greater distances from traffic. For many crops the edible internal por-
tions of the plant (e.g., kernels of corn and wheat) have considerably less lead than the
outer, more exposed parts such as stems, leaves, and husks. Atmospheric lead accounts for
about 30 percent of the total adult lead exposure, and 50 percent of the exposure for chil-
dren. Processed foods have greater lead concentrations than unprocessed foods, due to lead
inadvertently added during processing. Foods packaged in spidered cans have much greater lead
levels than foods packaged in other types of containers. About 45 percent of the baseline
adult exposure to lead results from the use of solder lead in packaging food and distributing
drinking water.
Significant amounts of lead in drinking water can result from contamination at the water
source and from the use of lead solder in the water distribution system. Atmospheric deposi-
tion has been shown to increase lead in rivers, reservoirs, and other sources of drinking
water; in some areas, however, lead pipes pose a more serious problem. Soft, acidic water in
homes with lead plumbing may have excessive lead concentrations. Besides direct consumption
of the water, exposure may occur when vegetables and other foods are cooked in water con-
taining lead.
All.of the categories of potential lead exposure discussed above may influence or be in-
fluenced by dust and soil. For example, lead in street dust is derived primarily from
vehicular emissions, while leaded house dust may originate from nearby stationary or mobile
sources. Food and water may include lead adsorbed from soil as well as deposited atmospheric
material. Flaking leadbased paint has been shown to increase soil lead levels. Natural con-
centrations of lead in soil average approximately 15 ug/g; this natural lead, in addition to
anthropogenic lead emissions, influences human exposure.
Americans living in rural areas away from sources of atmospheric lead consume 50 to 75 ug
Pb/day from all sources. Circumstances which can increase this exposure are: urban residence
(25 to 100 ug/day), family garden on high lead soil (800 to 2000 ug/day), houses with interior
lead-based paint (20 to 85 ug/day), and residence near a smelter (400 to 1300 ug/day). Occu-
pational settings, smoking and wine consumption also can increase consumption of lead accord-
ing to the degree of exposure.
A number of manmade materials are known to contain lead, the most important being paint
and plastics. Lead-based paints, although no longer used, are a major problem in older homes.
Small children who ingest paint flakes can receive excessive lead exposure. Incineration of
plastics may emit large amounts of lead into the atmosphere. Because of the increasing use of

SUMPB/D 1-51 9/30/83

-------
PRELIMINARY DRAFT
plastics, this source is likely to become more important. Other manmade materials containing
lead include colored dyes, cosmetic products, candle wicks, and products made of pewter and
silver.
The greatest occupational exposures are found in the lead smelting and refining indus-
tries. Excessive airborne lead concentrations and dust lead levels are occasionally found in
primary and secondary smelters; smaller exposures are associated with mining and processing of
the lead ores. Welding and cutting of metal surfaces coated with lead-based paint may also
result in excessive exposure. Other occupations with potentially high exposures to lead in-
clude the manufacture of lead storage batteries, printing equipment, alkyl lead, rubber pro-
ducts, plastics, and cans; individuals removing lead paint from walls and those who work in
indoor firing ranges may also be exposed to lead.
Environmental contamination by lead should be measured in terms of the total amount of
lead emitted to the biosphere. American industry contributes several hundred thousand tons of
lead to the environment each year: 35,000 tons from petroleum additives, 50,000 tons from am-
munition, 45,000 tons in glass and ceramic products, 16,000 tons in paint pigments, 8,000 tons
in food can solder, and untold thousands of tons of captured wastes during smelting, refining,
and coal combustion. These are uses of lead which are generally not recoverable, thus they
represent a permanent contamination of the human or natural environment. Although much of
this lead is confined to municipal and industrial waste dumps, a large amount is emitted to
the atmosphere, waterways, and soil, to become a part of the biosphere.
Potential human exposure can be expressed as the concentrations of lead in those environ-
mental components (air, dust, food, and water) that interface with man. It appears that, with
the exception of extraordinary cases of exposure, about 100 mg of lead are consumed daily by
each American. This amounts to only 8 tons, or less than 0.01 percent of the total environ-
mental contamination.
1.8 EFFECTS OF LEAD ON ECOSYSTEMS
The principle sources of lead entering an ecosystem are: the atmosphere (from automotive
emissions), paint chips, spent ammunition, the application of fertilizers and pesticides, and
the careless disposal of lead-acid batteries or other industrial products. Atmospheric lead
is deposited on the surfaces of vegetation as well as on ground and water surfaces. In ter-
restrial ecosystems, this lead is transferred to the upper layers of the soil surface, where
it may be retained for a period of several years. The movement of lead within ecosystems is
influenced by the chemical and physical properties of lead and by the biogeochemical pro-
perties of the ecosystem. Lead is non-degradable, but in the appropriate chemical' environ-
ment, may undergo transformations which affect its solubility (e.g., formation of lead sulfate
SUMPB/D 1-52 9/30/83

