oEPA
EPA/600/R-17/3 84F
September 2017
Update for Chapter 5 of the Exposure Factors Handbook
Soil and Dust Ingestion
National Center for Environmental Assessment
Office of Research and Development
U.S. Environmental Protection Agency
Washington, DC 20460

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
DISCLAIMER
This document has been reviewed in accordance with U.S. Environmental Protection
Agency policy and approved for publication. Mention of trade names or commercial products
does not constitute endorsement or recommendation for use.
September 2017
Page 5-ii

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
TABLE OF CONTENTS
LIST OF TABLES	5-vi
LIST OF FIGURES	5-vii
5. SOIL AND DUST INGESTION	5-1
5.1.	INTRODUCTION	5-1
5.2.	RECOMMENDATIONS	5-3
5.3.	KEY AND RELEVANT STUDIES	5-10
5.3.1.	Methodologies Used in Key Studies	5-10
5.3.1.1.	Tracer Element Methodology	5-10
5.3.1.2.	Biokinetic Model Comparison Methodology	5-12
5.3.1.3.	Activity Pattern Methodology	5-12
5.3.2.	Key Studies of Primary Analysis	5-12
5.3.2.1.	Vermeer and Frate (1979)—Geophagia in Rural Mississippi: Environmental
and Cultural Contexts and Nutritional Implications	5-13
5.3.2.2.	Calabrese et al. (1989)—How Much Soil Do Young Children Ingest: An
Epidemiologic Study/Barnes (1990)—Childhood Soil Ingestion: How Much
Dirt Do Kids Eat?/Calabrese et al. (1991)—Evidence of Soil Pica Behavior
and Quantification of Soil Ingested	5-13
5.3.2.3.	Davis et al. (1990)—Quantitative Estimates of Soil Ingestion in Normal
Children between the Ages of 2 and 7 Years: Population-Based Estimates
Using Aluminum, Silicon, and Titanium as Soil Tracer Elements	5-14
5.3.2.4.	Calabrese et al. (1997a)—Soil Ingestion Estimates for Children Residing on
a Superfund Site	5-14
5.3.2.5.	Stanek et al. (1998)—Prevalence of Soil Mouthing/Ingestion among
Healthy Children Aged One to Six/Calabrese et al. (1997b)—Soil Ingestion
Rates in Children Identified by Parental Observation as Likely High Soil
Ingesters	5-15
5.3.2.6.	Davis and Mirick (2006)—Soil Ingestion in Children and Adults in the Same
Family	5-16
5.3.3.	Key Studies of Secondary Analysis	5-16
5.3.3.1.	Calabrese and Stanek (1995)—Resolving Intertracer Inconsistencies in Soil
Ingestion Estimation	5-17
5.3.3.2.	Stanek and Calabrese (1995a)—Soil Ingestion Estimates for Use in Site
Evaluations Based on the Best Tracer Method	5-17
5.3.3.3.	Hogan et al. (1998)—Integrated Exposure Uptake Biokinetic Model for
Lead in Children: Empirical Comparisons with Epidemiologic Data	5-18
5.3.3.4.	Ozkaynak et al. (2011)—Modeled Estimates of Soil and Dust Ingestion
Rates for Children	5-19
5.3.3.5.	Wilson et al. (2013)—Revisiting Dust and Soil Ingestion Rates Based on
Hand-to-Mouth Transfer	5-20
5.3.3.6.	Von Lindern et al. (2016)—Estimating Children's Soil/Dust Ingestion Rates
through Retrospective Analyses of Blood Lead Biomonitoring from the
Bunker Hill Superfund Site in Idaho	5-21
5.3.4.	Relevant Studies of Primary Analysis	5-22
5.3.4.1.	Dickins and Ford (1942)—Geophagy (Dirt Eating) among Mississippi
Negro School Children	5-22
5.3.4.2.	Ferguson and Keaton (1950)—Studies of the Diets of Pregnant Women in
Mississippi: II Diet Patterns	5-22
5.3.4.3.	Cooper (1957)—Pica: A Survey of the Historical Literature as Well as
Reports from the Fields of Veterinary Medicine and Anthropology, the
Present Study of Pica in Young Children, and a Discussion of Its Pediatric
and Psychological Implications	5-23
5.3.4.4.	Barltrop (1966)—The Prevalence of Pica	5-23
September 2017	page s-jii

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
TABLE OF CONTENTS (Continued)
5.3.4.5.	Bruhn and Pangborn (1971)—Reported Incidence of Pica among Migrant
Families	5-23
5.3.4.6.	Robischon (1971)—Pica Practice and Other Hand-Mouth Behavior and
Children's Developmental Level	5-23
5.3.4.7.	Bronstein and Dollar (1974)—Pica in Pregnancy	5-23
5.3.4.8.	Hook (1978)—Dietary Cravings and Aversions during Pregnancy	5-24
5.3.4.9.	Binder etal. (1986)—Estimating Soil Ingestion: The Use of Tracer Elements
in Estimating the Amount of Soil Ingested by Young Children	5-24
5.3.4.10.	Clausing et al. (1987)—AMethod for Estimating Soil Ingestion by Children
	5-25
5.3.4.11.VanWijnenetal.	(1990)—Estimated Soil Ingestion by Children	5-25
5.3.4.12.Calabrese	et al. (1990)—Preliminary Adult Soil Ingestion Estimates:
Results of a Pilot Study	5-26
5.3.4.13.Cooksey	(1995)—Pica and Olfactory Craving of Pregnancy: How Deep Are
the Secrets?	5-26
5.3.4.14.	Smulian et al. (1995)—Pica in a Rural Obstetric Population	5-27
5.3.4.15.Grigsby	et al. (1999)—Chalk Eating in Middle Georgia: A Culture-Bound
Syndrome of Pica?	5-27
5.3.4.16.	Ward and Kutner (1999)—Reported Pica Behavior in a Sample of Incident
Dialysis Patients	5-27
5.3.4.17.Simpson et al. (2000)—Pica During Pregnancy in Low-Income Women
Born in Mexico	5-28
5.3.4.18.	Obialo etal. (2001)—Clay Pica Has No Hematologic or Metabolic Correlate
to Chronic Hemodialysis Patients	5-28
5.3.4.19.Klitzman	et al. (2002)—Lead Poisoning among Pregnant Women in New
York City: Risk Factors and Screening Practices	5-28
5.3.4.20.	Doyle etal. (2012)—A Soil Ingestion Pilot Study of a Population Following
a Traditional Lifestyle Typical of Rural or Wilderness Areas	5-28
5.3.4.21.Irvine et al. (2014)—Soil Ingestion Rate Determination in a Rural
Population in Alberta, Canada Practicing a Wilderness Lifestyle	5-29
5.3.4.22.Lumish	et al. (2014)—Gestational Iron Deficiency is Associated with Pica
Behaviors in Adolescents	5-29
5.3.4.23.Jang	et al. (2014)—General Factors of the Korean Exposure Factors
Handbook	5-29
5.3.4.24.	Chien et al. (2015)—Soil Ingestion Rates for Children under 3 Years Old in
Taiwan	5-30
5.3.4.25.Wang	et al. (2015)—Quantification of Soil/Dust (SD) on the Hands of
Children from Hubei Province, China Using Hand Wipes	5-31
5.3.4.26.Lin	et al. (2015)—Pica during Pregnancy among Mexican-Born Women: A
Formative Study	5-31
5.3.4.27.Ma	et al. (2016)—Estimation of the Daily Soil/Dust (SD) Ingestion Rate of
Children from Gansu Province, China via Hand-to-Mouth Contact Using
Tracer Elements	5-32
5.3.5. Relevant Studies of Secondary Analysis and Other Relevant Information	5-32
5.3.5.1.	Wong (1988)—The Role of Environmental and Host Behavioral Factors in
Determining Exposure to Infection with Ascaris lumbricoides and Trichuris
trichiura/Calabrese and Stanek (1993)—Soil Pica: Not a Rare Event	5-33
5.3.5.2.	Calabrese and Stanek (1992b)—What Proportion of Household Dust is
Derived from Outdoor Soil?	5-33
5.3.5.3.	Stanek and Calabrese (1995b)—Daily Estimates of Soil Ingestion in
Children	5-33
September 2017
Page 5-iv

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
TABLE OF CONTENTS (Continued)
5.3.5.4.	Calabrese et al. (1996)—Methodology to Estimate the Amount and Particle
Size of Soil Ingested by Children: Implications for Exposure Assessment at
Waste Sites	5-34
5.3.5.5.	Stanek et al. (1999)—Soil Ingestion Estimates for Children in Anaconda
Using Trace Element Concentrations in Different Particle Size Fractions	5-34
5.3.5.6.	Stanek and Calabrese (2000)—Daily Soil Ingestion Estimates for Children
at a Superfund Site	5-34
5.3.5.7.	Stanek et al. (2001a)—Biasing Factors for Simple Soil Ingestion Estimates
in Mass Balance Studies of Soil Ingestion	5-35
5.3.5.8.	Stanek et al. (2001b)—Soil Ingestion Distributions for Monte Carlo Risk
Assessment in Children	5-35
5.3.5.9.	Von Lindern et al. (2003)—Assessing Remedial Effectiveness through the
Blood Lead: Soil/Dust Lead Relationship at the Bunker Hill Superfund Site
in the Silver Valley of Idaho	5-35
5.3.5.10.Layton	and Beamer (2009)—Migration of Contaminated Soil and Airborne
Particulates to Indoor Dust	5-36
5.3.5.11.	Gavrelis et al. (2011)—An Analysis of the Proportion of the U. S. Population
That Ingests Soil or Other Non-Food Substances	5-36
5.3.5.12.Stanek et al. (2012a)—Meta-Analysis of Mass-Balance Studies of Soil
Ingestion in Children	5-37
5.3.5.13.	Wilson et al. (2015)—Estimation of Sediment Ingestion Rates Based on
Hand-to-Mouth Contact and Incidental Surface Water Ingestion	5-37
5.3.5.14.	Wilson et al. (2016)—Estimation of Dust Ingestion Rates inUnits of Surface
Area per Day Using a Mechanistic Hand-to-Mouth Model	5-38
5.3.5.15.Fawcett	et al. (2016)—A Meta-Analysis of the Worldwide Prevalence of
Pica during Pregnancy and the Postpartum Period	5-38
5.4.	LIMITATIONS OF STUDY METHODOLOGIES	5-39
5.4.1.	Tracer Element Methodology	5-39
5.4.2.	Biokinetic Model Comparison Methodology	5-42
5.4.3.	Activity Pattern Methodology	5-43
5.4.4.	Environmental and Household Interventions	5-44
5.4.5.	Key Studies: Representativeness of the U.S. Population	5-44
5.5.	DERIVATION OF RECOMMENDED SOIL AND DUST INGESTION VALUES	5-46
5.5.1.	Central Tendency Soil and Dust Ingestion Recommendations	5-47
5.5.2.	Upper Percentile, Soil Pica, and Geophagy Recommendations	5-49
5.6.	REFERENCES FOR CHAPTER 5	5-52
APPENDIX A	A-l
APPENDIX B	B-l
APPENDIX C	C-l
APPENDIX D	D-l
September 2017
Page 5-v

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
LIST OF TABLES
Table 5-1. Recommended Values for Daily Soil, Dust, and Soil + Dust Ingestion (mg/day)	5-7
Table 5-2. Confidence in Recommendations for Ingestion of Soil and Dust	5-8
Table 5-3. Soil, Dust, and Soil + Dust Ingestion Estimates for Amherst, MA Study Children
(1 to <4 Years)	5-58
Table 5-4. Amherst, MA Soil Pica Child's Daily Ingestion Estimates by Tracer and by Week (mg/day).... 5-59
Table 5-5. Estimated Soil Ingestion for Sample of Washington State Children (2-7 years; N= 101)	5-59
Table 5-6. Soil Ingestion Estimates for 64 Anaconda Children (Ages 1-4 Years)	5-60
Table 5-7. Soil Ingestion Estimates for Massachusetts Child Displaying Soil-Pica Behavior (mg/day)	5-60
Table 5-8. Average Daily Soil and Dust Ingestion Estimate for Children 1-3 Years (mg/day)	5-61
Table 5-9. Mean and Median Soil Ingestion (mg/day) by Family Members	5-61
Table 5-10. Positive/Negative Error (Bias) in Soil Ingestion Estimates in Calabrese et al. (1989) Study:
Effect on Mean Soil Ingestion Estimate (mg/day)	5-62
Table 5-11. Comparison of Soil Ingestion Estimates (mg/day) from Two Sites	5-63
Table 5-12. Predicted Soil and Dust Ingestion Rates for Children Age 3 to <6 Years (mg/day)	5-64
Table 5-13. Age-Dependent Probability Density Functions Used to Estimate Dust and Soil Ingestion
Rates via the Activity Pattern Modeling Approach	5-65
Table 5-14. Soil and Dust Ingestion Rates, Estimated Using a Probabilistic Activity Pattern Modeling
Approach	5-66
Table 5-15. Age-Specific Central Tendency Soil/Dust Ingestion Rates for Four Scenarios That Best
Predict Observed Blood Lead Levels (mg/day)	5-66
Table 5-16. Age-Specific Distributions of Soil and Dust Intake Rates for the Four Partition Scenarios
(mg/day)	5-67
Table 5-17. Estimated Daily Soil Ingestion for East Helena, MT Children Ages 1-3 years ( V =59)	5-68
Table 5-18. Estimated Soil Ingestion for Sample of Dutch Nursery School Children, Ages 2-4 Years	5-69
Table 5-19. Estimated Soil Ingestion for Sample of Dutch Hospitalized, Bedridden Children, Ages 2-4
Years	5-70
Table 5-20. Van Wijnen et al. (1990) Limiting Tracer Method (LTM) Soil Ingestion Estimates for
Sample of Dutch Children	5-70
Table 5-21. Estimated Geometric Mean Limiting Tracer Method (LTM) Soil Ingestion Values of
Children Attending Daycare Centers According to Age, Weather Category, and Sampling
Period	5-71
Table 5-22. Items Ingested by Low-Income Mexican-Born Women (Ages 18-42 Years) Who
Practiced Pica during Pregnancy While in the United States (N = 46)	5-72
Table 5-23. Soil Ingestion Rates for the Four Most Reliable Tracers (Aluminum, Silicon, Lanthanum,
and Cerium), Aluminum and Silicon Combined, and All Four Tracers Combined (mg/day)	5-72
Table 5-24. Soil Ingestion Rates for Aluminum, Silicon, and the Four Most Reliable Tracers
(Aluminum, Silicon, Lanthanum, and Cerium) Combined (mg/day)	5-73
Table 5-25. Estimated Soil Ingestion for Six Jamaican Children Displaying Soil Pica	5-73
Table 5-26. Distribution of Average (Mean) Daily Soil Ingestion Estimates per Child for 64 Children,
Ages 1 to <4 Years (mg/day)	5-74
Table 5-27. Estimated Distribution of Individual Mean Daily Soil Ingestion Based on Data for 64 Subjects
(Ages 1 to <4 Years) Projected over 365 Days	5-74
Table 5-28. Distribution of Daily Soil Ingestion (mg/day) Over 7 Days, 64 Children (1-4 years) from
Anaconda, MT	5-75
Table 5-29. Prevalence of Nonfood Consumption by Substance fromNHANES I and NHANES II	5-76
Table 5-30. Results of Meta-Analysis on Soil Ingestion (mg/day)	5-77
Table 5-31. Age-Dependent Probability Density Functions Used to Estimate Sediment Ingestion Rates
Using an Activity Pattern Modeling Approach	5-78
Table 5-32. Estimated Sediment Ingestion Rates (mg/hour) Using an Activity Pattern Modeling
Approach	5-79
Table 5-33. Dust Ingestion Rates at Residential Settings Based on an Activity Pattern Modeling
Approach	5-79
September 2017
Page 5-vi

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
LIST OF TABLES (CONTINUED)
Table 5-34. Summary of Estimates of Soil and Dust Ingestion by General Population Children and
Adults from Key Studies Using the Tracer Study, Biokinetic Modeling, and Activity
Pattern Methodologies (mg/day)	5-80
Table A-l. Terms Used in Literature Searches	A-l
Table B-l. Distributions of Soil Ingestion Rates (mg/day) Based on Different Methods of Analyzing
Data from the Tracer Studies	B-l
Table C-l. Key Soil Ingestion Studies Used in Developing Soil + Dust Ingestion Recommendation for
Use in Risk Assessment	C-l
Table D-l. Studies on the Prevalence of Ingesting Soil, Dust, or Other Nonfood Substances	D-l
September 2017
Page 5-vii

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
LIST OF FIGURES
Figure 5-1. Prevalence of Nonfood Substance Consumption by Age, NHANES I and NHANES II	5-83
Figure 5-2. Prevalence of Nonfood Substance Consumption by Race, NHANES I and NHANES II	5-84
Figure 5-3. Prevalence of Nonfood Substance Consumption by Income, NHANES I (1971-1975) and	
NHANES II (1976-1980)	 5-85
September 2017
Page 5-viii

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
5. SOIL AND DUST INGESTION
5.1. INTRODUCTION
This document is an update to Chapter 5 (Soil
and Dust Ingestion) of the Exposure Factors
Handbook: 2011 Edition. New information that has
become available since 2011 has been added, and the
recommended values have been revised, as needed,
to reflect the additional information. The chapter
includes a comprehensive review of the scientific
literature through 2016. The new literature was
identified via formal literature searches conducted
by EPA library services as well as targeted internet
searches conducted by the authors of this chapter.
Appendix Aprovides a list of the key terms that were
used in the literature searches. Revisions to this
chapter have been made in accordance with the
approved quality assurance plan for the Exposure
Factors Handbook.
The ingestion of soil and dust is a potential route
of exposure to environmental chemicals for both
adults and children. Children, in particular, may
ingest significant quantities of soil and dust due to
their tendency to play on the floor indoors and on the
ground outdoors and their tendency to mouth objects
or their hands. For example, children may ingest soil
and dust through deliberate hand-to-mouth
movements, or unintentionally by eating food or
mouthing objects that have dropped on the floor.
Adults may also ingest soil or dust particles that
adhere to food, cigarettes, or their hands. Other
vulnerable populations may include pregnant
women and populations engaging in wilderness and
traditional rural lifestyles. Thus, understanding soil
and dust ingestion patterns is an important part of
estimating overall exposures to environmental
chemicals.
Currently, knowledge of soil and dust ingestion
patterns within the United States is limited. Only a
few researchers have attempted to quantify soil and
dust ingestion patterns in U.S. adults or children.
This chapter explains the concepts of soil and
dust ingestion, soil pica, and geophagy; defines these
terms for the purpose of this handbook's exposure
factors; and presents available data from the
literature on the amount of soil and dust ingested.
The Centers for Disease Control and
Prevention's Agency for Toxic Substances and
Disease Registry (ATSDR) held a workshop in June
2000 in which a panel of soil ingestion experts
developed definitions for soil ingestion, soil pica,
and geophagy to distinguish aspects of soil ingestion
patterns that are important from a research
perspective (ATSDR, 2001). This chapter uses the
definitions developed by participants in that
workshop:
Soil ingestion is the consumption of soil. This may
result from various behaviors including, but not
limited to, mouthing, contacting dirty hands,
eating dropped food, or consuming soil directly.
Soil pica is the recurrent ingestion of unusually high
amounts of soil (i.e., on the order of
1,000-5,000 mg/day or more).
Geophagy is the intentional ingestion of earths and
is usually associated with cultural practices.
Some studies are of a behavior known as "pica,"
and the subset of "pica" that consists of ingesting
soil. A general definition of the concept of pica is
that of ingesting nonfood substances, or ingesting
large quantities of certain particular foods.
Definitions of pica often include references to
recurring or repeated ingestion of these substances.
Soil pica is specific to ingesting materials that are
defined as soil, such as clays, yard soil, and flower-
pot soil. Although soil pica has been observed
among children and adults, information about the
prevalence of pica behavior is limited. Gavrelis et al.
(2011) reported that the prevalence of nonfood
substance consumption varies by age, race, and
income level. The behavior was most prevalent
among children 1 to <3 years (Gavrelis et al., 2011).
Geophagy, on the other hand, is an extremely rare
behavior, especially among children, as is soil pica
among adults. One distinction between geophagy
and soil pica that may have public health
implications is the fact that surface soils generally
are not the main source of geophagy materials.
Instead, geophagy is typically the consumption of
clay from known, uncontaminated sources, whereas
soil pica involves the consumption of surface soils,
usually the top 2-3 inches (ATSDR, 2001).
Researchers in many different disciplines have
hypothesized motivations for human soil pica or
geophagy behavior, including alleviating hunger,
nutritional deficiencies, or gastrointestinal distress
(Young, 2010), a desire to remove toxins or
self-medicate (Starks and Slabach, 2012), and other
physiological or cultural influences (Danford, 1982).
Bruhn and Pangborn (1971) and Harris and Harper
(1997) suggest a religious context for certain
geophagy or soil ingestion practices. Geophagy is
characterized as an intentional behavior, whereas
soil pica should not be limited to intentional soil
ingestion, primarily because children can consume
large amounts of soil from their typical behaviors
September 2017
Page 5-1