-------
PRELIMINARY DRAFT
in soils), its bioavailability (e.g., chelation with humic substances), or its toxicity (e.g.,
chemical methylation). Although the situation is extremely complex, it is reasonable to state
that most plants cannot survive in soil containing 10,000 ug lead/g dry weight if the pH is
below 4.5 and the organic content is below 5 percent.
There is wide variation in the mass transfer of lead from the atmosphere to terrestrial
ecosystems. Smith and Siccama (1981) report 270 g/ha-yr in the Hubbard Brook forest of New
Hampshire, Lindberg and Harriss (1981) found 50 g/ha-yr in the Walker Branch watershed of
Tennessee; and Elias et al. (1976) found 15 g/ha-yr in a remote subalpine ecosystem of
California. Jackson and Watson (1977) found 1,000,000 g/ha-yr near a smelter in southeastern
Missouri. Getz et al. (1979) estimated 240 g/ha*yr by wet precipitation alone in a rural eco-
system largely cultivated, and 770 g/ha-yr in an urban ecosystem.
One factor causing great variation is remoteness from source, which translates to lower
air concentrations, smaller particles, and greater dependence on wind as a mechanism of depo-
sition. Another factor is type of vegetation cover. Deciduous leaves may, by the nature of
their surface and orientation in the wind stream, be more suitable deposition surfaces than
conifer needles.
There are three known conditions under which lead may perturb ecosystem processes (see
Figured 1-12). At soil concentrations of 1000 ug/g or higher, delayed decomposition may
result from the elimination of a single population of decomposer microorganisms. Secondly, at
concentrations of 500-1000 ug/g, populations of plants, microorganisms, and invertebrates may
shift toward lead tolerant populations of the same or different species. Finally, the normal
biogeochemical process which purifies and repurifies calcium in grazing and decomposer food
chains may be circumvented by the addition of lead to vegetation and animal surfaces. This
third effect can be measured at all ambient atmospheric concentrations of lead.
Some additional effects may occur due to the uneven distribution of lead in ecosystems.
It is known that lead accumulates in soil, especially soil with high organic content.
Although no firm documentation exists, it is reasonable to assume from the known chemistry of
lead in soil that: (1) other metals may be displaced from binding sites on the organic
matter; (2) the chemical breakdown of inorganic soil fragments may be retarded by interference
of lead with the action of fulvic acid on iron bearing crystals; and (3) lead in soil may be
in equilibrium with moisture films surrounding soil particles and thus available for uptake by
plants.
Two principles govern ecosystem functions: (1) energy flows through an ecosystem; and
(2) nutrients cycle within an ecosystem. Energy usually enters the ecosystem in the form of
sunlight and leaves as heat of respiration. Unlike energy, nutrient and non-nutrient elements
are recycled by the ecosystem and transferred from reservoir to reservoir in a pattern usually

SUMPB/D 1-53 9/30/83

-------
                                     PRELIMINARY DRAFT
                                                                        GRAZERS
                                 INORGANIC
                                 NUTRIENTS
     Rgure 1-12. This figure depicts cycling processes within the major components of a
     terrestrial ecosystem, i.e. primary producers, grazers and decomposers. Nutrient and
     non-nutrient elements are stored in reservoirs within these components. Processes
     that take place within reservoirs regulate the flow of elements between reservoirs
     along established pathways. The rate of flow is in part a function of the concentra-
     tion in the preceding reservoir. Lead accumulates in decomposer reservoirs which
     have a high binding capacity for this metal. It is likely that the rate of flow away
     from these reservoirs has increased in past decades and will continue to increase for
     some time until the decomposer reservoirs are in equilibrium with the entire
     ecosystem. Inputs to and outputs from the ecosystem as a whole are not shown.

     Source: Adapted from Swift et al. (1979).
SUMPB/D
                                          1-54
                                                                                   9/30/83

-------
PRELIMINARY DRAFT
referred to as a biogeochemfcal cycle (Brewer, 1979, p. 139). The reservoirs correspond ap-
proximately to the food webs of energy flow. Although elements may enter (e.g., weathering of
soil) or leave the ecosystem (e.g., stream runoff), the greater fraction of the available mass
of the element is usually cycled within the ecosystem.
Ecosystems have boundaries. These boundaries may be as distinct as the border of a pond
or as arbitrary as an imaginary circle drawn on a map. Many trace metal studies are conducted
in watersheds where some of the boundaries are determined by topography. For atmospheric in-
puts to terrestrial ecosystems, the boundary is usually defined as the surface of vegetation,
exposed rock or soil. Non-nutrient elements differ little from nutrient elements in their
biogeochemical cycles. Quite often, the cycling patterns are similar to those of a major
nutrient. In the case of lead, the reservoirs and pathways are very similar to those of
calcium.
Naturally occurring lead from the earth's crust is commonly found in soils and the atmos-
phere. Lead may enter an ecosystem by weathering of parent rock or by deposition of atmos-
pheric particles. This lead becomes a part of the nutrient medium of plants and the diet of
animals. All ecosystems receive lead from the atmosphere.
In prehistoric times, the contribution of lead from weathering of soil was probably about
4g Pb/ha«yr and from atmospheric deposition about 0.02 g Pb/ha-yr. Weathering rates are pre-
sumed to have remained the same, but atmospheric inputs are believed to have increased to
180 g/ha-yr in natural and some cultivated ecosystems, and 3000 g/ha-yr in urban ecosystems
and along roadways. In every terrestrial ecosystem of the Northern Hemisphere, atmospheric
lead deposition now exceeds weathering by a factor of at least 10, sometimes by as much as
1000.
Many of the effects of lead on plants, microorganisms, and ecosystems arise from the fact
that lead from atmospheric and weathering inputs is retained by soil. Geochemical studies
show that less than 3 percent of the inputs to a watershed leave by stream runoff. Lead in
natural soils now accumulates on the surface at an annual rate of 5-10 percent of the natural
lead. One effect of cultivation is that atmospheric lead is mixed to a greater depth than the
0-3 cm of natural soils.
Most of the effects on grazing vertebrates stem from the deposition of atmospheric par-
ticles on vegetation surfaces. Atmospheric deposition may occur by either of two mechanisms.
Wet deposition (precipitation scavenging through rainout or washout) generally transfers lead
directly to the soil. Dry deposition transfers particles to all exposed surfaces. Large par-
ticles (>4 pm) are transferred by gravitational mechanisms, small particles (<0.5 urn) are also
deposited by wind-related mechanisms.
If the air concentration is known, ecosystem inputs from the atmosphere can be predicted
over time and under normal conditions. These inputs and those from the weathering of soil
SUMPB/D 1-55 9/30/83