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
and because differentiating intentional and
unintentional behavior in young children is difficult
(ATSDR, 2001). Some researchers have investigated
populations that may be more likely than others to
exhibit soil pica or geophagy behavior on a recurring
basis. These populations might include pregnant
women who exhibit soil pica behavior (Simpson et
al., 2000), adults and children who practice
geophagy (Vermeer and Frate, 1979),
institutionalized children (Wong, 1988), and
children with developmental delays (Danford,
1983), autism (Kinnell, 1985), or celiac disease
(Korman, 1990). However, identifying specific soil
pica and geophagy populations remains difficult due
to limited research on this topic. ATSDR (2001) has
estimated that 33% of children ingest more than 10
grams of soil 1 or 2 days a year. No information was
located regarding the prevalence of geophagy
behavior.
Because some soil and dust ingestion may be a
result of hand-to-mouth behavior, soil properties that
relate to adherence to the skin may be important. For
example, soil particle size, organic matter content,
moisture content, and other soil properties may
affect the amount of soil that adheres to the skin and
is available for ingestion. Soil particle sizes range
from 50-2,000 (am for sand, 2-50 jam for silt, and
are <2 (am for clay (USDA, 1999), while typical
atmospheric dust particle sizes are in the range of
0.001-30 (am (U.S. OSHA, 1987). Studies on
particle size have indicated that finer soil particles
(generally <63 (am in diameter) tend to be adhered
more efficiently to human hands, whereas adhered
soil fractions are independent of organic matter
content or soil origin (Choate et al., 2006; Yamamoto
et al., 2006). For soils with higher moisture content,
a greater number of large particles have been shown
to adhere to the skin (Choate et al., 2006). Ikegami
et al. (2014) found that approximately 90% of the
particles of playground soil that adhered to
children's hands were less than 100 (am in size.
Beamer et al. (2012) and Bergstrom et al. (2011)
found that concentrations of contaminants (e.g.,
metals) in soil may differ according to particle size.
Cao et al. (2012) also described the importance of
considering particle size when evaluating exposures
to indoor settled dust.
In this handbook, soil, indoor settled dust, and
outdoor settled dust are defined generally as the
following:
Soil. Particles of unconsolidated mineral and/or
organic matter from the earth's surface that are
located outdoors, or are used indoors to support
plant growth. It includes particles that have
settled onto outdoor objects and surfaces
(outdoor settled dust).
Indoor Settled Dust. Particles in building interiors
that have settled onto objects, surfaces, floors,
and carpeting. These particles may include soil
particles that have been tracked or blown into the
indoor environment from outdoors, as well as
organic matter.
Outdoor Settled Dust. Particles that have settled
onto outdoor objects and surfaces due to either
wet or dry deposition. Note that it may not be
possible to distinguish between soil and outdoor
settled dust because outdoor settled dust
generally is present on the uppermost surface
layer of soil.
For the purposes of this handbook, soil ingestion
includes both soil and outdoor settled dust, and dust
ingestion includes indoor settled dust only.
Several methodologies related to soil and dust
ingestion are represented in the literature. Two
methodologies combine biomarker measurements
with measurements of the biomarker substance's
presence in environmental media. An additional
methodology offers modeled estimates of soil/dust
ingestion from activity pattern data from
observational studies (e.g., videography) or from the
responses to survey questionnaires about children's
activities, behaviors, and locations.
The first of the biomarker methodologies is the
"tracer element" methodology. This method uses
measured quantities of specific elements present in
feces, urine, food and medications, yard soil, house
dust, and sometimes community soil and dust. This
information is used in combination with certain
assumptions about the elements' behavior in the
gastrointestinal tract to produce estimates of soil and
dust quantities ingested (Davis et al., 1990).
The second biomarker methodology is the
"biokinetic model comparison " methodology. This
method compares results from a biokinetic model of
lead exposure and uptake that predicts blood lead
levels, with biomarker measurements of lead in
blood (Von Lindern et al., 2003). The model
predictions are made using assumptions about
ingested soil and dust quantities that are based, in
part, on results from early versions of the first
methodology. Therefore, the comparison with actual
measured blood lead levels serves to confirm, to
some extent, the assumptions about ingested soil and
dust quantities used in the biokinetic model. Lead
isotope ratios have also been used as a biomarker to
September 2017
Page 5-2

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
study sources of lead exposures in children. This
technique involves measurements of different lead
isotopes in blood and/or urine, food, water, and
house dust and compares the ratio of different lead
isotopes to infer sources of lead exposure that may
include dust or other environmental exposures
(Manton et al., 2000). However, application of lead
isotope ratios to derive estimates of dust ingestion by
children has not been attempted. Therefore, it is not
discussed any further in this chapter.
The third, "activity pattern" methodology,
combines information from hand-to-mouth and
object-to-mouth behaviors with microenvironment
data (i.e., time spent at different locations) to derive
estimates of soil and dust ingestion. Behavioral
information often comes from data obtained using
videography techniques or from responses to survey
questions obtained from adults, caregivers, and/or
children. Surveys often include questions about
hand-to-mouth and object-to-mouth behaviors, soil
and dust ingestion behaviors, frequency, and
sometimes quantity (Barltrop, 1966). Moya and
Phillips (2014) provide a review of these three
methodologies used to estimate soil and dust
ingestion.
A fourth methodology uses assumptions
regarding ingested quantities of soil and dust that are
based on a general knowledge of human behavior,
and potentially supplemented or informed by data
from other methodologies (Hawley, 1985; Kissel et
al., 1998a; Wong et al., 2000). This methodology is
not discussed in this chapter because it yields
rudimentary estimates of soil ingestion.
Another approach used to estimate soil/dust
ingestion consists of measurements of soil/dust
loadings on surfaces (mass per surface area) and
concentrations of contaminants on those surfaces.
Estimates of soil/dust ingestion can be made by
making assumptions about children's hand-to-mouth
and object-to-mouth behavior, surfaces contacted,
fraction of soil/dust transferred, exposure time,
saliva extraction efficiency, and assumptions about
the amount of soil/dust reloading on skin or surfaces.
This approach results in a different metric of
soil/dust ingestion in units of area contacted/day,
which can then be used with corresponding
contaminant concentrations of soil/dust per surface
area. This approach is described in more detail by
Wilson etal. (2016).
The recommendations for soil, dust, and
soil + dust ingestion rates are provided in the next
section, along with a summary of the confidence
ratings for these recommendations. The
recommended values are based on key studies
identified by the U.S. Environmental Protection
Agency (U.S. EPA) for this factor. As described in
Chapter 1 of the Exposure Factors Handbook: 2011
Edition (U.S. EPA, 2011), key studies represent the
most up-to-date and scientifically sound for deriving
recommendations for exposure factors, whereas
other studies are designated "relevant," meaning
applicable or pertinent, but not necessarily the most
important. For example, studies that provide
supporting data or information related to the factor
of interest (e.g., pica prevalence), or have study
designs or approaches that make the data less
applicable to the population of interest (e.g., studies
not conducted in the United States) have been
designated as relevant rather than key. Key studies
were selected based on the general assessment
factors described in Chapter 1 of the Handbook.
Following the recommendations, a description of
the three methodologies used to estimate soil and
dust ingestion is provided, followed by a summary
of key and relevant studies. Because strengths and
limitations of each one of the key and relevant
studies relate to the strengths and limitations
inherent of the methodologies themselves, they are
discussed at the end of the key and relevant studies.
5.2. RECOMMENDATIONS
Table 5-1 provides the recommended soil and
dust ingestion rates for use in human health risk
assessments. The soil ingestion recommendations in
Table 5-1 are intended to represent ingestion of a
combination of soil and outdoor settled dust, without
distinguishing between these sources. The source of
the soil in these recommendations could be outdoor
soil, indoor containerized soil used to support
growth of indoor plants, or a combination of both
outdoor soil and containerized indoor soil. The
inhalation and subsequent swallowing of soil
particles is accounted for in these recommended
values; therefore, this pathway does not need to be
considered separately. These recommendations are
called "soil." The dust ingestion recommendations in
Table 5-1 include soil tracked into the indoor setting,
indoor settled dust, and air-suspended particulate
matter that is inhaled and swallowed. "Dust"
recommendations are provided in the event that
assessors need recommendations for an indoor or
inside a transportation vehicle scenario in which
dust, but not outdoor soil, is the exposure medium of
concern. The soil + dust recommendations would
include soil, either from outdoor or containerized
indoor sources, dust that is a combination of outdoor
settled dust, indoor settled dust, and air-suspended
particulate matter that is inhaled, subsequently
trapped in mucous and moved from the respiratory
September 2017
Page 5-3

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
system to the gastrointestinal tract, and a soil-origin
material located on indoor floor surfaces that was
tracked indoors by building occupants.
Many of the key studies predated the age groups
recommended for children by U.S. EPA (2005) and
were performed on groups of children of varying
ages. As a result, central tendency and upper
percentile recommendations could only be
developed for some combined age categories, as
shown in Table 5-1. Published estimates from the
key studies have been rounded to one significant
figure in Table 5-1.
An important factor to consider when using the
recommended values described in the following
sections is that they are limited to estimates of soil
and dust quantities ingested. The scope of this
chapter is limited to quantities of soil and dust taken
into the gastrointestinal tract, and does not extend to
issues regarding bioavailability of environmental
contaminants present in that soil and dust.
Information from other sources is needed to address
bioavailability. In addition, as more information
becomes available regarding gastrointestinal
absorption of environmental contaminants,
adjustments to the soil and dust ingestion exposure
equations may need to be made to better represent
the direction of movement of those contaminants
within the gastrointestinal tract.
To place the recommended values into context, it
may be useful to compare the soil ingestion rates to
common measurements. For example, the central
tendency recommendation of 40 mg/day or
0.040 g/day of either soil only or dust only for
general population children 1 to <6 years old would
be equivalent to approximately 1/8 of an aspirin
tablet per day because the average aspirin tablet is
approximately 325 mg. Likewise, the central
tendency recommendation of 80 mg/day or 0.080
g/day, for soil and dust combined, would be
equivalent to approximately 1/4 of an aspirin tablet.
The 50 g/day ingestion rate recommended to
represent geophagy behavior would be roughly
equivalent to 150 aspirin tablets per day.
5.2.1 General Population Soil and Dust Ingestion
Rates
The key studies described in Sections 5.3.2 and
5.3.3 were used to recommend values for soil and
dust ingestion for adults and children in the general
population. Table 5-1 shows the central tendency and
upper percentile recommendations for daily
ingestion of soil + dust, soil only, and dust only in
mg/day. Section 5.5 and Table 5-34 provide
additional details on the derivation of these
recommended values. The recommended values for
soil ingestion only and dust ingestion only are based
on the assumption that 45% of the soil + dust
ingestion can be attributed to soil and 55% can be
attributed to dust. This assumption is based on the
defaults used in EPA's Integrated Exposure and
Uptake Biokinetic (IEUBK) model (U.S. EPA,
1994a). According to U.S. EPA (1994a), the
assumption is based on the relative likelihood of
contact with soil/dust in indoor and outdoor
locations and "represents [EPA's] best judgement of
a properly weighted ratio for this purpose." All
recommended values have been rounded to one
significant figure due to data limitations.
In general, the "central tendency"
recommendations reflect an arithmetic mean
(average) of estimated values within a study, across
studies within a methodology, and across the three
methodologies. However, in some of the tracer
studies, (Stanek and Calabrese, 1995a), the central
tendency value used was the average of the median
values for the four best tracer elements or the
average of the median of three tracers (see Section
5.3.3.2). For others (Calabrese etal., 1997a; 1997b)
the central tendency value represents the average of
the best tracer or the average based on aluminum and
silicon. Upper percentile recommendations for daily
ingestion are also provided in mg/day. Note that
there is considerably more uncertainty related to the
upper percentile soil and dust ingestion rate
estimates than for the average estimates. Biases due
to the errors (e.g., sampling errors, measurement
errors, analytical errors) are more likely to affect the
upper percentile estimates than the average
estimates. Upper percentile recommendations for the
general population are provided for soil, dust, and
soil + dust ingestion. These values are based on the
95th percentile values from the key studies.
The recommended central tendency soil + dust
ingestion estimate for general population infants
0 to <6 months old is 40 mg/day, and the central
tendency estimate for 6 months to <1 year of age is
70 mg/day. If a central tendency estimate is needed
for soil or dust only, the recommended values are
both 20 mg/day for infants 0 to <6 months (i.e., 18
mg/day soil and 22 mg/day dust, both rounded to one
significant figure is 20 mg/day). For infants 6
months to <1 year, the recommended soil only
estimate is 30 mg/day and the dust only estimate is
40 mg/day.
For risk assessment involving children 1 to <2
and 2 to <6 years of age, the recommended central
tendency soil + dust ingestion rates are 90 mg/day
and 60 mg/day, respectively. For soil only, the
recommended central tendency values are
40 mg/day for 1- to <2-year-olds and 30 mg/day for
September 2017
Page 5-4

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
2- to <6-year-olds, respectively. For dust only, the
recommended central tendency values are
50 mg/day and 30 mg/day for the two age groups,
respectively.
When assessing risks for children 1 to <6 years
of age who are not expected to exhibit soil pica or
geophagy behavior, the recommended central
tendency soil + dust ingestion estimate is 80 mg/day.
If an estimate for soil only or indoor dust only is
needed, the recommendation is 40 mg/day.
For children 6 to <12 years old without pica, the
recommended central tendency soil + dust ingestion
rate is 60 mg/day. For either soil or dust alone, the
estimate is 30 mg/day. For children 12 to <20 years
and adults, the recommended central tendency
values for use in risk assessment is 30 mg/day,
rounded to one significant figure. This
recommendation is based on data for adults from
Davis and Mirick (2006) and data from Wilson et al.
(2013) that indicated that central tendency soil + dust
ingestion rates ages 12 to <20 years and adults are
similar. For soil only, the recommended value is
10 mg/day, and for dust only, the recommended
value is 20 mg/day.
The upper percentile recommendations for
soil + dust ingestion among the general population,
are based on the 95th percentile values obtained from
key studies as shown in Table 5-1, rounded to one
significant figure. The recommended values: are
100 mg/day for infants 0 to <6 months (50 mg/day
soil and 60 mg/day dust), and 200 mg/day for
children 6 months to <1 year, 1 to <2 years, 2 to
<6 years, 1 to <6 years, and 6 to <12 years
(90 mg/day soil and 100 mg/day dust). For ages
12 years through adults, the recommended upper
percentile value is 100 mg/day (50 mg/day soil and
60 mg/day dust).
5.2.2. Soil Pica and Geophagy
Ingestion rates for "soil pica," and for individuals
who exhibit "geophagy" are also provided in Table
5-1. The soil-pica and geophagy recommendations
are likely to represent an acute high soil ingestion
episode or behavior. The soil pica ingestion estimate
in the literature for children up to age 6 years ranges
from 1,000 to 6,000 mg/day, averaged over the study
period (ATSDR, 2001; Barnes, 1990; Calabrese et
al., 1989, 1991, 1997b; Stanek et al., 1998). Due to
the short-term nature of these studies and the limited
amount of data available for children exhibiting pica
behavior, the lower end of this range of 1,000
mg/day is recommended for soil pica for children
1 to <6 years old. However, it is important to note
that soil ingestion for these children exhibiting soil
pica behavior has been reported as high as 20 to
25 g/day on any given day (Calabrese et al., 1997b).
Note too that the recommended soil pica value may
be more appropriate for acute exposures. Currently,
no data are available for soil pica behavior for
children less than 12 months or in children ages
6 to <21 years. Because pica behavior may occur
among some children ages ~1 to 21 years old
(Hyman et al., 1990), it is prudent to assume that, for
some children, soil pica behavior may occur at any
age up to 21 years. While pica may also occur among
adults, no key studies were available for developing
recommended intake rates for adults who exhibit
pica.
The recommended geophagy soil estimate is
50,000 mg/day (50 grams) for both adults and
children (Vermeer and Frate, 1979). It is important
to note that this value may be more representative of
acute exposures. Risk assessors should use this value
for soil ingestion in for individuals or populations
known to exhibit geophagy behaviors.
5.2.3.	Wilderness or Traditional Rural Lifestyles
Information on soil ingestion among special
populations, such as those engaging in wilderness
lifestyles in Canada, are presented as relevant studies
in Sections 5.3.4 (Doyle et al., 2012; Irvine et al.,
2014). Data from these studies may be appropriate
for high soil contact scenarios. For rural populations
following traditional rural or wilderness lifestyles as
described in these studies, adult soil ingestion rates
may be somewhat higher than those of the general
population. Based on these two studies the adult
mean soil + dust ingestion rate is 50 mg/day and the
upper percentile soil + dust ingestion rate is
200 mg/day. Based on personal communication with
the authors of these two studies, the 95th percentile
of the combined data sets was calculated to be 239
mg/day for aluminum and silicon (for all four tracers
the value would be 243 mg/day) (personal
communication between M. Stifelman, EPA, and J.
Doyle, University of Ottawa, Canada). Rounding to
one significant figure, the upper percentile value
would also be 200 mg/day. Assuming that soil
represents 45% and dust represents 55% of the
soil + dust value, the mean and upper percentile soil
only values would be 20 mg/day and 90 mg/day,
respectively. The mean and upper percentile dust
only values would be 30 mg/day and 100 mg/day,
respectively.
5.2.4.	Confidence Ratings
Section 5.4 gives a detailed explanation of the
limitations of the various study methodologies,
which are reflected in the confidence ratings for the
recommendations shown in Table 5-2. Individual
September 2017
Page 5-5

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
evaluations of the quality of studies are provided in
the specific discussions for each of the individual
studies. The confidence ratings are low due to the
relatively limited data on which the
recommendations are based and the limitations and
uncertainties associated with tracer studies described
in Section 5.4.1, and the assumptions needed to
develop quantitative estimates using the biokinetic
modeling (see Section 5.4.2) and activity pattern
modeling approaches (see Section 5.4.3). Other
uncertainties pertain to the representativeness of the
populations studied. A more detailed discussion
about the general assessment factors used to evaluate
the confidence in the recommendations is provided
in Chapter 1 of the Handbook. For the estimates of
soil only and dust only, an additional uncertainty
pertains to the assumption that the proportion of
soil + dust represented by soil only (45%) and dust
only (55%) is the same for all age groups.
September 2017
Page 5-6

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
Table 5-1. Recommended Values for Daily Soil, Dust, and Soil + Dust Ingestion (mg/day)a
Age Group
Soil + Dust
General	General
Population	Population
Central	Upper
Tendency"1	Percentile6
Soilb
General	General
Population Population
Central	Upper Soil
Tendencyf	Percentilef Pica8 Geophagy11
50
—
—
20
60
90
—
—
40
100
90
1,000
50,000
50
100
90
1,000
50,000
30
100
90
1,000
50,000
40
100
90
1,000
50,000
30
100
50
—
50,000
20
60
Dustc
General	General
Population Population
Central	Upper
Tendencyf	Percentilef
<6 months
6 months to <1
year
1	to <2 years
2	to <6 years
1 to <6 years
6 to <12 years
12 years
through adult
40
70 (60-80)
90
60
80 (60-100)
60 (60-60j
30 (4-50)i
100
200
200
200
200
200
100)
20
30
40
30
40
30
10
Ranges are provided in parentheses, when applicable, and represent the range of means from the various studies. Ranges
are not provided for age groups for which the recommendations are based on a single study.
Includes soil and outdoor settled dust.
Includes indoor settled dust only.
Based on the average of the central tendency values from the various studies for each of the three methodologies (tracer,
biokinetic modeling, activity pattern), averaged over the three methods. Recommendation for <6 months of age based on
Wilson et al. (2013) (note that data for 0 to <7 months in Wilson et al. [2013] were used to represent the 0 to <6 months
age group). Recommendations for children 6 months to <1 year based on the average of values from Hogan et al. (1998)
and von Lindern et al. (2016). Recommendations for 1- to 2 year-olds and 2- to <6-year-olds based on von Lindern et al.
(2016). Recommendations for children ages 1 to <6 years based on the average of values from Calabrese et al. (1989)
as reanalyzed in Stanek and Calabrese 1995a (mean of the median values for the best 4 tracers for each child); Calabrese
et al. (1997a) (average of the best tracer for each child); Calabrese et al. (1997b) (average of aluminum and silicon);
Davis et al. (1990) as reanalyzed by Stanek and Calabrese, 1995a (mean of the median values for 3 tracers for each child);
Hogan etal. (1998); Ozkaynak et al. (2011); von Lindern etal. (2016); and Wilson et al. (2013). The recommendations for
ages 12 years to adults are based on the average of data for teens (ages 12 to <20 years), adults, and seniors from Wilson
et al. (2013) and on adults from Davis and Mirick (2006). All recommended values were rounded to one significant figure.
See Table 5-34 for additional details.
Based on the average of the 95th percentile values from the various studies for each of the three methodologies (tracer,
biokinetic modeling, activity pattern), averaged over the three methods. Based on the 95th percentile values for the
same studies as used for the central tendency estimates except for age 12 years through adults. Upper percentile
recommendation for 12 years of age through adults based on the assumption that the ratio of the 95th percentile to the
mean value for adults is the same as the average of the ratios of 95th percentiles to means for all other age groups (i.e.,
average ratio of the 95th percentile to mean recommendations = 3.2). See Table 5-34 for additional details.
Estimates of soil and dust were derived from the soil + dust values assuming 45% soil and 55% dust, rounded to one
significant figure.
Professional judgement based on: ATSDR (2001); Barnes (1990); Calabrese et al. (1997b, 1991, 1989); Stanek et al.
(1998).
Vermeer and Frate (1979).
Range based on two studies with estimates of 55 and 56 mg/day; both of these estimates round to 60 mg/day.
Soil + dust ingestion rates may be higher for adults following a traditional rural or wilderness lifestyle. Based on Doyle
et al. (2012) and Irvine et al. (2014) the central tendency adult soil + dust ingestion rates is 50 mg/day (20 mg/day soil and
30 mg/day dust) and the upper percentile rate is 200 mg/day (90 mg/day soil and 100 mg/day dust).
= No data.
September 2017
Page 5-7

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
Table 5-2. Confidence in Recommendations for Ingestion of Soil and Dust"
General Assessment
Factors
Rationale
Rating
Soundness
Adequacy of
Approach
Minimal (or defined)
Bias
Applicability and
Utility
Exposure Factor of
Interest
Representativeness
Currency
Data Collection
Period
Low
The methodologies have serious limitations. No single study captured all of the information
needed (quantities ingested, frequency of high soil ingestion episodes, prevalence of high
soil ingestion). Sample selection may have introduced some bias in the results (i.e.,
children near smelter or Superfund sites, volunteers in nursery schools). The total number
of children in key studies were 241 (tracer studies; Calabrese et al., 1989; Davis et al., 1990;
Calabrese et al., 1997a,b) and 2,599 (biokinetic modeling; Hogan et al., 1998; von Lindern
et al., 2016). Modeled estimates were based on 1,000 simulated individuals (Ozkaynak et
al., 2011) or 200,000 trials (Wilson et al., 2013). Models may be sensitive to assumptions
and the quality and availability of input variables. The response rates for in-person
interviews and telephone surveys were often not stated in published articles. Only two key
studies provided data for adults.
Numerous sources of measurement error exist in the tracer element studies and the
biokinetic model comparison studies. Some input variables for the modeled estimates are
uncertain. Some of the assumptions used in the modeling studies may underestimate soil
ingestion rates. Knowledge of soil and dust contamination may have affected the results of
some of the studies.
Low
The key tracer studies focused on the soil exposure factor, with little or no focus on the dust
exposure factor. The biokinetic model comparison studies accounted for both soil and dust
ingestion, but also addressed other factors (e.g., exposure via dietary intake, inhalation).
The activity pattern studies focused on soil and dust ingestion.
The study samples may not be representative of the United States in terms of race, ethnicity,
socioeconomics, and geographical location; studies focused on specific areas. One key
study was from Canada (Wilson et al., 2013), but some of the assumptions were derived
from U.S. populations.
Most of the tracer element studies were conducted in the 1980s and 1990s; activity pattern
modeling studies are more recent; biokinetic modeling studies have more recent publication
dates, but were generally based on older data.
Tracer element studies' data collection periods may not represent long-term behaviors.
Biokinetic model comparison and survey response studies represent longer term behaviors.
Data used in modeled simulation estimates may not represent long-term behaviors.
September 2017
Page 5-8

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
Table 5-2. Confidence in Recommendations for Ingestion of Soil and Dust (Continued)
General Assessment
Factors
Rationale
Rating
Clarity and
Completeness
Accessibility
All key studies are available from the peer-reviewed literature.
Medium
Reproducibility
For methodologies used by more than one research group, reproducible results were
obtained in some instances.