-------
PRELIMINARY DRAFT
determine the concentration of lead in the nutrient media of plants, animals, and micro-
organisms. It follows that the concentration of lead in the nutrient medium determines the
concentration of lead in the organism and this in turn determines the effects of lead on the
organism. The fundamental nutrient medium of a terrestrial ecosystem is the soil moisture
film which surrounds organic and inorganic soil particles. This film of water is in equi-
librium with other soil components and provides dissolved inorganic nutrients to plants.
Studies have shown the lead content of leafy vegetation to be 90 percent anthropogenic,
even in remote areas (Crump and Barlow, 1980; Elias et al., 1976, 1978). The natural lead
content of nuts and fruits may be somewhat higher than leafy vegetation, based on internal
lead concentrations of modern samples (Elias et al. 1982).
Because lead in soil is the source of most effects on plants, microorganisms, and eco-
systems, it is important to understand the processes that control the accumulation of lead in
soil. Major components of soil are: (1) fragments of inorganic parent rock material;
(2) secondary inorganic minerals; (3) organic constituents, primarily humic substances, which
are residues of decomposition or products of decomposer organisms; (4) Fe-Mn oxide films,
which coat the surfaces of all soil particles and have a high binding capacity for metals;
(5) soil microorganisms, most commonly bacteria and fungi, although protozoa and soil algae may
also be found; and (6) soil moisture, the thin film of water surrounding soil particles which
is the nutrient medium of plants.
The concentration of lead ranges from 5 to 30 ug/g in the top 5 cm of most soils not
adjacent to sources of industrial lead, although 5 percent of the soils contain as much .as
800 ug/g. Aside from surface deposition of atmospheric particles, plants in North America
average about 0.5-1 ug/g dw (Peterson, 1978) and animals roughly 2 ug/g (Forbes and Sanderson,
1978). Thus, soils contain the greater part of total ecosystem lead. In soils, lead in
parent rock fragments is tightly bound within the crystalline structures of the inorganic soil
minerals. It is released to the ecosystem only by surface contact with soil moisture films.
Hutchinson (1980) has reviewed the effects of acid precipitation on the ability of soils
to retain cations. Excess calcium and other metals are leached from the A horizon of soils by
rain with a pH more acidic than 4.5. Most soils in the eastern United States are normally
acidic (pH 3.5-5.2) and the leaching process is a part of the complex equilibrium maintained
in the soil system. By increasing the leaching rate, acid rain can reduce the availability of
nutrient metals to organisms dependent on the top layer of soil. It appears that acidifica-
tion of soil may increase the rate of removal of lead from the soil, but not before several
major nutrients are removed first. The effect of acid rain on the retention of lead by soil
moisture is not known.
Atmospheric lead may enter aquatic ecosystems by wet or dry deposition or by the
erosional transport of soil particles. In waters not polluted by industrial, agricultural, or
SUMPB/D 1-56 9/30/83

-------
PRELIMINARY DRAFT
municipal effluents, the lead concentration is usually less than 1 ug/1- Of this amount,
approximately 0.02 M9/1 Is natural lead and the rest is anthropogenic lead, probably of atmos-
pheric origin (Patterson, 1980). Surface waters mixed with urban effluents may frequently
reach lead concentrations of 50 M9/1. and occasionally higher. In still water, lead is
removed front the water column by the settling of lead-containing particulate matter, by the
formation of insoluble complexes, or by the adsorption of lead onto suspended organic
particles. The rate of sedimentation is determined by temperature, pH, oxidation-reduction
potential, ionic competition, the chemical form of lead in water, and certain biological acti-
vities (Jenne and Luoma, 1977). McNurney et al. (1977) found 14 pg Pb/g in stream sediments
draining cultivated areas and 400 ug/g in sediments associated with urban ecosystems.

1.8.1 Effects on Plants
Some physiological and biochemical effects of lead on vascular plants have been detected
under laboratory conditions at concentrations higher than normally found in the environment.
The commonly reported effects are the inhibition of photosynthesis, respiration or cell elon-
gation, all of which reduce the growth of the plant (Koeppe, 1981). Lead may also induce pre-
mature senescence, which may affect the long-term survival of the plant or the ecological suc-
cess of the plant population. Most of the lead in or on a plant occurs on the surfaces of
leaves and the trunk or stem. The surface concentration of lead in trees, shrubs, and grasses
exceeds the internal concentration by a factor of at least five (Elias et al, 1978). There is
little or no evidence of lead uptake through leaves or bark. Foliar uptake, if it does occur,
cannot account for more than 1 percent of the uptake by roots, and passage of lead through
bark tissue has not been detected (Arvik and Zimdahl, 1974; reviewed by Koeppe, 1981; Zimdahl,
1976). The major effect of surface lead at ambient concentrations seems to be on subsequent
components of the grazing food chain and on the decomposer food chain following litterfall
(Elias et al., 1982).
Uptake by roots is the only major pathway for lead into plants. The amount of lead that
enters plants by this route is determined by the availability of lead in soil, with apparent
variations according to plant species. Soil cation exchange capacity, a major factor, is de-
termined by the relative size of the clay and organic fractions, soil pH, and the amount of
Fe-Mn oxide films present (Nriagu, 1978). Of these, organic humus and high soil pH are the
dominant factors in immobilizing lead. Under natural conditions, most of the total lead in
soil would be tightly bound within the crystalline structure of inorganic soil fragments, un-
available to soil moisture. Available lead, bound on clays, organic colloids, and Fe-Mn
films, would be controlled by the slow release of bound lead from inorganic rock sources. Be-
cause lead is strongly immobilized by humic substances, only a small fraction (perhaps 0.01 -
percent in soils with 20 percent organic matter, pH 5.5) is released to soil moisture.
SUMPB/D 1-57 9/30/83

-------
PRELIMINARY DRAFT
Two defensive mechanims appear to exist in the roots of plants for removing lead from the
stream of nutrients flowing to the above-ground portions of plants. Lead may be deposited
with cell wall material exterior to the individual root cells, or may be sequestered in organ-
elles within the root cells. Any lead not captured by these mechanisms would likely move with
nutrient metals cell-to-cell through the symplast and into the vascular system. Uptake of
lead by plants may be enhanced by symbiotic associations with mycorrhizal fungi. The three
primary factors that control the uptake of nutrients by plants are the surface area of the
roots, the ability of the root to absorb particular ions, and the transfer of ions through the
soil. The symbiotic relationship between mycorrhizal fungi and the roots of higher plants can
increase the uptake of nutrients by enhancing all three of these factors.
The translocation of lead to aboveground portions of the plant is not clearly understood.
Lead may follow the same pathway and be subject to the same controls as a nutrient metal such
as calcium. There may be several mechanisms that prevent the translocation of lead to other
plant parts. The primary mechanisms may be storage in cell organelles or adsorption on cell
walls. Some lead passes into the vascular tissue, along with water and dissolved nutrients,
and is carried to physiologically active tissue of the plant. Evidence that lead in contami-
nated soils can enter the vascular system of plants and be transported to aboveground parts
may be found in the analysis of tree rings. These chronological records confirm that lead can
be translocated in proportion to the concentrations of lead in soil.
Because most of the physiologically active tissue of plants is involved in growth, main-
tenance, and photosynthesis, it is expected that lead might interfere with one or more of
these processes. Indeed, such interferences have been observed in laboratory experiments at
lead concentrations greater than those normally found in the field, except near smelters or
mines (Koeppe, 1981). Inhibition of photosynthesis by lead may be by direct interference with
the light reaction or the indirect interference with carbohydrate synthesis. Miles et al.
(1972) demonstrated substantial inhibition of photosystem II near the site of water splitting,
a biochemical process believed to require manganese. Devi Prasad and Devi Prasad (1982) found
10 percent inhibition of pigment production in three species of green algae at 1 ug/g, in-
creasing to 50 percent inhibition at 3 ug/g. Bazzaz et al. (1974, 1975) observed reduced net
photosynthesis which may have been caused indirectly by inhibition of carbohydrate synthesis.
The stunting of plant growth may be by the inhibition of the growth hormone IAA (indole-
3-ylacetic acid). Lane et al. (1978) found a 25 percent reduction in elongation at 10 ug/g
lead as lead nitrate in the nutrient medium of wheat coleoptiles. Lead may also interfere
with plant growth by reducing respiration or inhibiting cell division. Miller and Koeppe
(1971) and Miller et al. (1975) showed succinate oxidation inhibition in isolated mitochondria
as well as stimulation of exogenous NADH oxidation with related mitochondrial swelling.