Quality Assurance
For some studies, information on quality assurance/quality control was limited or absent.

Variability and
Uncertainty
Variability in
Population
Tracer element and activity pattern methodology studies characterized variability among
study sample members; the IEUBK model used in the biokinetic approach uses average soil
ingestion rates. Day-to-day and seasonal variability was not very well characterized.
Numerous factors that may influence variability have not been explored in detail.
Low
Minimal
Uncertainty
Estimates are highly uncertain. Tracer element study designs appear to introduce biases in
the results. Modeled estimates may be sensitive to input variables.

Evaluation and
Review
Peer Review
All key studies appeared in peer-reviewed journals.
Medium
Number and
Agreement of
Studies
14 key studies, but some key studies are reanalyses of previously published data.
Researchers using similar methodologies obtained generally similar results. While there is
general agreement between researchers using different methodologies, estimates based on
the activity pattern methodology generally yield somewhat lower estimates than both the
tracer and biokinetic modeling approaches.

Overall Rating

Low
a See Section 1.5.2 in Chapter 1 of the Exposure Factors Handbook: 2011 Edition for a detailed
description of the evaluation criteria used in this table.
September 2017
Page 5-9

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
5.3. KEY AND RELEVANT STUDIES
The key tracer element, biokinetic model
comparison, and survey response studies are
summarized in the following sections. Certain
studies were considered "key" and were used as a
basis for developing the recommendations, using
judgment about the study's design features,
applicability, and utility of the data to U.S. soil and
dust ingestion rates, clarity and completeness, and
characterization of uncertainty and variability in
ingestion estimates. Because the studies often were
performed for reasons unrelated to developing
long-term soil and dust ingestion recommendations,
their attributes that were characterized as
"limitations" in this chapter might not be limitations
when viewed in the context of the study's original
purpose. However, when studies are used for
developing a soil or dust ingestion recommendation,
EPA has categorized some studies' design or
implementation as preferable to others. In general,
EPA chose studies designed either with a census or
randomized sample approach over studies that used
a convenience sample, or other nonrandomized
approach, as well as studies that more clearly
explained various factors in the study's
implementation that affect interpretation of the
results. However, in some cases, studies that used a
nonrandomized design contain information that is
useful for developing exposure factor
recommendations (e.g., if they are the only studies
of children in a particular age category), and thus
may have been designated as "key" studies. Other
studies were considered "relevant" but not "key"
because they provide useful information for
evaluating the reasonableness of the data in the key
studies or provide supporting information, but in
EPA's judgment they did not meet the same level of
soundness, applicability and utility, clarity and
completeness, and characterization of uncertainty
and variability that the key studies did or they may
not be representative of the U.S. general population.
In addition, studies that did not contain information
that can be used to develop a specific
recommendation for mg/day soil and dust ingestion
were classified as relevant rather than key. However,
some studies classified as "relevant" may be used as
the basis for recommendations for particular
exposure settings (e.g., Doyle et al., 2012, Irvine et
al., 2014 for populations engaging in rural or
wilderness lifestyles).
Some studies are reanalyses of previously
published data. For this reason, the sections that
follow are organized into key and relevant studies of
primary analysis (i.e., studies in which researchers
have developed primary data pertaining to soil and
dust ingestion) and key and relevant studies of
secondary analysis (i.e., studies in which researchers
have interpreted previously published results, or data
that were originally collected for a different
purpose).
The three methodologies described in this
chapter to derive soil and dust ingestion rates have
limitations. Because some of these limitations apply
equally to all the studies within each methodology,
they are discussed in more detail in Section 5.4
separately from the study summaries. Additional
limitations specific to each study are described
within each study summary. The discussion of
limitations does not imply that the studies were
conducted inappropriately, rather they are
limitations inherent in these methodologies.
5.3.1. Methodologies Used in Key
Studies
5.3.1.1. Tracer Element Methodology
The tracer element methodology attempts to
quantify the amounts of soil ingested by analyzing
samples of soil and dust from residences and/or
children's play areas, and feces or urine. The soil,
dust, fecal, and urine samples are analyzed for the
presence and quantity of tracer elements—typically,
aluminum, silicon, titanium, and other elements. A
key underlying assumption is that these elements are
not metabolized into other substances in the body or
absorbed from the gastrointestinal tract in significant
quantities, and thus their presence in feces and urine
can be used to estimate the quantity of soil ingested
by mouth. Although they are sometimes called mass
balance studies, none of the studies attempt to
quantify amounts excreted in perspiration, tears,
glandular secretions, or shed skin, hair, or finger- and
toenails, nor do they account for tracer element
exposure via the dermal or inhalation routes, and
thus they are not a complete "mass balance"
methodology. Early studies using this methodology
did not always account for the contribution of tracer
elements from nonsoil substances (food,
medications, and nonfood sources such as
toothpaste) that might be swallowed. U.S. studies
using this methodology in or after the mid to late
1980s account for, or attempt to account for, tracer
element contributions from these nonsoil sources.
Some study authors adjust their soil ingestion
estimate results to account for the potential
contribution of tracer elements found in household
dust as well as soil.
Empirical estimates of soil ingestion rates in
children have been made by back-calculating the
September 2017
Page 5-10

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
mass of soil and/or dust a subject would need to
ingest to achieve a tracer element mass measured in
collected excreta (i.e., feces and urine). The
following is a general expression for the trace
element ("tracer") mass balance:
Mingested in soil ^feces+urine ^non—soil or dust
(Eqn. 5-1)
where:
Mingestedinsoii = mass of tracer in soil or dust that is
ingested (mg)
Mfeces + urine = mass of tracer measured in feces and
urine (mg)
- ^ 1 non:-,oil or dust = mass of tracer measured in nonsoil
or dust (e.g., food, water, medicine,
toothpaste) (mg)
Dividing the mass of tracer in soil or dust that is
ingested by the measured tracer concentration in soil
(mg/g) yields an estimate of the mass of soil
ingested, 5" (g):
r Mingested in soil
°soil or dust
where:
S	= mass of soil or dust ingested (g)
Csoil or dust	= concentration of tracer in soil
or dust (mg/g)
(Eqn. 5-2)
The U.S. tracer element researchers have all
assumed a certain offset, or lag time between
ingestion of food, medication, and soil, and the
resulting fecal and urinary output. The lag times used
are typically 24 or 28 hours (Davis and Mirick,
2006; Stanek et al., 2001a; Stanek and Calabrese,
1995b); thus, these researchers subtract the previous
day's food and medication tracer element quantity
ingested from the current day's fecal and urinary
tracer element quantity that was excreted. When
compositing food, medication, fecal, and urine
samples across the entire study period, daily
estimates can be obtained by dividing the total
estimated soil ingestion by the number of days in
which fecal and/or urine samples were collected. A
variation of the algorithm that provides slightly
higher estimates of soil ingestion is to divide the total
estimated soil ingestion by the number of days on
which feces were produced, which by definition
would be equal to or less than the total number of
days of the study period's fecal sample collection.
Substituting tracer element dust concentrations
for tracer element soil concentrations yields a dust
ingestion estimate. Because the actual nonfood,
nonmedication quantity ingested is a combination of
soil and dust, the unknown true soil and dust
ingestion is likely to be somewhere between the
estimates based on soil concentrations and those
based on dust concentrations. Tracer element
researchers have described ingestion estimates for
soil that actually represent a combination of soil and
dust, but were calculated based on tracer element
concentrations in soil. Similarly, they have described
ingestion estimates for dust that are actually for a
combination of soil and dust, but were calculated
based on tracer element concentrations in dust. Other
variations on these general soil and dust ingestion
algorithms have been published in attempts to
account for time spent indoors, time spent away from
the house, and other factors that might influence the
relative proportion of soil versus dust.
Each individual's soil and dust ingestion can be
represented as an unknown constant in a set of
simultaneous equations of soil or dust ingestion
represented by different tracer elements. To date,
only two of the U.S. research teams (Lasztity et al.,
1989; Barnes, 1990) have published estimates
calculated for pairs of tracer elements using
simultaneous equations.
The U.S. tracer element studies have been
performed for only short-duration study periods, and
only for 33 adults (Davis and Mirick, 2006) and
241 children (101 inDavisetal. [1990], 12 of whom
were studied again in Davis and Mirick [2006]; 64 in
Calabrese et al. [1989] and Barnes [1990]; 64 in
Calabrese et al. [1997a]; and 12 in Calabrese et al.
[1997b]). The studies provide information on
quantities of soil and dust ingested for the studied
groups for short time periods, but provide limited
information on overall prevalence of soil ingestion
by U.S. adults and children, and limited information
on the frequency of higher soil ingestion episodes.
While there are advantages to using the tracer
method (e.g., estimates are provided based on
empirical data vs. modeling; direct measurements),
there are also sources of uncertainty associated with
this method. For example, error sources sometime
cause individual soil or dust ingestion estimates for
some tracers to be negative, and in some studies, this
resulted in median or mean "mass balance" soil
ingestion estimates that were also negative for some
tracers. Authors of these studies have averaged both
September 2017
Page 5-11

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
negative and positive numbers together in their
estimation of soil ingestion rates. For soil and dust
ingestion estimates based on each particular tracer,
or averaged across tracers, the net impact of
competing upward and downward sources of error is
unclear. Other sources of error can influence the
estimates in an upward direction (e.g., not
accounting for all nonsoil/dust sources of the tracer
elements). A more detailed discussion of the
uncertainties and limitations associated with the
tracer method is provided in Section 5.4.1.
5.3.1.2. Biokinetic Model Comparison
Methodology
The Biokinetic Model Comparison methodology
compares direct measurements of a biomarker, such
as blood or urine levels of a toxicant, with
predictions from a biokinetic model of oral, dermal,
and inhalation exposure routes with air, food, water,
soil, and dust toxicant sources. An example is to
compare measured children's blood lead levels with
predictions from the IEUBK model. Where
environmental contamination of lead in soil, dust,
and drinking water has been measured and those
measurements can be used as model inputs for the
children in a specific community, the model's
assumed soil and dust ingestion values can be
evaluated by comparing the model's predictions of
blood lead levels with those children's measured
blood lead levels. It should be noted, however, that
such confirmation of the predicted blood lead levels
would be confirmation of the net impact of all model
inputs, and not just soil and dust ingestions. Under
the assumption that (actual) blood lead levels of
various groups of children studied were accurately
measured, and those measured blood lead levels are
consistent with biokinetic model predictions for
those groups of children, then the model's default
assumptions may correspond to the central tendency,
or typical, children in an assessed group of children.
Nevertheless, the model's default assumptions for
biokinetics and intake rates can be useful for
predicting outcomes for highly exposed children if
the higher exposure occurs as increased
concentrations in the relevant media, and if the
default population variability is relevant for the
group of children under consideration. Use of the
IEUBK in this way assumes that blood lead can be
used as a suitable biomarker for soil and dust
ingestion. An advantage of this method is that it can
be used to indirectly estimate long-term soil and dust
intake. A detailed discussion on the limitations and
uncertainties associated with this method is provided
in Section 5.4.2.
5.3.1.3. Activity Pattern Methodology
The activity pattern methodology combines
information on hand-to-mouth and object-to-mouth
activities (microactivities) and time spent at various
locations (microenvironments) with assumptions
about transfer parameters (e.g., soil-to-skin
adherence, saliva removal efficiency) and other
exposure factors (e.g., frequency of hand washing)
to derive estimates of soil and dust ingestion. This
methodology has been used in U.S. EPA's stochastic
human exposure and dose simulation (SHEDS)
model. The SHEDS model is a probabilistic model
that can simulate cumulative (multiple chemicals) or
aggregate (single chemical) residential exposures for
a population of interest over time via multiple routes
of exposure for different types of chemicals and
scenarios, including those involving soil ingestion
(U.S. EPA, 2010).
The activity pattern methodology includes
observational studies as well as surveys of adults,
children's caretakers, or children themselves, via
in-person or mailed questionnaires that ask about
mouthing behavior and ingestion of various nonfood
items and time spent in various microenvironments.
There are three general approaches to gather data on
children's mouthing behavior: real-time hand
recording, in which trained observers manually
record information (Davis et al., 1995);
video-transcription, in which trained videographers
tape a child's activities and subsequently extract the
pertinent data manually or with computer software
(Black et al., 2005); and questionnaire, or survey
response, techniques (Stanek et al., 1998).
An advantage of this method is that it does not
require collection of biologic samples. Also, soil and
dust ingestion can be estimated separately. One of
the limitations of this approach includes the
availability and quality of the input variables.
Ozkaynak et al. (2011) found that the model is most
sensitive to dust loadings on carpets and hard floor
surfaces, soil-to-skin adherence factors, hand
mouthing frequency, and hand washing frequency
(Ozkaynak et al., 2011). A detailed discussion of the
limitations and uncertainties associated with this
method is provided in Section 5.4.3.
5.3.2. Key Studies of Primary Analysis
The sections that follow provide summaries of
key studies in which researchers have developed
primary data pertaining to soil and dust ingestion.
September 2017
Page 5-12

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
5.3.2.1.	Vermeer and Frate (1979)—Geophagia
in Rural Mississippi: Environmental and
Cultural Contexts and Nutritional
Implications
Vermeer and Frate (1979) performed a survey
response study in Holmes County, MS in the 1970s
(date unspecified). Questions about geophagy
(defined as regular consumption of clay over a
period of weeks) were asked of household members
(N = 229 in 50 households; 56 were women, 33 were
men, and 140 were children or adolescents) of a
subset of a random sample of nutrition survey
respondents. Caregiver responses to questions about
115 children under 13 years old indicate that
geophagy was likely to be practiced by a minimum
of 18 (16%) of these children; however, 16 of these
18 children were 1 to 4 years old, and only 2 of the
18 were older than 4 years. Of the 56 women, 32
(57%) reported eating clay. There was no reported
geophagy among 33 men or 25 adolescent study
subjects questioned.
In a separately administered survey, geophagy
and pica data were obtained from 142 pregnant
women over a period of 10 months. Geophagy was
reported by 40 of these women (28%), and an
additional 27 respondents (19%) reported other pica
behavior, including the consumption of laundry
starch, dry powdered milk, and baking soda.
The average daily amount of clay consumed was
reported to be about 50 grams, for the adult and child
respondents who acknowledged practicing
geophagy. Quantities were usually described as
either portions or multiples of the amount that could
be held in a single, cupped hand. Clays for
consumption were generally obtained from the B soil
horizon, or subsoil rather than an uppermost layer, at
a depth of 50 to 130 cm.
5.3.2.2.	Calabrese et aL (1989)—How Much Soil
Do Young Children Ingest: An
Epidemiologic Study/Barnes
(1990)—Childhood	Soil Ingestion: How
Much Dirt Do Kids Eat?/Calabrese et aL
(1991)—Evidence	of Soil Pica Behavior
and Quantification of Soil Ingested
Calabrese et al. (1989) and Barnes (1990) studied
soil ingestion among children using eight tracer
elements—aluminum, barium, manganese, silicon,
titanium, vanadium, yttrium, and zirconium. A
nonrandom sample of 1-, 2-, and 3-year-olds
(30 males and 34 females) from the greater Amherst,
MA area was studied, presumably in 1987. The
children were predominantly from two-parent
households where the parents were highly educated.
The study was conducted over a period of 8 days
spread over 2 weeks. During each week, duplicate
samples of food, beverages, medicines, and vitamins
were collected on Monday through Wednesday,
while excreta, excluding wipes and toilet paper, were
collected for four 24-hour cycles running from
Monday/Tuesday through Thursday/Friday. Soil and
dust samples were also collected from the children's
homes and play areas. Study participants were
supplied with toothpaste, baby cornstarch, diaper
rash cream, and soap with low levels of most of the
tracer elements. Quality control of the analysis
yielded recoveries between 88.1% and 100.2% for
all tracers except zirconium, which had a low
recovery.
Table 5-3 shows the published mean soil
ingestion estimates ranging from -294 mg/day
based on manganese to 459 mg/day based on
vanadium, median soil ingestion estimates ranging
from -261 mg/day based on manganese to
96 mg/day based on vanadium, and 95th percentile
estimates ranged from 106 mg/day based on yttrium
to 1,903 mg/day based on vanadium. Maximum
daily soil ingestion estimates ranged from
1,391 mg/day based on zirconium to 7,281 mg/day
based on manganese. Dust ingestion estimates
calculated using tracer concentrations in dust were
often, but not always, higher than soil ingestions
calculated using tracer concentrations in soil.
Data for the uppermost 23 subject-weeks (the
highest soil ingestion estimates, averaged over the
4 days of excreta collection during each of the
2 weeks) were published in Calabrese et al. (1991).
One child's soil pica behavior was estimated in
Barnes (1990) using both the subtraction/division
algorithm and the simultaneous equations method.
On two particular days during the second week of
the study period, the child's aluminum-based soil
ingestion estimates were 19 g/day (18,700 mg/day)
and 36 g/day (35,600 mg/day), silicon-based soil
ingestion estimates were 20 g/day (20,000 mg/day)
and 24 g/day (24,000 mg/day), and simultaneous-
equation soil ingestion estimates were 20 g/day
(20,100 mg/day) and 23 g/day (23,100 mg/day)
(Barnes, 1990). By tracer, averaged across the entire
week, this child's estimates ranged from
approximately 10 to 14 g/day during the second
week of observation, excluding zirconium, which
presented limitations with the analytical protocol
(Calabrese et al., 1991, see Table 5-4), and averaged
6 g/day across the entire study period. Additional
information about this child's apparent ingestion of
soil versus dust during the study period was
published in Calabrese and Stanek (1992a).
September 2017
Page 5-13

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
5.3.2.3. Davis et aL (1990)—Quantitative
Estimates of Soil Ingestion in Normal
Children between the Ages of 2 and 7
Years: Population-Based Estimates Using
Aluminum, Silicon, and Titanium as Soil
Tracer Elements
Davis et al. (1990) used a tracer element
technique to estimate soil ingestion among children.
In this study, 104 children between the ages of 2 and
7 years were randomly selected from a three-city
area in southeastern Washington State. Soil and dust
ingestion was evaluated by analyzing soil and house
dust, feces, urine, and duplicate food, dietary
supplement, medication, and mouthwash samples
for aluminum, silicon, and titanium. Data were
collected for 101 of the 104 children during July,
August, or September 1987. In each family, data
were collected over a 7-day period, with 4 days of
excreta sample collection. Dried soil samples were
passed successively through a 20- and 60-mesh (850
and 250 |im. respectively; ASTM, 2017) stainless
steel sieve. Participants were supplied with
toothpaste with known tracer element content. In
addition, information on dietary habits and
demographics was collected to identify behavioral
and demographic characteristics that influence soil
ingestion rates among children. The amount of soil
ingested on a daily basis was estimated using
Equation 5-3:
„ — (((DWf +DWp)xEf-)+2Eu-)-(DWfd xEfd)
O j 0 	
i,e
Esoil

(Eqn. 5-3)
where:

SUe =
soil ingested for child /' based on

tracer e (grams)
DWf =
feces dry weight (grams)
DWP =
feces dry weight on toilet

paper (grams)
Ef =
tracer concentration in feces (|ig/g)
Eu =
tracer amount in urine (|ig)
DW/d =
food dry weight (grams)
Efd
tracer concentration in food (|ig/g)
Esoil
tracer concentration in soil (|ig/g)
The tracer amount in urine (/•.'„) was multiplied
by a factor of 2 to account for the fact that parents
were asked to collect half of the total daily urine
output. The soil ingestion rates were corrected by
adding the amount of tracer in vitamins and
medications to the amount of tracer in food and
adjusting the food, fecal, and urine sample weights
to account for missing samples. Food, fecal, and
urine samples were composited over a 4-day period,
and estimates for daily soil ingestion were obtained
by dividing the 4-day composited tracer quantities
by 4. Davis et al. (1990) reported that recoveries for
most analyses were within the quality control limits,
±20% for laboratory samples and ±25% for the
matrix spiked samples.
Soil ingestion rates were highly variable,
especially those based on titanium. Mean daily soil
ingestion estimates were 38.9 mg/day for aluminum,
82.4 mg/day for silicon, and 245.5 mg/day for
titanium (see Table 5-5). Median values were
25.3 mg/day for aluminum, 59.4 mg/day for silicon,
and 81.3 mg/day for titanium. The investigators also
evaluated the extent to which differences in tracer
concentrations in house dust and yard soil impacted
estimated soil ingestion rates. The value used in the
denominator of the soil ingestion estimate equation
was recalculated to represent a weighted average of
the tracer concentration in yard soil and house dust
based on the proportion of time the child spent
indoors and outdoors, using an assumption that the
likelihood of ingesting soil outdoors was the same as
that of ingesting dust indoors. The adjusted mean
soil/dust ingestion rates were 64.5 mg/day for
aluminum, 160.0 mg/day for silicon, and
268.4 mg/day for titanium. Adjusted median
soil/dust ingestion rates were 51.8 mg/day for
aluminum, 112.4 mg/day for silicon, and
116.6 mg/day for titanium. The authors also
investigated whether nine behavioral and
demographic factors could be used to predict soil
ingestion. They found family income less than
$15,000/year and swallowing toothpaste to be
predictors with silicon-based estimates, residing in
one of the three cities to be a significant predictor
with aluminum-based estimates, and washing the
face before eating significant for titanium-based
estimates.
5.3.2.4. Calabrese et aL (1997a)—Soil Ingestion
Estimates for Children Residing on a
Superfund Site
Calabrese et al. (1997a) estimated soil ingestion
rates for children residing on a Superfund site using
a methodology in which eight tracer elements were
analyzed. The methodology used in this study is
similar to that employed in Calabrese et al. (1989),
except that rather than using barium, manganese, and
vanadium as three of the eight tracers, the
researchers replaced them with cerium, lanthanum,
September 2017
Page 5-14

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
and neodymium. A total of 64 children ages 1 to 4
years (36 males, 28 females) were selected for this
study of the Anaconda, MT area. The study was
conducted for seven consecutive days during
September or September and October, apparently in
1992, shortly after soil was removed and replaced in
some residential yards in the area. Duplicate samples
of meals, beverages, and over-the-counter medicines
and vitamins were collected over the 7-day period,
along with fecal samples. In addition, soil and dust
samples were collected from the children's home and
play areas. Soil samples were sieved through a 2-mm
nylon mesh. A subsample was ground and sieved
using a 200-mesh (75 |im: ASTM, 2017) screen.
Dust samples were sieved to separate fine dust from
larger pieces. Toothpaste containing nondetectable
levels of the tracer elements, with the exception of
silica, was provided to all of the children. Infants
were provided with baby cornstarch, diaper rash
cream, and soap, which were found to contain low
levels of tracer elements.
Because of the high degree of intertracer
variability, Calabrese et al. (1997a) also derived
estimates based on the "best tracer methodology"
(BTM). This BTM uses food:soil (F:S) tracer
concentration ratios in order to correct for errors
caused by misalignment of tracer input and outputs,
ingestion of nonfood sources, and nonsoil sources
(Stanek and Calabrese, 1995a). A low F:S ratio is
desired because it minimizes transit time errors. The
BTM did not use the results from cerium, lanthanum,
and neodymium despite these tracers having low
F:S ratios because the soil concentrations for these
elements were found to be affected by particle size
and more susceptible to source errors. Calabrese et
al. (1997a) noted that estimates based on aluminum,
silicon, and yttrium in this study may result in lower
soil ingestion estimates than the true value because
the apparent residual negative errors found for these
three tracers for a large majority of subjects. It was
noted that soil ingestion estimates for this population
may be lower than estimates found by previous
studies in the literature because of families'
awareness of contamination from the Superfund site,
which may have resulted in altered behavior.
Soil ingestion estimates were also examined
based on various demographic characteristics. There
were no statistically significant differences in soil
ingestion based on age, sex, birth order, or house
yard characteristics (Calabrese et al., 1997a).
Although not statistically significant, soil ingestion
rates were generally higher for females, children
with lower birth number, children with parents
employed as laborers, service professionals,
homemakers, unemployed, and children with pets
(Calabrese et al., 1997a).
Table 5-6 shows the estimated soil and dust
ingestion by each tracer element and by the BTM.
Based on the best tracer, the mean soil ingestion rate
was 65.5 mg/day.
5.3.2.5. Stanek et al. (1998)—Prevalence of Soil
Mouthing/Ingestion among Healthy
Children Aged One to Six/Calabrese et ah
(1997b)—Soil Ingestion Rates in
Children Identified by Parental
Observation as Likely High Soil Ingesters
Stanek et al. (1998) conducted a survey response
study using in-person interviews of parents of
children attending well visits at three western
Massachusetts medical clinics in August,
September, and October of 1992. Of 528 children
ages 1 to 7 years with completed interviews, parents
reported daily mouthing or ingestion of sand and
stones in 6%, daily mouthing or ingestion of soil and
dirt in 4%, and daily mouthing or ingestion of dust,
lint, and dustballs in 1%. Parents reported more than
weekly mouthing or ingestion of sand and stones in
16%, more than weekly mouthing or ingestion of soil
and dirt in 10%, and more than weekly mouthing or
ingestion of dust, lint and dustballs in 3%. Parents
reported more than monthly mouthing or ingestion
of sand and stones in 27%, more than monthly
mouthing or ingestion of soil and dirt in 18%, and
more than monthly mouthing or ingestion of dust,
lint, and dustballs in 6%.
Calabrese and colleagues performed a follow-up
tracer element study (Calabrese et al., 1997b) for a
subset (N= 12) of the Stanek et al. (1998) children
(ages 1 to 3) whose caregivers had reported daily
sand/soil ingestion ( Y = 17). The time frame of the
follow-up tracer study relative to the original survey
response study was not stated; the study duration
was 7 days. Of the 12 children in Calabrese et al.
(1997b), one exhibited behavior that the authors
believed was clearly soil pica; Table 5-7 shows
estimated soil ingestion rates for this child during the
study period. Estimates ranged from -10 mg/day to
7,253 mg/day depending on the tracer. The mean soil
ingestion rate for the pica child, using aluminum and
silicon as tracers, is approximately 1,000 mg/day
(rounded to one significant figure).
Table 5-8 presents the estimated average daily
soil and dust ingestion estimates for the 12 children
studied. Estimates calculated based on soil tracer
element concentrations only for the 12 subjects
ranged from -15 to 1,783 mg/day based on
aluminum, -46 to 931 mg/day based on silicon, and
September 2017
Page 5-15

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
-47 to 3,581 mg/day based on titanium. Estimated
average daily dust ingestion estimates ranged from
-39 to 2,652 mg/day based on aluminum,
~51 to 3,145 mg/day based on silicon, and
—98 to 3,632 mg/day based on titanium. Quantities
for soil and dust are presented separately and assume
that the entire quantity of residual fecal tracers
originates entirely from soil or dust. Calabrese et al.
(1997b) questioned the validity of retrospective
caregiver reports of soil pica on the basis of the tracer
element results.
5.3.2.6. Davis and Mirick (2006)—Soil Ingestion
in Children and Adults in the Same
Family
Davis and Mirick (2006) calculated soil
ingestion for children and adults in the same family
using a tracer element approach. Data were collected
one year after the Davis et al. (1990) study was
conducted. Samples were collected and prepared for
laboratory analysis and then stored for a 2-year
period prior to tracer element quantification with
laboratory analysis. Analytical recovery values for
spiked samples were within the quality control limits
of ±25%. The 20 families in this study were a
nonrandom subset of the 104 families who
participated in the soil ingestion study by Davis et al.
(1990). Data collection issues resulted in sufficiently
complete data for only 19 of the 20 families
consisting of a child participant from the Davis et al.
(1990) study ages 3 to 7, inclusive, and a female and
male parent or guardian living in the same house.
Duplicate samples of all food and medication items
consumed, and all feces excreted, were collected for
11 consecutive days. Urine samples were collected
twice daily for 9 of the 11 days; for the remaining
2 days, attempts were made to collect full 24-hour
urine specimens. Soil and house dust samples were
also collected. Soil and dust samples were passed
successively through a 20- and 60-mesh (850 and
250 |im. respectively, ASTM, 2017) stainless steel
sieves. Only 12 children had sufficiently complete
data for use in the soil and dust ingestion estimates.
Tracer elements for this study included
aluminum, silicon, and titanium. Toothpaste was
supplied for use by study participants. In addition,
parents completed a daily diary of activities for
themselves and the participant child for
4 consecutive days during the study period.
Table 5-9 shows soil ingestion rates for all three
family member participants. The mean and median
estimates for children for all three tracers ranged
from 36.7 to 206.9 mg/day and 26.4 to 46.7 mg/day,
respectively, and fall within the range of those
reported by Davis et al. (1990). Adult soil ingestion
estimates ranged from 23.2 to 624.9 mg/day for
mean values and from 0 to 259.5 mg/day for median
values. This is based on 33 adults with complete
food, excreta, and soil data. Adult soil ingestion
estimates were more variable than those of children
in the study regardless of the tracer. The authors
believed that this higher variability may have
indicated an important occupational contribution of
soil ingestion in some, but not all, of the adults. As
in previous studies, the soil ingestion estimates were
the highest for titanium. Although toothpaste is a
known source of titanium, the titanium content of the
toothpaste used by study participants was not
determined.
Only three of a number of behaviors examined
for their relationship to soil ingestion were found to
be associated with increased soil ingestion in this
study:
•	Reported eating of dirt (for children),
•	Occupational contact with soil (for adults),
and
•	Hand washing before meals (for both
children and adults).
Several typical childhood behaviors, however,
including thumb-sucking, furniture licking, and
carrying around a blanket or toy were not associated
with increased soil ingestion for the participating
children. Among both parents and children, neither
nail-biting nor eating unwashed fruits or vegetables
was correlated with increased soil ingestion.
However, because duplicate food samples were used
to "correct" for dietary intake of tracers, accounting
for soil ingestion from eating unwashed fruits or
vegetables was not possible. Although eating
unwashed fruits or vegetables was not reflected in
the soil ingestion estimates in this study, the authors
noted that it is a behavior that could lead to soil
ingestion. When investigating correlations within
the same family, a child's soil ingestion was not
found to be associated with either parent's soil
ingestion, nor did the mother and father's soil
ingestion appear to be correlated.
5.3.3. Key Studies of Secondary Analysis
The following sections provide summaries of
key studies of secondary analysis (i.e., studies in
which researchers have interpreted previously
published results, or data that were originally
collected for a different purpose).
September 2017
Page 5-16

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
5.3.3.1. Calabrese and Stanek (1995)—Resolving
Intertracer Inconsistencies in Soil
Ingestion Estimation
Calabrese and Stanek (1995) explored sources
and magnitude of positive and negative errors in soil
ingestion estimates for children on a subject-week
and trace element basis. Calabrese and Stanek
(1995) identified possible sources of positive errors
as follows:
•	Ingestion of high levels of tracers before the
start of the study and low ingestion during the
study period and
•	Ingestion of element tracers from a nonfood
or nonsoil source during the study period.
Possible sources of negative bias were identified
as follows:
•	Ingestion of tracers in food that are not
captured in the fecal sample either due to
slow lag time or not having a fecal sample
available on the final study day and
•	Sample measurement errors that result in
diminished detection of fecal tracers, but not
in soil tracer levels.
The authors developed an approach that
attempted to reduce the magnitude of error in the
individual trace element ingestion estimates. Results
from a previous study conducted by Calabrese et al.
(1989) were used to quantify these errors based on
the following criteria: (1) a lag period of 28 hours
was assumed for the passage of tracers ingested in
food to the feces (this value was applied to all
subject-day estimates), (2) a daily soil ingestion rate
was estimated for each tracer for each 24-hour day a
fecal sample was obtained, (3) the median
tracer-based soil ingestion rate for each subject-day
was determined, and (4) negative errors due to
missing fecal samples at the end of the study period
were also determined. Also, upper- and lower-bound
estimates were determined based on criteria formed
using an assumption of the magnitude of the relative
standard deviation presented in another study
conducted by Stanek and Calabrese (1995b). Daily
soil ingestion rates for tracers that fell beyond the
upper and lower ranges were excluded from
subsequent calculations, and the median soil
ingestion rates of the remaining tracer elements were
considered the best estimate for that particular day.
The magnitude of positive or negative error for a
specific tracer per day was derived by determining
the difference between the value for the tracer and
the median value.
Table 5-10 presents the estimated magnitude of
positive and negative error for six tracer elements in
the children's study (conducted by Calabrese et al.,
1989). The original nonnegative mean soil ingestion
rates (see Table 5-3) ranged from a low of 21 mg/day
based on zirconium to a high of 459 mg/day based
on vanadium. The adjusted mean soil ingestion rate
after correcting for negative and positive errors
ranged from 97 mg/day based on yttrium to 208
mg/day based on titanium. Calabrese and Stanek
(1995) concluded that correcting for errors at the
individual level for each tracer element provides
more reliable estimates of soil ingestion.
5.3.3.2. Stanek and Calabrese (1995a)—Soil
Ingestion Estimates for Use in Site
Evaluations Based on the Best Tracer
Method
Stanek and Calabrese (1995a) recalculated soil
ingestion rates for adults and children from two
previous studies using data for eight tracers from
Calabrese et al. (1989) and three tracers from Davis
et al. (1990). Recalculations were performed using
the BTM. This method selected the "best" tracer(s),
by dividing the total amount of tracer in a particular
child's duplicate food sample by tracer concentration
in that child's soil sample to yield a (F:S) ratio. The
F:S ratio was small when the tracer concentration in
food was low compared to the tracer concentration
in soil. Small F:S ratios were desirable because they
lessened the impact of transit time error (the error
that occurs when fecal output does not reflect food
ingestion, due to fluctuation in gastrointestinal
transit time) in the soil ingestion calculation.
For adults, Stanek and Calabrese (1995a) used
data for eight tracers from the Calabrese et al. (1989)
study to estimate soil ingestion by the BTM. The
lowest F:S ratios were zirconium and aluminum and
the element with the highest F:S ratio was
manganese. For soil ingestion estimates based on the
median of the lowest four F:S ratios, the tracers
contributing most often to the soil ingestion
estimates were aluminum, silicon, titanium, yttrium,
vanadium, and zirconium. Using the median of the
soil ingestion rates based on the best four tracer
elements, the average adult soil ingestion rate was
estimated to be 64 mg/day with a median of
87 mg/day. The 95th percentile soil ingestion
September 2017
Page 5-17

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
estimate was 142 mg/day. These estimates are based
on 18 subject weeks for the 6 adult volunteers
described in Calabrese et al. (1989).
The BTM used a ranking scheme of F:S ratios to
determine the best tracers for use in the ingestion rate
calculation. To reduce the impact of biases that may
occur as a result of sources of fecal tracers other than
food or soil, the median of soil ingestion estimates
based on the four lowest F:S ratios was used to
represent soil ingestion.
Using the lowest four F:S ratios for each
individual child, calculated on a per-week
("subject-week") basis, the median of the soil
ingestion estimates from the Calabrese et al. (1989)
study most often included aluminum, silicon,
titanium, yttrium, and zirconium. Table 5-11
presents the soil ingestion estimates based on the
median values for aluminum, silicon, and titanium
for each child; the median of the best four tracers for
each child, and the best tracer for each child. Based
on the median of soil ingestion estimates from the
best four tracers, the mean soil ingestion rate for
children was 132 mg/day and the median was
33 mg/day. The 95th percentile value was
154 mg/day.
For the 101 children in the Davis et al. (1990)
study, the mean soil ingestion rate was 69 mg/day
and the median soil ingestion rate was 44 mg/day
(see Table 5-11). The 95th percentile estimate was
246 mg/day. These data are based on the three tracers
(i.e., aluminum, silicon, and titanium) from the
Davis et al. (1990) study. When the results for the
128 subject-weeks in Calabrese et al. (1989) and
101 children in Davis et al. (1990) were combined,
soil ingestion for children was estimated to be
104 mg/day (mean); 37 mg/day (median); and
217 mg/day (95th percentile), using the BTM.
5.3.3.3. Hogan et al. (1998)—Integrated
Exposure Uptake Biokinetic Model for
Lead in Children: Empirical
Comparisons with Epidemiologic Data
Hogan et al. (1998) used the IEUBK model, to
compare model predictions of blood lead levels with
epidemiological data to serve as one component of
the model validation. Environmental lead
measurement data from 478 children (38 were 0.5 to
<1 year; 440 were 1 to <7 years) across three
epidemiological studies were used as input to the
IEUBK model. Model results were compared to
blood lead levels from the same children. These
children were a subset of the entire population of
children living in three historic lead smelting
communities (Palmerton, PA; Madison County, IL;
and southeastern Kansas/southwestern Missouri),
whose environmental lead exposures (soil and dust
lead levels) had been studied as part of public health
evaluations in these communities. The study
populations were, in general, random samples of
children 6 months to 7 years of age. Children who
had lived in their residence for less than 3 months or
those reported by their parents to be away from home
more than 10 hours per week (>20 hours/week for
the Pennsylvania data set) were excluded due to lack
of information regarding lead exposure at the
secondary location. The nature of the soil and dust
exposures for the residential study population were
typical, with the sample size considered sufficiently
large to ensure that a wide enough range of
children's behavior would be spanned by the data.
Comparisons were made for a number of exposure
factors, including age, location, time spent away
from home, time spent outside, and whether or not
children took food outside to eat.
The IEUBK model is a biokinetic model for
predicting children's blood lead levels that uses
measurements of lead content in house dust, soil,
drinking water, food, and air. Model users use default
assumptions for the lead contents and intake rates for
each exposure medium (including soil) when they do
not have specific information for each child.
Hogan et al. (1998) compared children's
measured blood lead levels with biokinetic model
predictions (IEUBK version 0.99d) of blood lead
levels, using the children's measured drinking water,
soil, and dust lead contamination levels together
with default IEUBK model inputs for soil and dust
ingestion, relative proportions of soil and dust
ingestion, lead bioavailability from soil and dust, and
other model parameters. Thus, the default soil and
dust ingestion rates, and other default assumptions in
the model, were tested by comparing measured
blood lead levels with the model's predictions for
those children's blood lead levels. Most IEUBK
model kinetic and intake parameters were drawn
independently from published literature (White et
al., 1998; U.S. EPA, 1994b). Elimination parameters
in particular had relatively less literature to draw
upon (few data in children) and were fixed through
a calibration exercise using a data set with children's
blood lead levels paired with measured
environmental lead exposures in and around their
homes, while holding the other model parameters
constant.
Results for all community-wide children
6 months to 7 years of age were as follows: for
Palmerton, PA (TV =34), the geometric mean
measured blood lead levels (6.8 ng/dL) were slightly
over-predicted by the model (7.5 ng/dL); for
September 2017
Page 5-18