SUMPB/D 1-58 9/30/83

-------
PRELIMINARY DRAFT
Hassett et al. (1976), Koeppe (1977), and Malone et al. (1978) described significant inhibi-
tion of lateral root initiation in corn. The interaction of lead with calcium has been shown
by several authors, most recently by Garland and Wilkins (1981), who demonstrated that barley
seedlings (Hordeum vulgare). which were growth inhibited at 2 ug Pb/g sol. with no added
calcium, grew at about half the control rate with 17 pg Ca/g sol. This relation persisted up
to 25 M9 Pb/g sol. and 500 jjg Ca/g sol.
These studies of the physiological effects of lead on plants all show some effect at con-
centrations from 2 to 10 ug/g in the nutrient medium of hydroponically-grown agricultural
plants. It is certain that no effects would have been observed at these concentrations had
the lead solutions been added to normal soil, where the lead would have been bound by humic
substances. There is no firm relationship between soil lead and soil moisture lead, because
each soil type has a unique capacity to retain lead and to release that lead to the soil
moisture film surrounding the soil particle. Once in soil moisture, lead seems to pass freely
to the plant root according to the capacity of the plant root to absorb water and dissolved
substances.
It seems reasonable that there may be a direct correlation between lead in hydroponic
media and lead in soil moisture. Hydroponic media typically have an excess of essential
nutrients, including calcium and phosphorus, so that movement of lead from hydroponic media to
plant root would be equal to or slower than movement from soil moisture to plant root.
Even.under the best of conditions where soil has the highest capacity to retain lead,
most plants would experience reduced growth rate (inhibition of photosynthesis, respiration,
or cell elongation) in soils containing 10,000 pg Pb/g or greater. Concentrations approaching
this value typically occur around smelters and near major highways. These conclusions pertain
to soil with the ideal composition and pH to retain the maximum amount of lead. Acid soils or
soils lacking organic matter would inhibit plants at much lower lead concentrations.
The rate at which atmospheric lead accumulates in soil varies from 1.1 mg/m2-yr average
global deposition to 3000 mg/m2-yr near a smelter. Assuming an average density of 1.5 g/cm3,
undisturbed soil to a depth of 2 cm (20,000 cnrVm2) would incur an increase in lead concen-
tration at a rate of 0.04 to 100 ug/g soil-yr. This means remote or rural area soils may
never reach the 10,000 ug/g threshold but that undisturbed soils closer to major sources may
be within range in the next 50 years.
Some plant species have developed populations tolerant to high lead soils. Using popu-
lations taken from mine waste and uncontaminated control areas, some authors have quantified
the degree of tolerance of Agrostis tenuis (Karataglis, 1982) and Festuca rubra (Wong, 1982)
under controlled laboratory conditions. Root elongation was used as the index of tolerance.
At 36 ug Pb/g nutrient solution, all populations of A. tenuis were completely inhibited. At
12 pg Pb/g, the control populations from low lead soils were completely inhibited, but the
SUMPB/D 1-59 9/30/83

-------
PRELIMINARY DRAFT
populations from mine soils achieved 30 percent of their normal growth (growth at no lead in
nutrient solution). At 6 ug/g, the control populations achieved 10 percent of their normal
growth, tolerant populations achieved 42 percent. There were no measurements below 6 |jg/g.
These studies support the conclusion that inhibition of plant growth begins at a lead concen-
tration of less than 1 ug/g soil moisture and becomes completely inhibitory at a level between
3 and 10 ug/g. Plant populations that are genetically adapted to high lead soils may achieve
50 percent of their normal root growth at lead concentrations above 3 ug/g.
When soil conditions allow lead concentrations in soil moisture to exceed 2-10 ug/g, most
plants experience reduced growth due to the inhibition of one or more physiological processes.
Excess calcium or phosphorus may reverse the effect. Plants that absorb nutrients from deeper
soil layers may receive less lead. Acid rain is not likely to release more lead until after
major nutrients have been depleted from the soil. A few species of plants have the genetic
capability to adapt to high lead soils.
Tyler (1972) explained three ways in which lead might interfere with the normal decompo-
sition processes in a terrestrial ecosystem. Lead may be toxic to specific groups of decom-
posers, it may deactivate en2ymes excreted by decomposers to break down organic matter, or it
may bind with the organic matter to render it resistant to the action of decomposers. Because
lead in litter may selectively inhibit decomposition by soil bacteria at 2000-5000 ug/g,
forest floor nutrient cycling processes may be seriously disturbed near lead smelters. This
is especially important because approximately 70 percent of plant biomass enters the de-
composer food chain. If decomposition of the biomass is inhibited, then much of the energy
and nutrients remain unavailable to subsequent components of the food chain. There is also
the possibility that the ability of soil to retain lead would be reduced, as humic substances
are byproducts of bacterial decomposition. Because they are interdependent, the absence of
one decomposer group in the decomposition food chain seriously affects the success of sub-
sequent groups, as well as the rate at which plant tissue decomposes. Each group may be
affected in a different way and at different lead concentrations. Lead concentrations toxic
to decomposer microbes may be as low as 1 to 5 ug/g or as high as 5000 ug/g. Under conditions
of mild contamination, the loss of one sensitive bacterial population may result in its
replacement by a more lead-tolerant strain. Decayed decomposition has been reported near
smelters, mine waste dumps, and roadsides. This delay is generally in the breakdown of litter
from the first stage (0^ to the second (02), with intact plant leaves and twigs accumulating
at the soil surface. The substrate concentrations at which lead inhibits decomposition appear
to be very low.
The conversion of ammonia to nitrate in soil is a two-step process mediated by two genera
of bacteria, Nitrosomonas and Nitrobacter. Nitrate is required by all plants, although some