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
southeastern Kansas/southwestern Missouri
(N= 111), the blood lead levels (5.2 ng/dL) were
slightly under-predicted (4.6 |ig/dL). and for
Madison County, IL ( Y = 333), the geometric mean
measured blood lead levels matched the model
predictions (5.9 |ig/dL measured and predicted),
with very slight differences in the 95% confidence
interval. Geometric mean model predictions were
within 1 ng/dL of the observed geometric mean
blood lead levels in the three populations studied.
Hogan et al. (1998) noted that results may vary
depending on the ability to identify children's
playing areas and the use of different environmental
sampling methods. In addition, interactions between
socioeconomic status, race, and sex can vary across
communities and make generalizations of results
difficult. For 31 children 6 to 12 months old in the
Madison County, IL site only, Hogan et al. (1998)
reported a predicted geometric mean blood lead level
1.4-fold higher than the geometric mean blood lead
levels observed.
Default soil and dust ingestion rates used in this
version of the IEUBK model were: 135 mg/day for
1-, 2-, and 3-year-olds; 100 mg/day for 4-year-olds;
90 mg/day for 5-year-olds; and 85 mg/day for
6-year-olds (U.S. EPA 1994b, 2007). These values
represent mean soil and dust ingestion rates;
distributional data are not used in the model. The
time-averaged daily soil + dust ingestion rate for
these 6 years of life was 113 mg/day. Hogan et al.
(1998) did not provide information on the particle
sizes of the soil analyzed for this study. Because
particle size may be an important factor in estimating
the concentrations of elements in soil, this adds
uncertainties to the results. Regardless of this and
other uncertainties, these results suggest that the
combination of assumptions used as model input
parameters, including soil and dust ingestion rates,
are roughly accurate for 440 1- to <7-year-old
children in the three locations studied.
5.3.3.4. Ozkaynak et al. (2011)—Modeled
Estimates of Soil and Dust Ingestion
Rates for Children
Ozkaynak et al. (2011) developed soil and dust
ingestion rates for children 3 to <6 years of age using
U.S. EPA's SHEDS model for multimedia pollutants
(SHEDS-Multimedia). The authors had two main
objectives for this research: (1) to demonstrate an
application of the SHEDS model while identifying
and quantifying the key factors contributing to the
predicted variability and uncertainty in the soil and
dust ingestion exposure estimates and (2) to compare
the modeled results to existing tracer-element field
measurements. The SHEDS model is a physically
based probabilistic exposure model, which combines
diary information on sequential time spent in
different locations and activities drawn from EPA's
Consolidated Human Activity Database (CHAD),
with micro-activity data (e.g., hand-to-mouth
frequency, hand-to-surface frequency),
surface/object soil or dust loadings, and other
exposure factors (e.g., soil-to-skin adherence, saliva
removal efficiency). The SHEDS model generates
simulated individuals, who are then followed
through time, generally up to one year. The model
computes changes to their exposure at the diary
event level.
For this study, an indirect modeling approach
was used in which soil and dust were assumed to first
adhere to the hands and remain until washed off or
ingested by mouthing. The object-to-mouth pathway
for soil/dust ingestion was also addressed. For this
application of the SHEDS model, however, other
avenues of soil/dust ingestion were not considered.
Outdoor matter was designated as "soil" and indoor
matter as "dust." Estimates for the distributions of
exposure factors such as activity, time outdoors,
environmental concentrations, soil-skin and
dust-skin transfer, hand washing frequency and
efficiency, hand-mouthing frequency, area of object
or hand mouthed, mouthing removal rates, and other
variables were obtained from the literature. These
input variables were used in this SHEDS model
application to generate estimates of soil and dust
ingestion rates for a simulated population of 1,000.
Both sensitivity and uncertainty analyses were
conducted. Based on the sensitivity analysis, the
model results are the most sensitive to dust loadings
on carpet and hard floor surfaces, soil-skin
adherence factor, hand mouthing frequency, and
mean number of hand washes per day. Based on 200
uncertainty simulations that were conducted, the
modeling uncertainties were seen to be
asymmetrically distributed around the 50th (median)
or the central variability distribution.
Table 5-12 shows the predicted soil and dust
ingestion rates. Mean total soil and dust ingestion
was predicted to be 68 mg/day, with approximately
60% originating from soil ingestion, 30% from dust
on hands, and 10% from dust on objects. It is
important to note that this is different from the
assumptions used in the IEUBK model of 55% dust
and 45% soil. Hand-to-mouth soil ingestion was
found to be the most important pathway, followed by
hand-to-mouth dust ingestion, then object-to-mouth
dust ingestion. The authors noted that these modeled
estimates were found to be consistent with other
soil/dust ingestion values in the literature, but
September 2017
Page 5-19

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
slightly lower than the central tendency value of
100 mg/day recommended in EPA's Child-Specific
Exposure Factors Handbook (U.S. EPA, 2008).
The SHEDS methodology can be applied to
specific study populations of interest, a wide range
of input parameters, and can generate a full range of
distributions. One advantage of this methodology is
that it produced both ingestion of soil and ingestion
of dust. However, data for some of the input
variables are lacking. Data needs include additional
information on the activities and environments of
children in younger age groups, including children
with high hand-to-mouth, object-to-mouth, and pica
behaviors, and information on skin adherence and
dust loadings on indoor objects and floors, and
information to evaluate temporal variability.
Assumptions used, particularly for the transfer of
dust from both bare floors and carpets to hands and
soil loadings to hands while playing outdoors, may
be considered low for these parameters compared to
other experimental data available. The model also
assumes that the same hand area is mouthed on each
occasion, which may lead to lower dust intake rates.
In addition, other age groups of interest were not
included because of lack of data for some of the input
variables.
5.3.3.5. Wilson et aL (2013)—Revisiting Dust and
Soil Ingestion Rates Based on Hand-to-
Mouth Transfer
Wilson et al. (2013) provided estimates of the
ingestion rates of indoor dust and outdoor soil for
Canadians using both deterministic and probabilistic
methods. An indirect modeling approach was used.
Based on data from multiple studies, Wilson et al.
(2013) estimated dust and soil ingestion using
measures of particle loading to indoor surfaces, the
fraction transferred to the hands, hand surface areas,
the fraction of hand surface area that may be
mouthed or contact food, the frequency of
hand-to-mouth contacts, the amount dissolved in
saliva, and the exposure time. The following
equations were used to estimate indoor dust and
outdoor soil ingestion, respectively:
DIG = DSL x FTSS x SAhand x FSAfingers x FQ x SE x ET
(Eqn. 5-4)
and
SIR = SLhands x SAhand x FSAfingers xFQxSExET
(Eqn. 5-5)
where:
DIG
DSL
ET
FQ
FSAfingers
FTSS
SAhand
SE
SIR
SLhands
dust ingestion rate (mg/day)
dust surface loading
(mg/cm2)
exposure time (hours/day)
frequency of hand-to-mouth
events (events/hour)
fractional surface area of hand
mouthed
fraction of dust transferred from
surfaces to the skin
surface area of one hand (cm2)
saliva extraction factor (unitless)
soil ingestion rate (mg/day)
soil loading (mass of soil
adhering to hands) (mg/cm2)
The input parameters used in these equations are
provided in Table 5-13. For FTSS, it was assumed
that contact occurred with hard surfaces (e.g.,
nonporous floor surfaces such as tile or hardwood,
countertops, tables, window sills) 50% of the time
and with soft surfaces (e.g., carpets, sofas, beds)
50% of the time, except for infants for whom contact
was assumed to occur with soft surfaces only (i.e.,
100% of the time). In addition, it was assumed that
the front surface of the four fingers and the thumb of
one hand was the area mouthed.
Mean dust ingestion rates were similar within
age groups when using either the deterministic or
probabilistic approach based on 200,000 trials (see
Table 5-14), ranging from 2.2 mg/day for teenagers
(ages 12 to 19 years) to 41 mg/day fortoddlers (ages
7 months to 4 years). Mean soil ingestion rates
ranged from 1.2 mg/day for seniors (ages 60+ years)
to 23 mg/day for children (ages 5 to 11 years) using
the probabilistic approach (see Table 5-14). Mean
soil ingestion rates using the deterministic approach
were similar. Combined dust and soil ingestion rates
ranged from 3.7 mg/day for teenagers (ages 12 to 19
years) to 61 mg/day for toddlers (ages 7 months to 4
years). The 95thpercentile combined soil and dust
ingestion rates ranged from 12.6 mg/day for
teenagers to 204 mg/day for toddlers.
Ingestion rates were estimated for a wide range
of age groups, and separate ingestion estimates were
provided for both dust and soil. However, the study
used input values from multiple studies to generate
the dust and soil ingestion rates, and each individual
study is expected to have its own shortcomings,
which also apply to this analysis. Some of these
limitations include uncertainties with regard to the
methodologies for dust collection in hard and soft
surfaces in the studies selected for the analysis,
September 2017
Page 5-20

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
assumptions of a single hand contact with surfaces,
assumptions about the parts of the hand mouthed,
and not accounting for effects of soil and hand
moisture on soil and dust loadings on the skin. Also,
the model does not account for object-to-mouth
contact. Many of these assumptions used by Wilson
et al. (2013) in the model would tend to
underestimate amount of soil and dust ingested.
5.3.3.6. Von Lindern etal. (2016)—Estimating
Children's Soil/Dust Ingestion Rates
through Retrospective Analyses of Blood
Lead Biomonitoring from the Bunker
Hill Superfund Site in Idaho
Von Lindern et al. (2016) conducted an analysis
to estimate age-specific soil/dust ingestion rates
using the IEUBK model for lead in children and soil
and house dust lead bioavailability data from the
Bunker Hill Mining and Metallurgical Complex
Superfund Site in Idaho. A total of 271 soil/dust
samples archived during 1986-2002, including
193 house dust samples, 73 yard soil samples, and
5 quality control samples, were retrieved from
storage. The samples were matched with blood lead
data and information on the child's age and sex,
home location, and property remediation status.
After sieving to 80-mesh (180 |im: ASTM, 2017),
they were analyzed for total lead and in vitro
bioaccessibility. In vivo relative and absolute
bioavailability was calculated from in vitro
bioaccessibility results. A total of 2,176 records of
blood/soil/dust/lead concentrations were available.
Structural equations modeling (SEM), which is a
statistical multivariate methodology, was used to
determine soil and dust partitions, age-specific soil
and dust intake rates, and lead uptake from sources
other than soil and dust (i.e., air, diet, and water). Von
Lindern et al. (2016) calculated the total lead uptake
by dividing the measured blood lead levels (|ig/L) by
the age-specific biokinetic slope factors in the
IEUBK model. Total lead uptake is the sum of the
uptakes from the four components: air, water, diet,
and soil/dust. Soil and dust intake rates were
calculated by assigning partition coefficients (i.e.,
fractional contributions to total soil/dust ingestionby
each source) using Equation 5-6.
IRsd = 1,000 x {UPsd/[(Cd x PTd x ABSd) +
(Cys x PTys x ABSys) + (Ccs x PTCS x ABScs) +
(Cns x PTns x i4B5ns)]}
(Eqn. 5-6)
where:
IRsd = soil/dust intake rate (mg/day)
UPsd = uptake from soil/dust (|ig/day)
C = concentration from the various
sources (i.e., d = dust; ys = yard
soil; cs = community soil; ns =
neighborhood soil) (mg/kg)
PT = partition coefficient for the various
sources (unitless)
ABS = absolute bioavailability for the
various sources (unitless)
Soil and dust intake rates were calculated for four
source partition (PT) scenarios: (1) the IEUBK
model default of 55% dust and 45% soil (55/45); (2)
the original Bunker Hill Superfund Site model of
40% dust, 30% yard soil, and 30% geometric mean
community soil (40/30/30GM); (3) the same
partition as in scenario 2, but using the arithmetic
mean (40/30/30AM); and (4) the SEM using
50% dust, 25% yard soil, 10% neighborhood soil,
and 15% community soil (50/25/10/15). Mean soil
and dust absolute bioavailability (ABS) was
estimated to be 33% (SD ± 4%) and 28% (SD ± 6%),
respectively. These values are similar to the
30% value recommended in the IEUBK model as a
default.
Central tendency age-specific soil and dust intake
rates for the four partition scenarios were estimated
using Equation 5-5, and are presented in Table 5-15.
All partition scenarios produced similar central
tendency intake rates (von Lindern et al., 2016). For
children ages 0.5 to 9 years, the mean soil and dust
ingestion rates ranged from 47 mg/day to
100 mg/day. Among the age groups evaluated,
children ages 1 to <2 years had the highest mean soil
and dust ingestion rates (i.e., 89-100 mg/day), and
children older than 2 years, had estimated soil and
dust ingestion rates averaging approximately
60 mg/day. Age-specific distributions of soil and
dust intake rates for the four partition scenarios are
shown in Table 5-16. The 95th percentile soil and
dust intake rates ranged from 120 mg/day to
493 mg/day. The 55/45 partition scenario
consistently produced higher 95th percentile soil and
dust intake rates compared to the other scenarios.
These age specific soil/dust intake rates were input
to the IEUBK model to compare predicted and
observed blood lead levels (von Lindern et al.,
2016). Linear regression analyses indicated that the
50/25/10/15 and the 40/30/30 average intake rates
were the partition scenarios that best fit the blood
lead levels predicted by the IEUBK model to the
observed blood lead levels. The data from these two
September 2017
Page 5-21

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
partition scenarios were used in developing the
recommended soil and dust ingestion values shown
in Tables 5-1 and 5-34.
This study provides soil and dust intake rates for
various age groups including very young children.
Blood lead data were collected for a number of
consecutive years. The strengths of this study are
that the soil and dust ingestion estimates are based
on a large number of samples, and that the estimates
represent long-term exposures. One of the
limitations of this study is that partition coefficients
that were derived may not be representative of all
age groups, all neighborhoods in the study, or other
populations. Partition of soil and dust materials
contributing to an individual child's lead uptake can
be expected to vary strongly among residences and
among children. The authors also acknowledged that
education and intervention programs may have
resulted in children's temporary reduction in
soil/dust ingestion rates at this site (von Lindern et
al., 2016). Increased knowledge and concern about
lead exposure may lead to residents taking steps to
reduce their child's lead exposure through more
frequent hand washing, more household vacuuming,
and changing the areas where children play. This is
evidenced by a decline in the estimated average
ingestion rates from 1988 to 2002 when the analysis
was performed for the different partition scenarios
by year. Other uncertainties include not accounting
for the variability in the children's blood lead levels
with time, a limited number of soil and dust
measurements, the representativeness of the soil and
dust measurements of the children's play areas, and
measurement error. In addition, the equations
derived by von Lindern et al. (2016) for the analysis
would be highly sensitive to low lead media
concentrations and less sensitive to higher lead
concentrations.
5.3.4. Relevant Studies of Primary
Analysis
The following studies are classified as relevant
rather than key. They either do not provide a
quantitative estimate of soil ingestion, or they are
estimates generated for a population that may not be
representative of the U.S. general population.
Studies that provide data for special populations
such as those engaging in wilderness lifestyles in
Canada (Doyle et al., 2012; Irvine et al., 2014) are
presented in this section. Data from these studies
may be appropriate for high soil contact scenarios.
The general population tracer element studies
described in this section are not designated as key
because the methodology to account for nonsoil
tracer exposures (e.g., food, medicine) was not as
well developed as the methodology in the U.S. tracer
element studies described in Sections 5.3.2 and
5.3.3. However, the method of Clausing et al. (1987)
and data of van Wijnen et al. (1990) were used in
developing biokinetic model default soil and dust
ingestion rates (U.S. EPA, 1994a) used in the Hogan
et al. (1998) study and von Lindern et al. (2016),
which were designated as key. In most cases, the
survey response studies were of a nonrandomized
design, provided insufficient information to
determine important details regarding study design,
or provided no data to allow quantitative estimates
of soil and/or dust ingestion rates.
5.3.4.1.	Dickins and Ford (1942)—Geophagy
(Dirt Eating) among Mississippi Negro
School Children
Dickins and Ford (1942) conducted a survey
response study of rural Black school children (4th
grade and above) in Oktibbeha County, MS in
September 1941. A total of 52 of 207 children (18 of
69 boys and 34 of 38 girls) studied gave positive
responses to questions administered in a test-taking
format regarding having eaten dirt in the previous 10
to 16 days. The authors stated that the study sample
likely was more representative of the higher
socioeconomic levels in the community because
older children from lower socioeconomic levels
sometimes left school in order to work and because
children in the lower grades, who were more
socioeconomically representative of the overall
community, were excluded from the study. Clay was
identified as the predominant type of soil eaten.
5.3.4.2.	Ferguson and Keaton (1950)—Studies of
the Diets of Pregnant Women in
Mississippi: II Diet Patterns
Ferguson and Keaton (1950) conducted a survey
response study of a group of 361 pregnant women
receiving health care at the Mississippi State Board
of Health, who were interviewed regarding their diet,
including the consumption of clay or starch. All of
the women were from the lowest economic and
educational level in the area, and 92% were Black.
Of the Black women, 27% reported eating clay and
41% eating starch. In the group of White women,
7 and 10% reporting clay- and starch-eating,
respectively. The amount of starch eaten ranged from
2-3 small lumps to 3 boxes (24 ounces) per day. The
amount of clay eaten ranged from one tablespoon to
one cup per day.
September 2017
Page 5-22

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
5.3.4.3.	Cooper (1957)—Pica: A Survey of the
Historical Literature as Well as Reports
from the Fields of Veterinary Medicine
and Anthropology, the Present Study of
Pica in Young Children, and a
Discussion of Its Pediatric and
Psychological Implications
Cooper (1957) conducted a nonrandomized
survey response study in the 1950s of children age
7 months or older referred to a Baltimore, MD
mental hygiene clinic. For 86 out of 784 children
studied, parents or caretakers gave positive
responses to the question, "Does your child have a
habit, or did he ever have a habit, of eating dirt,
plaster, ashes, etc.?" and identified dirt, or dirt
combined with other substances, as the substance
ingested. Cooper (1957) described a pattern of pica
behavior, including ingesting substances other than
soil, being most common between ages 2 and 4 or
5 years, with one of the 86 children ingesting clay at
age 10 years and 9 months.
5.3.4.4.	Barltrop (1966)—The Prevalence of Pica
Barltrop (1966) conducted a randomized survey
response study of children born in Boston, MA
between 1958 and 1962, inclusive, whose parents
resided in Boston and who were neither illegitimate
nor adopted. A stratified random subsample of 500
of these children was contacted for in-person
caregiver interviews, in which a total of 186 families
(37%) participated. A separate stratified subsample
of 1,000 children was selected for a mailed survey in
which 277 (28%) of the families participated.
Interview-obtained data regarding caregiver reports
of pica (in this study is defined as placing nonfood
items in the mouth and swallowing them) behavior
in all children ages 1 to 6 years in the 186 families
( Y = 439) indicated 19 had ingested dirt (defined as
yard dirt, house dust, plant-pot soil, pebbles, ashes,
cigarette ash, glass fragments, lint, and hair
combings) in the preceding 14 days. These data do
not appear to have been corrected for unequal
selection probability in the stratified random sample,
nor were they corrected for nonresponse bias.
Interviews were conducted in the March/April time
frame, presumably in 1964. Mail-survey obtained
data regarding caregiver reports of pica in the
preceding 14 days indicated that 39 of 277 children
had ingested dirt, presumably using the same
definition as above. Barltrop (1966) mentions
several possible limitations of the study, including
nonparticipation bias and respondents' memory, or
recall, effects.
5.3.4.5.	Bruhn and Pangborn (1971)—Reported
Incidence of Pica among Migrant
Families
Bruhn and Pangborn (1971) conducted a survey
among 91 low income families of migrant
agricultural workers in California in May through
August 1969. Families in two labor camps (Madison
camp, 10 miles west of Woodland, and Davis camp,
10 miles east of Davis) were of Mexican descent, and
families in one camp (Harney Lane camp 17 miles
north of Stockton) were "Anglo." Participation was
34 of 50 families at the Madison camp, 31 of
50 families at the Davis camp, and 26 of 26 families
at the Harney Lane camp. Respondents for the
studied families (primarily wives) gave positive
responses to open-ended questions such as "Do you
know of anyone who eats dirt or laundry starch?"
Bruhn and Pangborn (1971) apparently asked a
modified version of this question pertaining to the
respondents' own or relatives' families. They
reported 18% (12 of 65) of Mexican families'
respondents as giving positive responses for
consumption of "dirt" among children within the
Mexican respondents' own or relatives' families.
They reported 42% (11 of 26) of "Anglo" families'
respondents as giving positive responses for
consumption of "dirt" among children within the
Anglo respondents' own or relatives' families.
5.3.4.6.	Robischon (1971)—Pica Practice and
Other Hand-Mouth Behavior and
Children's Developmental Level
A survey response sample of 19- to 24-month-old
children examined at an urban well-child clinic in
the late 1960s or 1970 in an unspecified location
indicated that 48 of the 130 children whose
caregivers were interviewed, exhibited pica behavior
(defined as "ate nonedibles more than once a week").
The specific substances eaten were reported for 30
of the 48 children. All except 2 of the 30 children
habitually ate more than one nonedible substance.
The soil and dust-like substances reported as eaten
by these 30 children were: ashes (17), "earth" (5),
dust (3), fuzz from rugs (2), clay (1), and
pebbles/stones (1). Caregivers for some of the study
subjects (between 0 and 52 of the 130 subjects, exact
number not specified) reported that the children "ate
nonedibles less than once a week."
5.3.4.7.	Bronstein and Dollar (1974)—Pica in
Pregnancy
The frequency and effects of pica behavior was
investigated by Bronstein and Dollar (1974) in
410 pregnant, low-income women from both urban
September 2017
Page 5-23

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
(TV = 201) and rural (N= 209) areas in Georgia. The
women selected were part of the Nutrition
Demonstration Project, a study investigating the
effect of nutrition on the outcome of the pregnancy,
conducted at the Eugene Talmadge Memorial
Hospital and University Hospital in Augusta, GA.
During their initial prenatal visit, each patient was
interviewed by a nutrition counselor who questioned
her food frequency, social and dietary history, and
the presence of pica. Patients were categorized by
age, parity, and place of residence (rural or urban).
Of the 410 women interviewed, 65 (16%) stated
that they practiced pica. A variety of substances were
ingested, with laundry starch being the most
common. There was no significant difference in the
practice of pica between rural and urban women,
although older rural women (20-35 years) showed a
greater tendency to practice pica than younger rural
or urban women (<20 years). The number of
previous pregnancies did not influence the practice
of pica. The authors noted that the frequency of pica
among rural patients had declined from a previous
study conducted 8 years earlier, and attributed the
reduction to a program of intensified nutrition
education and counseling provided in the area. No
specific information on the amount of pica
substances ingested was provided by this study, and
the data are more than 30 years old.
5.3.4.8. Hook (1978)—Dietary Cravings and
Aversions during Pregnancy
Hook (1978) conducted interviews of
250 women who had each delivered a live infant at
two New York hospitals; the interviews took place in
1975. The mothers were first asked about any
differences in consumption of seven beverages
during their pregnancy, and the reasons for any
changes. They were then asked, without mentioning
specific items, about any cravings or aversions for
other foods or nonfood items that may have
developed at any time during their pregnancy.
Nonfood items reportedly ingested during
pregnancy were ice, reported by three women, and
chalk from a river clay bank, reported by one
woman. In addition, one woman reported an
aversion to nonfood items (specific nonfood item not
reported). No quantity data were provided by this
study.
5.3.4.9. Binder et al. (1986)—Estimating Soil
Ingestion: The Use of Tracer Elements in
Estimating the Amount of Soil Ingested
by Young Children
Binder et al. (1986) used a tracer technique
modified from a method previously used to measure
soil ingestion among grazing animals to study the
ingestion of soil among children 1 to 3 years of age
who wore diapers. The children were studied during
the summer of 1984 as part of a larger study of
residents living near a lead smelter in East Helena,
MT. Fecal samples from diapers were collected over
a 3-day period from 65 children (42 males and
23 females), and composited samples of soil were
obtained from the children's yards. Both excreta and
soil samples were analyzed for aluminum, silicon,
and titanium. These elements were found in soil but
were thought to be poorly absorbed in the gut and to
have been present in the diet only in limited
quantities. Excreta measurements were obtained for
59 of the children. Soil ingestion by each child was
estimated on the basis of each of the three tracer
elements using a standard assumed fecal dry weight
of 15 g/day, and the following Equation 5-7:
f,t xFi
The = —	1	(Eqn. 5-7)
$i,e
where:
Ti,e = estimated soil ingestion for child /'
based on element e (g/day)
fi,e = concentration of element e in fecal
sample of child /' (mg/g)
Fi = fecal dry weight (g/day)
Siie = concentration of element e in child z"s
yard soil (mg/g)
The analysis assumed that (1) the tracer elements
were neither lost nor introduced during sample
processing, (2) the soil ingested by children
originates primarily from their own yards, and
(3) that absorption of the tracer elements by the
children occurred in only small amounts. The study
did not distinguish between ingestion of soil and
house dust, nor did it account for the presence of the
tracer elements in ingested foods or medicines.
The arithmetic mean quantity of soil ingested by
the children in the Binder et al. (1986) study was
estimated to be 181 mg/day (range 25 to 1,324)
based on the aluminum tracer, 184 mg/day (range 31
September 2017
Page 5-24

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
to 799) based on the silicon tracer, and 1,834 mg/day
(range 4 to 17,076) based on the titanium tracer (see
Table 5-17). The overall mean soil ingestion
estimate, based on the minimum of the three
individual tracer estimates for each child, was
108 mg/day (range 4 to 708). The median values
were 121 mg/day, 136 mg/day, and 618 mg/day for
aluminum, silicon, and titanium, respectively. The
95th percentile values for aluminum, silicon, and
titanium were 584 mg/day, 578 mg/day, and
9,590 mg/day, respectively. The 95thpercentile value
based on the minimum of the three individual tracer
estimates for each child was 386 mg/day.
The authors were not able to explain the
difference between the results for titanium and for
the other two elements, but they speculated that
unrecognized sources of titanium in the diet or in the
laboratory processing of stool samples may have
accounted for the increased levels. The frequency
distribution graph of soil ingestion estimates based
on titanium in the Binder et al. (1986) paper (not
provided here) shows that a group of 21 children had
particularly high titanium values
(i.e., >1,000 mg/day). The remainder of the children
showed titanium ingestion estimates at lower levels,
with a distribution more comparable to that of the
other elements.
5.3.4.10. Clausing et al. (1987)—A Method for
Estimating Soil Ingestion by Children
Clausing et al. (1987) conducted a soil ingestion
study with Dutch children using a tracer element
methodology. The study measured aluminum,
titanium, and acid-insoluble residue (AIR) contents
of fecal samples from children aged 2 to 4 years
attending a nursery school, and for samples of
playground dirt at that school. Over a 5-day period,
27 daily fecal samples were obtained for 18 children.
Soil samples from the direct surroundings of the
nursery school were sieved through a 250 |im screen.
Using the average soil concentrations present at the
school, and assuming a standard fecal dry weight of
10 g/day, soil ingestion was estimated for each
tracer. Six hospitalized, bedridden children served as
a control group, representing children who had very
limited access to soil; eight daily fecal samples were
collected from the hospitalized children.
Recoveries from analytical analyses ranged from
54-89% for titanium and aluminum. Without
correcting for the tracer element contribution from
background sources, represented by the hospitalized
children's soil ingestion estimates, the
aluminum-based soil ingestion estimates for the
school children in this study ranged from 23 to
979 mg/day, the AIR-based estimates ranged from
48 to 362 mg/day, and the titanium-based estimates
ranged from 64 to 11,620 mg/day. As in the Binder
et al. (1986) study, a fraction of the children (6/18)
showed titanium values above 1,000 mg/day, with
most of the remaining children showing
substantially lower values. Calculating an arithmetic
mean quantity of soil ingested based on each fecal
sample yielded 232 mg/day for aluminum;
129 mg/day for AIR, and 1,431 mg/day for titanium
(see Table 5-18). Based on the limiting tracer method
(LTM) and averaging across each fecal sample, the
arithmetic mean soil ingestion was estimated to be
105 mg/day with a population standard deviation of
67 mg/day (range 23 to 362 mg/day); geometric
mean soil ingestion was estimated to be 90 mg/day.
Use of the LTM assumed that "the maximum amount
of soil ingested corresponded with the lowest
estimate from the three tracers" (Clausing et al.,
1987).
The hospitalized children's arithmetic mean
aluminum-based soil ingestion estimate was
56 mg/day; titanium-based estimates included
estimates for three of the six children that exceeded
1,000 mg/day, with the remaining three children in
the range of 28 to 58 mg/day (see Table 5-19). AIR
measurements were not reported for the hospitalized
children. Using the LTM method, the mean soil
ingestion rate was estimated to be 49 mg/day with a
population standard deviation of 22 mg/day (range
26 to 84 mg/day). The geometric mean soil ingestion
rate was 45 mg/day. The hospitalized children's data
suggested a major nonsoil source of titanium for
some children and a background nonsoil source of
aluminum. However, conditions specific to
hospitalization (e.g., medications) were not
considered.
Clausing et al. (1987) estimated that the average
soil ingestion of the nursery school children was
56 mg/day, after subtracting the mean LTM soil
ingestion for the hospitalized children (49 mg/day)
from the nursery school children's mean LTM soil
ingestion (105 mg/day), to account for background
tracer intake from dietary and other nonsoil sources.
5.3.4.11. Van Wijnen et ah (1990)—Estimated Soil
Ingestion by Children
In a tracer element study by Van Wijnen et al.
(1990), soil ingestion among Dutch children ranging
in age from 1 to 5 years was evaluated using a tracer
element methodology. Van Wijnen et al. (1990)
measured three tracers (titanium, aluminum, and
AIR) in soil and feces. The authors estimated soil
ingestion based on the LTM, which assumed that soil
September 2017
Page 5-25

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
ingestion could not be higher than the lowest value
of the three tracers. LTM values represented soil
ingestion estimates that were not corrected for
dietary intake. Recoveries for aluminum and
titanium from soil materials mixed with fecal
samples were 94 and 97%, respectively.
An average daily feces dry weight of 15 grams
was assumed. A total of 292 children attending
daycare centers were studied during the first of two
sampling periods and 187 children were studied in
the second sampling period; 162 of these children
were studied during both periods (i.e., at the
beginning and near the end of the summer of 1986).
A total of 78 children were studied at campgrounds.
Soil samples were sieved through a 2-mm mesh. The
authors reported geometric mean LTM values
because soil ingestion rates were found to be skewed
and the log-transformed data were approximately
normally distributed. Geometric mean LTM values
were estimated to be 111 mg/day for children in
daycare centers and 174 mg/day for children
vacationing at campgrounds (see Table 5-20). For
the 162 daycare center children studied during both
sampling periods the arithmetic mean LTM was
162 mg/day, and the median was 114 mg/day. For
the 78 children at the campgrounds, the overall
arithmetic mean LTM was 213 mg/day and the
median was 160 mg/day.
Fifteen hospitalized children were studied and
used as a control group. These children's LTM soil
ingestion estimates were 74 (geometric mean),
93 (mean), and 110 (median) mg/day. The authors
assumed the hospitalized children's soil ingestion
estimates represented dietary intake of tracer
elements, and used rounded 95% confidence limits
on the arithmetic mean, 70 to 120 mg/day (midpoint
95 mg/day), to correct the daycare and campground
children's LTM estimates for dietary intake of
tracers. Although the authors suggested that
corrections should be made to the soil ingestion
values obtained from the daycare and campground
subjects by subtracting values obtained from the
hospitalized children (i.e., background), they do not
appear to have made these adjustments to the
arithmetic means. These corrections would result in
soil ingestion rates of 67 mg/day (162 mg/day minus
95 mg/day) for daycare children and 118 mg/day
(213 mg/day minus 95 mg/day) for campers. Van
Wijnen et al. (1990) showed corrected geometric
mean soil ingestion to range from 0 to 90 mg/day,
with a 90th percentile value of up to 190 mg/day for
the various age categories within the daycare group
and 30 to 200 mg/day, with a 90th percentile value of
up to 300 mg/day for the various age categories
within the camping group.
AIR was the limiting tracer in about 80% of the
samples. Among children attending daycare centers,
soil ingestion was also found to be higher when the
weather was good (i.e., <2 days/week precipitation)
than when the weather was bad (i.e., >4 days/week
precipitation) (see Table 5-21).
5.3.4.12.	Calabrese et aL (1990)—Preliminary
Adult Soil Ingestion Estimates: Results of
a Pilot Study
Calabrese et al. (1990) studied six adults to
evaluate the extent to which they ingest soil. This
adult study was originally part of the children soil
ingestion study (Calabrese et al., 1989) and was used
to validate part of the analytical methodology used
in the children's study. The participants were six
healthy adults, three males and three females,
25-41 years old. Each volunteer ingested one empty
gelatin capsule at breakfast and one at dinner
Monday, Tuesday, and Wednesday during the first
week of the study. During the second week, they
ingested 50 mg of sterilized soil within a gelatin
capsule at breakfast and at dinner (a total of 100 mg
of sterilized soil per day) for 3 days. For the third
week, the participants ingested 250 mg of sterilized
soil in a gelatin capsule at breakfast and at dinner (a
total of 500 mg of soil per day) during the 3 days.
Duplicate meal samples (food and beverage) were
collected from the six adults. The sample included
all foods ingested from breakfast Monday, through
the evening meal Wednesday during each of the
3 weeks. In addition, all medications and vitamins
ingested by the adults were collected. Total excretory
output was collected from Monday noon through
Friday midnight over three consecutive weeks.
Data obtained from the first week, when empty
gelatin capsules were ingested, were used to estimate
soil intake by adults. On the basis of recovery values,
aluminum, silicon, yttrium, and zirconium were
considered the most valid tracers. The mean values
for these four tracers were: aluminum, 110 mg/day;
silicon, 30 mg/day; yttrium, 63 mg/day; and
zirconium, 134 mg/day. A limitation of this study is
the small sample size. Thus, this study was classified
as relevant.
5.3.4.13.	Cooksey (1995)—Pica and Olfactory
Craving of Pregnancy: How Deep Are the
Secrets?
Postpartum interviews were conducted between
1992 and 1994 of 300 women at a mid-western
hospital, to document their experiences of pica
behavior. The majority of women were Black and
low-income, and ranged in age from 13 to 42 years.
September 2017
Page 5-26

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
In addition to questions regarding nutrition, each
woman was asked if during her pregnancy she
experienced a craving to eat ice or other things that
are not food.
Of the 300 women, 194 (65%) described
ingesting one or more pica substances during their
pregnancy, and the majority (78%) ate ice/freezer
frost alone or in addition to other pica substances.
Reported quantities of items ingested on a daily basis
were three to four 8-pound bags of ice, two to three
boxes of cornstarch, two cans of baking powder, one
cereal bowl of dirt, five quarts of freezer frost, and
one large can of powdered cleanser.
5.3.4.14. Smulian et aL (1995)—Pica in a Rural
Obstetric Population
In 1992, Smulian et al. (1995) conducted a
survey response study of pica in a convenience
sample of 125 pregnant women in Muscogee
County, GA, who ranged in age from 12 to 37 years.
Of these, 73 were Black, 47 were White, 4 were
Hispanic, and 1 was Asian. Interviews were
conducted at the time of the first prenatal visit, using
nondirective questionnaires to obtain information
regarding substances ingested as well as patterns of
pica behavior and influences on pica behavior. Only
women ingesting nonfood items were considered to
have pica. Ingestion of ice was included as a pica
behavior only if the ice was reported to be ingested
multiple times per day, if the ice was purchased
solely for ingestion, or if the ice was obtained from
an unusual source such as freezer frost.
The overall prevalence of pica behavior in this
study was 14.4% (18 of 125 women), and was
highest among Black women (17.8%). There was no
significant difference between groups with respect to
age, race, weight, or gestational age at the time of
enrollment in the study. The most common form of
pica was ice eating (pagophagia), reported by
44.4% of the patients. Nine of the women reported
information on the frequency and amount of the
substances they were ingesting. Of these women,
66.7% reported daily consumption and 33.3%
reported pica behavior three times per week. Soap,
paint chips, or burnt matches were reportedly
ingested 3 days per week. One patient ate ice
60 times per week. Women who ate dirt or clay
reported ingesting 0.5-1 pound per week. The
largest amount of ice consumed was five pounds per
day.
5.3.4.15.	Grigsby et ah (1999)—Chalk Eating in
Middle Georgia: A Culture-Bound
Syndrome of Pica?
Grigsby et al. (1999) investigated the ingestion
of kaolin, also known as white dirt, chalk, or white
clay, in the central Georgia Piedmont area as a
culture-bound syndrome. A total of 21 individuals
who consumed kaolin at the time or had a history of
consuming kaolin were interviewed, using a
seven-item, one-page interview protocol. All of
those interviewed were Black, ranging in age from
28 to 88 years (mean age of 46.5 years), and all were
female except for one.
Reasons for eating kaolin included liking the
taste, being pregnant, craving it, and to gain weight.
Eight respondents indicated that they obtained the
kaolin from others, five reported getting it directly
from the earth, four purchased it from a store, and
two obtained it from a kaolin pit mine. The majority
of the respondents reported that they liked the taste
and feel of the kaolin as they ate it. Only three
individuals reported knowing either males or White
persons who consumed kaolin. Most individuals
were not forthcoming in discussing their ingestion of
kaolin and recognized that their behavior was
unusual.
The study suggests that kaolin-eating is primarily
practiced by Black women who were introduced to
the behavior by family members or friends, during
childhood or pregnancy. The authors concluded that
kaolin ingestion is a culturally transmitted form of
pica, not associated with any other psychopathology.
Although information on kaolin eating habits and
attitudes were provided by this study, no quantitative
information on consumption was included, and the
sample population was small and nonrandom.
5.3.4.16.	Ward and Kutner (1999)—Reported Pica
Behavior in a Sample of Incident Dialysis
Patients
Structured interviews were conducted with a
sample of 226 dialysis patients in the metropolitan
Atlanta, GA area from September 1996 to September
1997. Interviewers were trained in nutrition data
collection methods, and patients also received a
3-day diet diary that they were asked to complete
and return by mail. If a subject reported a strong past
or current food or nonfood craving, a separate form
was used to collect information to determine whether
this was a pica behavior.
Pica behavior was reported by 37 of the dialysis
patients studied (16%), and most of these patients
(31 of 37) reported that they were currently
practicing some form of pica behavior. The patients'
September 2017
Page 5-27

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
race and sex were significantly associated with pica
behavior, with Black patients and women making up
86% and 84% of those reporting pica, respectively.
Those reporting pica behavior were also younger
than the remainder of the sample, and two patients
described a persistent craving for ice. Other pica
items reportedly consumed included starch, dirt,
flour, or aspirin.
5.3.4.17. Simpson etal. (2000)—Pica During
Pregnancy in Low-Income Women Born
in Mexico
Simpson et al. (2000) interviewed
225 Mexican-born women, aged 18-42 years (mean
age of 25 years), using a questionnaire administered
in Spanish. Subjects were recruited by approaching
women in medical facilities that served low-income
populations in the cities of Ensenada, Mexico
(N= 75), and Santa Ana, Bakersfield and East Los
Angeles, CA ( Y = 150). Criteria for participation
were that the women had to be Mexican-born, speak
Spanish as their primary language, and be pregnant
or have been pregnant within the past year. Only the
data for the women in the United States are included
in this handbook.
Pica behavior was reported in 31% of the women
interviewed in the United States. Table 5-22 shows
the items ingested and the number of women
reporting the pica behavior. Of the items ingested,
only ice was said to be routinely eaten outside of
pregnancy, and was only reported by U.S. women,
probably because none of the low-income women
interviewed in Mexico owned a refrigerator.
Removing the 12 women who reported eating only
ice from the survey lowers the percentage of U.S.
women who reported pica behavior to 23%. Women
said they engaged in pica behavior because of the
taste, smell, or texture of the items, for medicinal
purposes, or because of advice from someone, and
one woman reported eating clay for religious
reasons. Magnesium carbonate, a pica item not
found to be previously reported in the literature, was
reportedly consumed by 17% of women. The
amount of magnesium carbonate ingested ranged
from a quarter of a block to five blocks per day; the
blocks were approximately the size of a 35-mm film
box. No specific quantity information on the
amounts of pica substances ingested was provided in
the study.
5.3.4.18. Obialo et al. (2001)—Clay Pica Has No
Hematologic or Metabolic Correlate to
Chronic Hemodialysis Patients
A total of 138 dialysis patients at the Morehouse
School of Medicine, Atlanta, GA, were interviewed
about their unusual cravings or food habits. The
patients were Black and ranged in age from 37 to
78 years. Obialo et al. (2001) suggested that the
stress caused by end-stage renal disease may provide
a stimulus for pica, especially for those with cultural
predispositions.
Thirty of the patients (22%) reported some form
of pica behavior, while 13 patients (9.4%) reported
clay pica. The patients with clay pica reported daily
consumption of 225-450 g of clay.
5.3 A.\9.Klitzman et aL (2002)—Lead Poisoning
among Pregnant Women in New York
City: Risk Factors and Screening
Practices
Klitzman et al. (2002) interviewed 33 pregnant
women whose blood lead levels were >20 ng/dL as
reported to the New York City Department of Health
between 1996 and 1999. The median age of the
women was 24 years (range of 15 to 43 years), and
the majority were foreign born. The women were
interviewed regarding their work, reproductive, and
lead exposure history. A home visit was also
conducted and included a visual inspection and a
colorimetric swab test; consumable items suspected
to contain lead were sent to a laboratory for analysis.
Thirteen women (39%) reported pica behavior
during their current pregnancies. Of these, 10
reported eating soil, dirt, or clay; 2 reported
pulverizing and eating pottery; and 1 reported eating
soap. One of the women reported eating
approximately one quart of dirt daily from her
backyard for the past three months. No other
quantity data were reported.
5.3.4.20./.My/e et al. (2012)—A Soil Ingestion
Pilot Study of a Population Following a
Traditional Lifestyle Typical of Rural or
Wilderness Areas
Doyle et al. (2012) conducted a pilot study to
estimate soil ingestion among a Canadian Aboriginal
community living in a wilderness area. The study
was conducted over a 3-week period during 2011 in
the Nemiah Valley of British Columbia. The study
was conducted on traditional lands in cooperation
with the Xeni Gwet'in First Nation. Seven adults
were recruited, and four of these adults were
members of the Xeni Gwet'in community. During
the study, the subjects participated in traditional
September 2017
Page 5-28

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
daily activities (e.g., fishing, clearing, scouting,
hiking, cutting firewood, etc.). Soil ingestion was
estimated using a mass balance tracer methodology.
The tracers used for this study were aluminum,
barium, cerium, lanthanum, manganese, silicon,
thorium, titanium, uranium, vanadium, yttrium, and
zirconium. Daily soil ingestion was calculated for
each subject based on tracer concentrations in fecal
and soil samples, assuming a 24-hour transit time.
Soil ingestion rates were corrected for tracers
ingested via the diet and in medications. Participants
recorded their food and medicine intake in daily
logs. Soil, food, water, and fecal samples were
analyzed at a commercial laboratory.
Soil ingestion rates calculated for the four most
reliable tracers (aluminum, cerium, lanthanum,
silicon) are shown in Table 5-23. The most reliable
tracers were considered those that are poorly
absorbed in the gastrointestinal tract and those with
the lowest food: soil ratios. For all four tracers
combined, the mean, median, and 90th percentile soil
ingestion rates were 75, 50, and 211 mg/day,
respectively. This study provides quantitative soil
ingestion estimates for an adult Canadian population
following a traditional wilderness lifestyle.
However, as a pilot study, the sample population was
very small, and is not expected to be representative
of the U.S. population.
5.3.4.21.Irvine et aL (2014)—Soil Ingestion Rate
Determination in a Rural Population in
Alberta, Canada Practicing a Wilderness
Lifestyle
Irvine et al. (2014) estimated soil ingestion rates
among a group of First Nations people inhabiting a
wilderness area of Alberta, Canada using the mass
balance tracer method. The study was conducted
over a 3-week period in August 2012. Nine adults
within the Cold Lake First Nations Reserve who
practiced traditional activities (e.g., hunting, fishing,
and gathering) participated in the study. The tracers
used for this study were aluminum, barium, cerium,
lanthanum, manganese, silicon, thorium, titanium,
uranium, vanadium, yttrium, and zirconium. Soil
were sieved into multiple particle size fractions to
separate soil particles of <63 |im. Daily soil
ingestion was calculated for each subject based on
tracer concentrations in fecal and soil samples. A
24-hour transit time was assumed, and soil ingestion
rates were corrected for tracers ingested via the diet.
Four tracers were considered to be the most reliable
(aluminum, cerium, lanthanum, and silicon) because
of their low food: soil ratios, but aluminum and
silicon had lower coefficients of variance. Table 5-24
provides the estimated soil ingestion rates based on
aluminum and silicon, and on all four of the most
reliable tracers (aluminum, cerium, lanthanum, and
silicon combined). Mean soil ingestion rates for
these tracers ranged from 32 to 68 mg/day for this
adult population; 90th percentile values ranged from
152 to 231 mg/day. This study provides quantitative
soil ingestion estimates for an adult Canadian
population following a traditional wilderness
lifestyle. However, as a pilot study, the sample
population was very small, and is not expected to be
representative of the U.S. population.
5.3.4.22.Lumish	etaL (2014)—Gestational Iron
Deficiency is Associated with Pica
Behaviors in Adolescents
Lumish et al. (2014) examined pica behavior and
iron status among 158 pregnant adolescents
(<18 years of age) receiving prenatal care at a health
clinic in Rochester, NY in 2006-2009. The women
were mostly African-American (two-thirds) and
about 25% were Hispanic. At each visit, the women
were asked whether they craved any nonfood items,
and were asked to provide detailed information on
the items that they craved or ate. A total of 18
different items were reported to have been ingested,
including: ice; raw starches (flour and cornstarch);
powder (dust, vacuum powder from vacuum cleaner
bags, and baby powder); soap (soap, bar soap,
laundry soap, and powdered cleansers); plastic/foam
(stuffing from pillows/sofas and sponges); paper
(writing paper, toilet paper, and tissues); baking
soda/powder; and other (dirt and chalk). Overall,
46% of the teens reported ingesting one or more of
these items. Ice was the nonfood item most often
consumed (37% of all the pregnant adolescents),
while only 1.3% of the teens reported ingestion of
dirt/chalk. Lumish et al. (2014) also observed that a
significantly larger proportion of African-American
teens reported pica behavior than Caucasian teens.
This study provides prevalence data for a population
of pregnant adolescents. These data may not be
entirely representative of the U.S. population as a
whole or of pregnant women in general. Also, the
study provides prevalence data only. No information
on the quantity of substance ingested is provided.
5.3.4.23.	Jang et aL (2014)—General Factors of
the Korean Exposure Factors Handbook
Jang et al. (2014) reported on a soil ingestion
study that was conducted in Seoul, South Korea.
Feces samples were collected from 63 children, ages
0 to 7 years and analyzed for the following tracer
elements: aluminum, barium, manganese, silicon,
September 2017
Page 5-29

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
titanium, vanadium, yttrium, and zirconium. Soil
samples collected from the areas where the children
spent most of their time were analyzed for the same
tracer elements. Five children who spent no time
outdoors were used as controls to account for
exposures to tracer elements from sources other than
soil (e.g., food, medicine). Using the LTM,
aluminum was used to estimate soil ingestion rates.
Jang et al. (2014) reported that, after adjusting for the
environmental background levels based on the
control group, the estimated arithmetic mean soil
ingestion rate was 118 mg/day, and the geometric
mean was 29 mg/day. The 90th percentile value was
286 mg/day and the 95th percentile value was 898
mg/day. This study provides quantitative data for a
study population in South Korea. While it does not
appear that tracer elements in food or medicines
were accounted for, a control population was used to
adjust for exposures from sources other than soil. A
limitation of this study is that the study population
may not be representative of children in the United
States.
5.3.4.24. Chien etal. (2015)—Soil Ingestion Rates
for Children under 3 Years Old in
Taiwan
Chien et al. (2015) conducted a soil ingestion
tracer study of 66 children (33 boys and 33 girls)
under 3 years of age recruited from health centers
from northern, central, southern, and eastern Taiwan
from May 2011 to November 2012. Duplicate
24-hour food and liquid samples were collected for
each child on Day one. All feces were collected
beginning on Day 2 through Day 4 of the study.
Diapers were collected for those children who wore
diapers and feces were removed in the laboratory.
Urine samples were not collected. Outdoor soil
samples from the top 0-5 cm of soil were collected
from typical play areas. Household dust samples
were collected from the living room, bedroom, and
kitchen floors using a handheld vacuum cleaner.
Both soil and household dust samples were sieved
using a 100-mesh (150 |im: ASTM, 2017) screen.
Samples were analyzed for titanium and silicon. Soil
ingestion rates were calculated using Equation 5-7
adapted from Davis and Mirick (2006).
Sije =
V Esail '
(Eqn. 5-8)
where:
Si,e = amount of soil ingested (mg/day)
. I f = daily concentration of tracer
element in the feces (|ig/day )
Afood = daily concentration of tracer
element in the foods (|ig/day)
ESoii = concentration of tracer elements in
soils (g/kg)
The following equation was used to adjust the
daily fecal tracer concentration to account for
missing fecal samples.
Af — | Cf 2 x
DWfj 2
DWf y + DWf 3 + DWt
¦ + C,
vf,2
X
DW,
Vf, 4
'/,3
/,3
DWfi 2 + DWft + DWf
¦ + c,
/,3
Vf. 4
¦f, 4
DWi
f, 4
DWf_2 + DWf3 + DWf 4
x DWf ave
(Eqn. 5-9)
where:
Cf2
Cf.3
Cf4
DWf,2
DWf,3
DWf 4
DWfave
= concentration of tracer element in
the feces on Day 2 (|ig/day)
= concentration of tracer element in
the feces on Day 3 (|ig/day)
= concentration of tracer element in
the feces on Day 4 (|ig/day)
= dry weight of the feces on Day 2 (g)
= dry weight of the feces on Day 3 (g)
= dry weight of the feces on Day 4 (g)
= average dry weight of the feces on
Day 2 through Day 4 (g)
Activity pattern data were also collected by
videotaping each child for 2 hours and by
administering a questionnaire to parents or
caregivers. Most of the children spent more than four
times a week outdoors. About half of the children
spent under one hour a day outdoors. Children
washed their hands and took a bath or shower an
average of 3.98 and 1.24 times a day, respectively.
Recoveries for the analyses were 100 ± 20%. The
soil intake rates ranged from 0-82.6 mg/day for
silicon and 36.4—1,850 mg/day for titanium, with an
average of 9.6 mg/day (SD = 19.2 mg/day) and
957.1 mg/day (SD = 477 mg/day) for silicon and
titanium, respectively (Chien et al., 2015). These
estimates excluded children with soil intake rates
that were considered outliers. Negative soil intake
rate values were replaced by 0 mg/day in the
September 2017
Page 5-30

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
estimation of the average soil intake rates. Soil
intake rates based on silicon were positively
correlated with the total hand-to-mouth frequency
for both indoors and outdoors. Soil intake rates for
children with hand-to-mouth contact >15 contacts/hr
was estimated to be 15.9 mg/day compared to
5.46 mg/day for children with <15 contacts/hour.
Soil intake rates were also statistically significantly
greater for children with higher hand washing
frequency (>4 events/day, mean =17.4 mg/day)
compared to children with less hand washing
frequency (<4 events/day, mean =5.6 mg/day).
While not suggested as an explanation for this
observation by Chien et al. (2015), this may due to
re-loading of soil on hands and subsequent ingestion.
Chien et al. (2015) found no significant difference
between children's age and gender and the estimated
soil intake rates. The authors stated that titanium
may not be a suitable tracer for estimating soil intake
rates in the study because other sources of titanium
were not considered, including: peeling paint chips,
toothpaste, and candy. They also noted that soil
intake rates estimated in this study are lower than
those in the United States. This may be due to several
factors. The authors stated that dust loadings in the
Taiwanese homes are lower than those in U.S.
homes. As in most other Asian countries, the
Taiwanese remove their shoes before entering the
house, therefore, reducing the amount of household
dust. The authors also found that the cleaning
frequency in the study homes was higher than those
found in studies performed in the United States. This
study found correlations between soil intake rates
and hand-to-mouth frequency behavior. The results,
however, may not be representative of children's
behavior in the United States. Another limitation is
that the sources of tracer elements, other than food,
were not accounted for in estimating of soil intake
rates.
5.3.4.25. Wang et aL (2015)—Quantification of
Soil/Dust (SD) on the Hands of Children
from Hubei Province, China Using Hand
Wipes
Wang et al. (2015) estimated soil and dust
ingestion for three age ranges of children:
kindergarten (ages 3 to <7 years), primary school (7
to <13 years), and middle school (13 to <17 years)
in Hubei Province in Central China. Hand wipe
samples and background area soil samples were both
analyzed for three tracer elements (cerium,
vanadium, and yttrium). Age-specific estimates of
soil/dust adhering to the hands (in mg) were
calculated as the amount of tracer element on the
hand (in ng) divided by the amount of tracer element
in the soil (in mg/kg). Based on the three tracer
elements, the amount of soil/dust on the hands was
estimated to range from 0.59-0.64 ng for
kindergarten-aged children, 1.42-1.64 ng for
primary school children, and 1.04-1.52 ng for
middle school children. Wang et al. (2015) used
these hand soil/dust estimates, along with
assumptions about the number of hand-to-mouth
contacts that occur over a 12-hour period, the
proportion of the hand area that contacts the mouth,
and the efficiency at which soil/dust is transferred
from the hand to the mouth to estimate soil/dust
ingestion rates. Hand-to-mouth contact was assumed
to occur 15, 7, and 2 times per hour for kindergarten,
primary, and middle school children, respectively.
The proportion of the hand that contacts the mouth
was assumed to be 0.1 (10%), and the transfer
efficiency was assumed to be 0.159. The mean rates
of hand soil/dust ingestion were estimated to be 1.79,
2.12, and 0.49 mg/day for kindergarten, primary, and
middle school children, respectively. This study
provides a novel method for estimating the amount
of hand soil/dust ingested. However, these soil/dust
ingestion estimates do not account for soil/dust
ingested from other pathways (e.g., object-to-mouth
contact, food-to-soil contact), and the population
evaluated in this study may not be representative of
the U.S. population. Additional limitations relate to
the uncertainties associated with the input
parameters for estimating soil/dust intake (e.g.,
transfer efficiency, proportion of the hand contacting
the mouth).
5.3.4.26.et al. (2015)—Pica during Pregnancy
among Mexican-Born Women: A
Formative Study
Lin et al. (2015) formed nine focus groups
involving 76 Mexican-born pregnant or <2-year
postpartum women. Three of the focus groups were
in central California ( Y = 23), and six were held in
Mexico (TV =53). The aim of the focus group
discussions was to obtain information on the
frequency and types of pica behavior among the
women. Pica was defined as "eating items that were
not food." Among the women in the California focus
group, 10 (43%) indicated that they had engaged in
pica, and 6 (60%) of these women reported pica
behavior during pregnancy. Among the Mexican
focus group, 18 (34%) reported engaging in pica,
and 16 (80%) of these women reported engaging in
pica during pregnancy. Commonly eaten items were
earth, bean stones, and adobe. Among the California
group, 5 (22%) had eaten earth. Among the Mexican
September 2017
Page 5-31

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
group, 6 (11%) had eaten earth. Some women
discussed their pica behavior in the context of
micronutrient deficiencies, or perceived health
consequences of unfulfilled cravings. This study
provides information on the pica behavior for
Mexican-born women living in the United States and
Mexico. However, this population may not be
representative of the general population of the
United States. Also, the study provides prevalence
data only. Data on the quantity of pica substances
ingested are not provided.
5.3.4.27 .Ma et al. (2016)—Estimation of the Daily
Soil/Dust (SD) Ingestion Rate of Children
from Gansu Province, China via Hand-to-
Mouth Contact Using Tracer Elements
Ma et al. (2016) estimated soil and dust ingestion
among 60 children between the ages of 3 and
12 years in the Gansu Province of China. An activity
pattern modeling approach was used to estimate the
soil and dust ingestion from hand-to-mouth contact
using the following equation:
IRhandSD ~ TAhandSD x TE X SAC X EF
(Eqn. 5-10)
where:
IRhandSD = daily ingestion of soil and dust from
hands (mg/day)
TAhandso = theoretical amount of hand soil and
dust (mg)
TE = transfer efficiency at each contact
SAC = proportion of the hand surface area
contacting the mouth at each event
EF = frequency of contact in a day (day 1)
The amount of soil on the hands was estimated
by collecting two hand wipes from each child (one
in the morning and one in the afternoon), and
analyzing them for three tracer elements (cerium,
yttrium, and vanadium). Background levels of these
tracers were also measured in soil samples from the
Gansu Province. Using these data, the amount of soil
and dust on the children's hands (7/1 si j) was
estimated as follows:
TAsd Ctracer-hands / Ctracer-soil	(Eqn. 5-11)
where:
Ctracer-hands = the amount of tracer on the
hands (ng)
Ctracer-soil = the amount of tracer in
background soil (mg/kg)
Transfer efficiency (TE) was assumed to be
0.159, based on a study by Kissel et al. (1998b), the
proportion of the hand surface contacting the mouth
per event was assumed to be 0.1, and the frequency
of contact (EF) was assumed to be 15 and
7 contacts/hour, for kindergarten and primary school
children, respectively, over a 12-hour exposure
period. The estimated mean soil and dust ingestion
rates from hand-to-mouth contact ranged from 6 to
10 mg/day and 95th percentile rates ranged from 12
to 18 mg/day.
The estimates provided by Ma et al. (2016) are
based on various assumptions related to
hand-to-mouth exposure (i.e., surface areas of the
hand contacted, transfer efficiency, the number of
contacts per day), which introduce a certain degree
of uncertainty. The amount of soil on the hands was
based on measurements of three tracers (i.e., cerium,
yttrium, and vanadium) from hand wipe samples,
which the authors believed to "more accurately
estimate the amounts of [soil and dust] on the hands"
than simply weighing the hand wipes before and
after wiping the children's hands because other
substances such as grease and other organic
ingredients are eliminated. The estimates account for
soil and dust ingestion from hand-to-mouth contact
only, and the population of children studied may not
be representative of children in the United States.
5.3.5. Relevant Studies of Secondary
Analysis and Other Relevant
Information
The following studies are classified as relevant
rather than key. This section includes studies of
secondary analysis using the three methodologies
discussed in Section 5.3.1, and data on the
prevalence of the soil ingestion behavior. The
secondary analysis literature on soil and dust
ingestion rates also gives important insights into
methodological strengths and limitations of the
studies. These methodological issues include
attempts to determine the origins of apparent
positive and negative bias in the methodologies,
including: food input/fecal output misalignment;
missed fecal samples; assumptions about the weight
of children's feces; particle sizes of, and relative
contributions of soils and dusts to total soil and dust
ingestion; and attempts to identify a "best" tracer
September 2017
Page 5-32

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
element or combination of tracer elements. The
secondary analysis literature also provides insights
into the data needs for assessing soil and dust
ingestion using activity pattern approaches. Potential
error from using short-term-study estimates for
long-term soil and dust ingestion behavior estimates
is also discussed.
5.3.5.1. Wong (1988)—The Role of
Environmental and Host Behavioral
Factors in Determining Exposure to
Infection with Ascaris lumbricoides and
Trichuris trichiura/Calabrese and
Stanek (1993)—Soil Pica: Not a Rare
Event
Calabrese and Stanek (1993) reviewed a tracer
element study conducted by Wong (1988) to
estimate the amount of soil ingested by two groups
of children. Wong (1988) studied a total of 52
children in two government institutions in Jamaica.
The younger group included 24 children with an
average age of 3.1 years (range of 0.3 to 7.5 years).
The older group included 28 children with an
average age of 7.2 years (range of 1.8 to 14 years).
One fecal sample was collected each month from
each subject over the 4-month study period. The
amount of silicon in dry feces was measured to
estimate soil ingestion.
An unspecified number of daily fecal samples
were collected from a hospital control group of
30 children with an average age of 4.8 years (range
of 0.3 to 12 years). Dry feces were observed to
contain 1.45% silicon, or 14.5 mg Si per gram of dry
feces. This quantity was used to correct measured
fecal silicon from dietary sources. Fecal silicon
quantities greater than 1.45% in the 52 studied
children were interpreted as originating from soil
ingestion.
For the 28 children in the older group, soil
ingestion was estimated to be 58 mg/day, based on
the mean minus one outlier, and 1,520 mg/day, based
on the mean of all the children. The outlier was a
child with an estimated average soil ingestion rate of
41,000 mg/day (41 g/day) over the 4 months.
Estimates of soil ingestion were higher in the
younger group of 24 children. The mean soil
ingestion of all the children was 470 ± 370 mg/day.
Due to some sample losses, of the 24 children
studied, only 15 had samples for each of the
4 months of the study. Over the entire 4-month study
period, 9 of 84 samples (or 10.5%) yielded soil
ingestion estimates in excess of 1 g/day.
Of the 52 children studied, 6 had 1-day estimates
of more than 1,000 mg/day. Table 5-25 shows the
estimated soil ingestion for these six children. The
article describes 5 of 24 (or 20.8%) in the younger
group of children as having a >1,000 mg/day
estimate on at least one of the four study days; in the
older group one child is described in this manner. A
high degree of daily variability in soil ingestion was
observed among these six children; three showed
soil pica behavior on 2, 3, and 4 days, respectively,
with the most consistent (4 out of 4 days) soil pica
child having the highest estimated soil ingestion, 3.8
to 60.7 g/day.
5.3.5.2.	Calabrese and Stanek (1992b)—What
Proportion of Household Dust is Derived
from Outdoor Soil?
Calabrese and Stanek (1992b) estimated the
amount of outdoor soil in indoor dust using
statistical modeling. The model used soil and dust
data from the 60 households that participated in the
Calabrese et al. (1989) study, by preparing scatter
plots of each tracer's concentration in soil versus
dust. Correlation analysis of the scatter plots was
performed. The scatter plots showed little evidence
of a consistent relationship between outdoor soil and
indoor dust concentrations. The model estimated the
proportion of outdoor soil in indoor dust using the
simplifying assumption that the following variables
were constants in all houses: the amount of dust
produced every day from both indoor and outdoor
sources, the proportion of indoor dust due to outdoor
soil, and the concentration of the tracer element in
dust produced from indoor sources. Using these
assumptions, the model predicted that 31.3% by
weight of indoor dust came from outdoor soil. This
model was then used to adjust the soil ingestion
estimates from Calabrese et al. (1989).
5.3.5.3.	Stanek and Calabrese (1995b)—Daily
Estimates of Soil Ingestion in Children
Stanek and Calabrese (1995b) presented a
methodology that links the physical passage of food
and fecal samples to construct daily soil ingestion
estimates from daily food and fecal trace-element
concentrations. Soil ingestion data for children
obtained from the Amherst study (Calabrese et al.,
1989) were reanalyzed by Stanek and Calabrese
(1995b). A lag period of 28 hours between food
intake and fecal output was assumed for all
respondents. Day 1 for the food sample
corresponded to the 24-hour period from midnight
on Sunday to midnight on Monday of a study week;
Day 1 of the fecal sample corresponded to the
24-hour period from noon on Monday to noon on
Tuesday. Based on these definitions, the food soil
September 2017
Page 5-33

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
equivalent was subtracted from the fecal soil
equivalent to obtain an estimate of soil ingestion for
a trace element. A daily overall ingestion estimate
was constructed for each child as the median of trace
element values remaining after tracers falling
outside of a defined range around the overall median
were excluded.
Table 5-26 presents adjusted estimates, modified
according to the input/output misalignment
correction, of mean daily soil ingestion per child
(mg/day) for the 64 study participants. The approach
adopted in this paper led to changes in ingestion
estimates from those presented in Calabrese et al.
(1989).
Estimates of children's soil ingestion projected
over a period of 365 days were derived by fitting
log-normal distributions to the overall daily soil
ingestion estimates using estimates modified
according to the input/output misalignment
correction (see Table 5-27). The estimated median
value of the 64 respondents' daily soil ingestion
averaged over a year was 75 mg/day, while the 95th
percentile was 1,751 mg/day. In developing the
365-day soil ingestion estimates, data that were
obtained over a short period of time (as is the case
with all available soil ingestion studies) were
extrapolated over a year. The 2-week study period
may not reflect variability in tracer element ingestion
over a year. This study was classified as relevant
because, while Stanek and Calabrese (1995b)
attempted to address the variability through
modeling of the long-term ingestion, new
uncertainties were introduced through the parametric
modeling of the limited subject day data.
5.3.5.4. Calabrese et ah (1996)—Methodology to
Estimate the Amount and Particle Size of
Soil Ingested by Children: Implications
for Exposure Assessment at Waste Sites
Calabrese et al. (1996) examined the hypothesis
that one cause of the variation between tracers seen
in soil ingestion studies could be related to
differences in soil tracer concentrations by particle
size. This study, published before the Calabrese et al.
(1997a) primary analysis study results, used
laboratory analytical results for the Anaconda, MT
soil's tracer concentration after it had been sieved to
a particle size of <250 |im in diameter (it was sieved
to <2 mm soil particle size in Calabrese et al.
[1997a]). The smaller particle size was examined
based on the assumption that children principally
ingest soil of small particle size adhering to
fingertips and under fingernails. For five of the
tracers used in the original study (aluminum, silicon,
titanium, yttrium, and zirconium), soil concentration
was not changed by particle size. However, the soil
concentrations of three tracers (lanthanum, cerium,
and neodymium) were increased 2- to 4-fold at the
smaller soil particle size. Soil ingestion estimates for
these three tracers were decreased by approximately
60% at the 95th percentile compared to the Calabrese
et al. (1997a) results.
5.3.5.5.	Stanek etaL (1999)—Soil Ingestion
Estimates for Children in Anaconda
Using Trace Element Concentrations in
Different Particle Size Fractions
Stanek et al. (1999) extended the findings from
Calabrese et al. (1996) by quantifying trace element
concentrations in soil based on sieving to particle
sizes of 100-250 |im and to particle sizes of 53 to
<100 |im. The earlier study (Calabrese et al., 1996)
used particle sizes of 0-2 jim and 1-250 |im. This
study used the data from soil concentrations from the
Anaconda, MT site reported by Calabrese et al.
(1997a). Results of the study indicated that soil
concentrations of aluminum, silicon, and titanium
did not increase at the two finer particle size ranges
measured. However, soil concentrations of cerium,
lanthanum, and neodymium increased by a factor of
2.5 to 4.0 in the 100-250 |im particle size range
when compared with the 0-2 |im particle size range.
There was not a significant increase in concentration
in the 53-100 ^m particle size range.
5.3.5.6.	Stanek and Calabrese (2000)—Daily Soil
Ingestion Estimates for Children at a
Superfund Site
Stanek and Calabrese (2000) reanalyzed the soil
ingestion data from the Anaconda study (Calabrese
et al., 1997a) to provide estimates of variability
between days and subjects, and daily soil ingestion
rates over a longer period. Stanek and Calabrese
(2000) attempted to address several sources of
uncertainties in this reanalysis including
identification of outliers and the estimation of soil
ingestion rates based on particle size of <250 |im.
Tracer-specific soil ingestion rates were estimated
for 8 tracer elements for up to 7 subject days for each
of the 64 children. The mean and median values were
estimated for all tracers for each day, and median and
mean estimates were generated over all subject days.
Table 5-28 summarizes these soil ingestion
estimates.
Assuming a log-normal distribution for the soil
ingestion estimates in the Anaconda study, average
long-term soil ingestion rates were predicted for
children for each of the eight trace elements. Using
September 2017
Page 5-34

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
"best linear unbiased predictors," the authors
predicted 95th percentile soil ingestion values over
time periods of 7 days, 30 days, 90 days, and
365 days. The 95th percentile soil ingestion values
were predicted to be 133 mg/day over 7 days,
112 mg/day over 30 days, 108 mg/day over 90 days,
and 106 mg/day over 365 days. Based on this
analysis, estimates of the distribution of longer term
average soil ingestion are expected to be narrower,
with the 95th percentile estimates being as much as
25% lower (Stanek and Calabrese, 2000). This study
was classified as relevant because, while Stanek and
Calabrese (2000) attempted to characterize the
variability between days for a given subject and
between subjects, the derivation of usual (long term)
intake based on limited short term subject day data
introduces new uncertainties.
5.3.5.7.	Stanek etaL (2001a)—Biasing Factors
for Simple Soil Ingestion Estimates in
Mass Balance Studies of Soil Ingestion
To identify and evaluate biasing factors for soil
ingestion estimates, the authors developed a
simulation model based on data from previous soil
ingestion studies. The soil ingestion data used in this
model were taken from Calabrese et al. (1989) (the
Amherst study), Davis et al. (1990) (southeastern
Washington State), Calabrese et al. (1997a) (the
Anaconda study), and Calabrese et al. (1997b) (soil
pica in Massachusetts), and relied only on the
aluminum and silicon trace element estimates
provided in these studies.
Of the biasing factors explored, the impact of
study duration was the most striking, with a positive
bias of more than 100% for 95th percentile estimates
in a 4-day tracer element study. A smaller bias was
observed for the impact of absorption of trace
elements from food. Although the trace elements
selected for use in these studies are believed to have
low absorption, whatever amount is not accounted
for will result in an underestimation of the soil
ingestion distribution. In these simulations, the
absorption of trace elements from food of up to 30%
was shown to negatively bias the estimated soil
ingestion distribution by less than 20 mg/day. No
biasing effect was found for misidentifying play
areas for soil sampling (i.e., ingested soil from ayard
other than the subject's yard).
5.3.5.8.	Stanek etaL (2001b)—Soil Ingestion
Distributions for Monte Carlo Risk
Assessment in Children
Stanek et al. (2001b) developed "best linear
unbiased predictors" to reduce the biasing effect of
short-term soil ingestion estimates. This study
estimated the long-term average soil ingestion
distribution using daily soil ingestion estimates from
children who participated in the Anaconda, MT
study. Trace element specific estimates on a subject
day were simulated assuming a normal distribution.
In this long-term (annual) distribution, the soil
ingestion estimates were: mean 31, median 24,
75 th percentile 42, 90th percentile 75, and
95th percentile 91 mg/day. This study was classified
as relevant because of the uncertainties introduced
by the assumptions used to derive long-term average
soil ingestion distribution. For example, the
methodology assumes that the variance of the trace
element estimates on a subject-day is identical for all
subjects and days and that the variance between days
in soil ingestion for a subject is the same for all
subjects.
5.3.5.9. Von Lindern et al. (2003)—Assessing
Remedial Effectiveness through the
Blood Lead:Soil/Dust Lead Relationship
at the Bunker Hill Superfund Site in the
Silver Valley of Idaho
Similar to Hogan et al. (1998), Von Lindern et al.
(2003) used the IEUBK model to predict blood lead
levels in a nonrandom sample of several hundred
children ages 0-9 years in an area of northern Idaho
from 1989-1998 during community-wide soil
remediation. Von Lindern et al. (2003) used the
IEUBK default soil and dust ingestion rates together
with observed house dust/soil lead levels (and
imputed values based on community soil and dust
lead levels, when observations were missing). The
authors compared the predicted blood lead levels
with observed blood lead levels and found that the
default IEUBK soil and dust ingestion rates and lead
bioavailability value over-predicted blood lead
levels, with the over-prediction decreasing as the
community soil remediation progressed. The authors
stated that the over-prediction may have been caused
either by a default soil and dust ingestion that was
too high, a default bioavailability value for lead that
was too high, or some combination of the two. They
also noted under-predictions for some children, for
whom follow up interviews revealed exposures to
lead sources not accounted for in the model
simulations (e.g., recreational exposures from
outside the site). In addition, some of these children
were socioeconomically disadvantaged, highly
mobile, and cared for at multiple locations.
Von Lindern et al. (2003) developed a statistical
model that apportioned the contributions of
community soils, yard soils of the residence, and
September 2017
Page 5-35

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
house dust to lead intake; the models' results
suggested that community soils contributed more
(50%) than neighborhood soils (28%) or yard soils
(22%) to soil found in house dust of the studied
children.
5.3.5.10.	Layton and Beamer (2009)—Migration
of Contaminated Soil and Airborne
Particulates to Indoor Dust
Layton and Beamer (2009) developed a
modeling and measurement framework to assess the
transport of contaminated soils and airborne
particulates indoors at a residence from outdoor
sources. The model accounted for the resuspension
and deposition processes, and the removal of
particles by cleaning and exhalation. The model
consisted of a two-compartment model (i.e., air and
floor) that simulated the dust fall, accumulation, and
loss of particles on floor surfaces using the
concentration of an inorganic tracer in those two
media. Monitoring data from the National Human
Exposure Assessment Survey (NHEXAS) in six
Midwestern states and a city in The Netherlands
were used to characterize the transport parameters of
the model. The analysis assumed a lognormal
distribution of soil track-in indoors with a geometric
mean of 0.1 g/day (GSD = 3) based on data from the
published literature. Results of the model indicated
that 60% of the arsenic indoors originated from
ambient air, while 40% was a result of soil
tracked-in.
5.3.5.11.	Gavrelis et ah (2011)—An Analysis of the
Proportion of the U.S. Population That
Ingests Soil or Other Non-Food
Substances
Gavrelis et al. (2011) evaluated the prevalence of
the U.S. population that ingests nonfood substances
such as soil, clay, starch, paint, or plaster. Data were
compiled from the National Health and Nutrition
Examination Survey (NHANES) collected from
1971 to 1975 (NHANES I) and 1976 to 1980
(NHANES II), which represent a complex, stratified,
multistage, probability-cluster design, and include
nationwide probability samples of approximately
21,000 and 25,000 study participants, respectively.
NHANES I surveyed people aged 1 to 74 years and
NHANES II surveyed those 6 months to 74 years.
The study population included women of
childbearing age, people with low income status, the
elderly, and preschool children, who represented an
oversampling of specific groups in the population
that were believed to have high risks for
malnutrition.
The survey questions were demographic,
socioeconomic, dietary, and health-related queries,
and included specific questions regarding soil and
nonfood substance ingestion. Survey questions for
children under 12 years asked whether they
consumed nonfood substances including dirt or clay,
starch, paint or plaster, and other materials
(NHANES I) or about consumption of clay, starch,
paint or plaster, dirt, and other materials
(NHANES II). For participants over 12 years of age,
the survey questions asked only about consumption
of dirt or clay, starch, and other materials
(NHANES I) or about nonfood substances including
clay, starch, and other materials (NHANES II).
Age groupings used in this analysis vary slightly
from the age group categories established by EPA
and described in Guidance on Selecting Age Groups
for Monitoring and Assessing Childhood Exposures
to Environmental Contaminants (U.S. EPA, 2005).
Other demographic parameters included sex
(including pregnant and nonpregnant females), race
(White, Black, and other), geography (urban and
rural, with "urban" defined as populations >2,500),
income level (ranging from $0-$9,999 up
to >$20,000, or not stated), and highest grade head
of household (population under 18 years) or
respondent (population >18 years) attended. For
statistical analysis, frequency estimates were
generated for the proportion of the total U.S.
population that reported consumption of dirt, clay,
starch, paint or plaster, or other materials
"considered unusual" using the appropriate NCHS
sampling weights and responses to the relevant
questions in NHANES I and II. NHANES I and II
were evaluated separately because the data sets did
not provide components of the weight variable
separately (i.e., probability of selection, nonresponse
adjustment weight, and poststratification weight).
Although the overall prevalence estimates were
higher in NHANES I compared with NHANES II,
similar patterns were generally observed across
substance types and demographic groups studied.
For NHANES I, the estimated prevalence of all
nonfood substance consumption in the United States
for all ages combined was 2.5% (95% confidence
interval [CI]: 2.2-2.9%), whereas for NHANES II,
the estimated prevalence of all nonfood substance
consumption in the United States for all ages
combined was 1.1%(95%CI: 1.0-1.2%). Table 5-29
provides the prevalence estimates by type of
substance consumed for all ages combined. By type
of substance, the estimated prevalence was greatest
for dirt and clay consumption and lowest for starch.
Figures 5-1, 5-2, and 5-3, respectively, show the
prevalence of nonfood substance consumption by
September 2017
Page 5-36

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
age, race, and income. The most notable differences
were seen across age, race (Black vs. White), and
income groups. For both NHANES I and II,
prevalence for the ingestion of all nonfood
substances decreased with increasing age, was
higher among Blacks (5.7%; 95% CI: 4.4-7.0%) as
compared to Whites (2.1%; 95% CI: 1.8-2.5%), and
was inversely related to income level, with
prevalence of nonfood consumption decreasing as
household income increased. The estimated
prevalence of all nonfood substances for the 1- to
<3-year age category was at least twice that of the
next oldest category (3 to <6 years). Prevalence
estimates were 22.7% (95% CI: 20.1-25.3%) for the
1- to <3-year age group based on NHANES I and
12% based on NHANES II. In contrast, prevalence
estimates for the >21-year age group was 0.7%
(95% CI: 0.5-1.0%) and 0.4% (95% CI: 0.3-0.5%)
for NHANES I and NHANES II, respectively. Other
differences related to geography (i.e., urban and
rural), highest grade level of the household head, and
sex were less remarkable. For NHANES I, for
example, the estimated prevalence of nonfood
substance consumption was only slightly higher
among females (2.9%; CI: 2.3-3.5%) compared to
males (2.1%; CI: 1.8-2.5%) of all ages. Forpregnant
females, prevalence estimates (2.5%;
95% CI: 0.0-5.6%) for those 12 years and over were
more than twice those for nonpregnant females
(1.0%; 95% CI: 0.7-1.4%).
5.3.5.1 l.Stanek et aL (2012a)—Meta-Analysis of
Mass-Balance Studies of Soil Ingestion
in Children
Stanek et al. (2012a) conducted a meta-analysis
of four major mass-balance soil ingestion studies in
U.S. children between 1 and 7 years of age. The four
studies included the Amherst study (Calabrese et al.,
1989),	the Washington State study (Davis et al.,
1990),	the Washington Family study (Davis and
Mirick, 2006), and the Anaconda study (Calabrese et
al., 1997a).
Data for only two trace elements were included
(Al and Si). Excluded from the study were 10% of
subjects identified as outliers. In addition, study
subjects with high soil intake (soil pica) were
excluded. Additional sources of variability and bias
between the four studies included in the
meta-analysis are discussed in Stanek et al. (2012b).
Mean, median, and 95th percentile soil ingestion
estimates based on 216 children were 25.5, 32.6, and
79.4 mg/day, respectively. Soil ingestion rates for
males and females were similar. As shown in Table
5-30, soil ingestion appeared to increase with age
with the youngest age group (1 to <2 years) having
the lowest soil ingestion rate.
This study provides age-specific soil ingestion
estimates based on an analysis of four U.S.
mass-balance soil ingestion studies for children.
However, this study was classified as relevant
because high-end values that were considered to be
biased were excluded. In addition, data for a child
with pica were also excluded. The study subjects
were all from a limited study area (northern United
States only) and data were collected only in the
summer and early fall.
5.3.5.13. Wilson et aL (2015)—Estimation of
Sediment Ingestion Rates Based on
Hand-to-Mouth Contact and Incidental
Surface Water Ingestion
Wilson et al. (2015) used a similar mechanistic
approach to that of Wilson et al. (2013) to estimate
sediment ingestion rates, except that greater
adherence of sediment to the hands was assumed
than for soil and dust. These sediment ingestion rates
were intended for use in recreational exposure
scenarios involving contact with sediments in
aquatic areas. Sediment ingestion was assumed to
occur as a result of direct contact with the sediment
and subsequent hand-to-mouth contact, as well as
incidental ingestion of surface water containing
suspended sediments. Both deterministic and
probabilistic methods were used to estimate
sediment ingestion based on the following
equations:
SDIRhM - SLhands X SAhand X FSAfi„gers X FQ X SE
(Eqn. 5-12)
and
SDIRwc =SSx SWIR	(Eqn. 5-13)
where:
FQ = frequency of hand-to-mouth events
(1/hour)
FSAfingerS = fractional surface area of the hands
(unitless)
SAhand = surface area of the hand (cm2)
SDIRhm = sediment ingestion rate from hand-
to-mouth contact (mg/hr)
SDIRwc = sediment ingestion from incidental
water consumption (mg/hr)
SE = saliva extraction factor (unitless)
September 2017
Page 5-37

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
SLhands = sediment adherence on hands
(mg/cm2)
SS = suspended sediment concentration
(mg/L)
SWIR = surface water ingestion rate (L/hr)
Table 5-31 presents the age-dependent input
values used in these calculations. Both deterministic
and probabilistic methods were used to estimate
sediment ingestion rates. The results based on the
probabilistic assessment are presented in Table 5-32.
The estimated sediment ingestion rates based on
hand-to-mouth contact using the probabilistic
approach were 72 mg/hour for toddlers 7-months to
4 years old, 57 mg/hour for children 5 to 11 years
old, 18 mg/hour for teens 12 to 19 years old, and
20 mg/hour for adults and seniors ages 20 to
60+years. The sediment ingestion rate based on
incidental surface water ingestion was 7.7 mg/hour
for all age groups. Slightly lower values were
observed using the deterministic approach.
This study provides sediment ingestion rates in
units of mg/hour which can be used to estimate
exposure to contaminants in sediment. Its limitations
are based on the uncertainties associated with the
input parameters used to model dust ingestion. Also,
as indicated by the author, the age groups and
receptor characteristics were intended to represent
the Canadian population, which may or may not be
representative of U.S. populations.
5.3.5.14. Wilson et aL (2016)—Estimation of Dust
Ingestion Rates in Units of Surface Area
per Day Using a Mechanistic Hand-to-
Mouth Model
Wilson et al. (2016) estimated dust ingestion
rates for various age groups of Canadian children on
the basis of surface area (m2/day). These dust
ingestion rates are intended to be used with
measured contaminant loadings in surface dust
(|ig/m2) to estimate exposure from ingestion of
surface dust. Wilson et al. (2016) used a similar
approach and input parameters as used in Wilson et
al. (2013). Dust ingestion was calculated using the
following equation:
DIG = FTSS x SAhand x FSAfmgers x FQ x-SEx-ET
(Eqn. 5-14)
where:
DIG = dust ingestion rate (m2/d)
ET = exposure time (h/d)
FQ = frequency of hand-to-mouth events
(events/hr)
FSAfi„gers = fractional surface area of hand
mouthed
FTSS = fraction of dust transferred from
indoor surfaces to hands
SAhand = surface area of one hand (cm2)
SE = saliva extraction factor (unitless)
The age-dependent parameters used were the
same as those used in Wilson et al. (2013), as shown
in Table 5-13. Age-specific exposure time (ET)
values were estimated as 24 hours/day minus the
time spent sleeping and the time spent outdoors. It
was assumed that contact occurred with hard
surfaces (e.g., carpets, countertops) 50% of the time
and with soft surfaces (e.g., carpets, sofas) 50% of
the time, except for infants for whom contact was
assumed to occur with soft surfaces only (i.e., 100%
of the time). Both deterministic and probabilistic
methods were used to estimate dust ingestion rates.
Results based on the probabilistic approach are
presented in Table 5-33. Dust ingestion rates for
children <11 years old were estimated to range from
0.025 m2/day for infants 0-6 months to 0.061 m2/day
for toddlers 7 months to 4 years.
This study provides dust ingestion rates in units
of m2/day which can be used with surface loadings
of dust (and corresponding contaminant
concentrations in the dust) in |ig/nr to estimate
exposure. Its limitations are based on the
uncertainties associated with the input parameters
used to model dust ingestion. The use of these dust
ingestion rates assumes that the surface loading
measurements are representative of the surfaces
contacted.
5.3.5.15.Fawcett et aL (2016)—A Meta-Analysis
of the Worldwide Prevalence of Pica
during Pregnancy and the Postpartum
Period
Fawcett et al. (2016) conducted a meta-analysis
using information found in 70 studies published
through February 2014 to develop worldwide pica
prevalence estimates among pregnant and
postpartum populations. Fawcett et al. (2016) also
characterized variations in prevalence based on
moderating variables (e.g., educational level,
geographic region, and ethnicity). Of the studies
evaluated, 33.8% were from North America, 5.6%
were from South America, 33.8% were from Africa,
18.3% were from the Middle East, 4.2% were from
September 2017
Page 5-38

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
Asia, and 4.2% were from Europe. Fawcett et al.
(2016) defined pica as "the purposeful consumption
of nonfood or nonnutritive substances," including
"earth (geophagia), starch (amylophagia), ice
(pagophagia), and a vast number of additional
substances (e.g., baking soda)." Postpartum was
defined as up to 12 months after delivery. The
overall prevalence of pica was estimated to be
27.8%, but Fawcett et al. (2016) suggested that this
is a poor indicator of pica prevalence among the
general population of pregnant and postpartum
because of the heterogeneity across studies. African
women had the highest prevalence of pica at 44.8%,
followed by North/South American at 23.0%, and
Eurasia at 17.5%. Based on nine U.S. studies that
included data on ethnicity, African-American
women in the United States were also more likely to
experience pica than non-African-American women.
Based on a subset of 29 studies that reported the
educational level of the women, education had a
negative association with pica prevalence (i.e.,
higher educational level is associated with lower
prevalence of pica). A subset of 31 studies reported
on pica behavior among women with anemia. These
women were one-and-a-half times more likely to
report pica than nonanemic women.
This study estimated pica based on a wide range
of studies, and provides information on the factors
that may affect pica prevalence among pregnant and
postpartum women. However, as stated by the
author, the overall pica prevalence of 27.8% is not a
good indicator of prevalence among the general
population due to the wide range of values observed
in the various studies evaluated. Also, pica was
defined as ingestion of nonfood substances and did
not represent soil/dust ingestion alone, but Fawcett
et al. (2016) indicated that the moderating variables
for geophagia did not differ from those of overall
pica.
5.4. LIMITATIONS OF STUDY
METHODOLOGIES
The three types of information needed to provide
recommendations to exposure assessors on soil and
dust ingestion rates among U.S. children include
quantities of soil and dust ingested, frequency of
high soil and dust ingestion episodes, and prevalence
of high soil and dust ingesters. The methodologies
provide different types of information: The tracer
element, biokinetic model comparison, and activity
pattern methodologies provide information on
quantities of soil and dust ingested; the tracer
element methodology provides limited evidence of
the frequency of high soil ingestion episodes; and the
survey response methodology can shed light on
prevalence of mouthing behavior and frequency of
high soil ingestion episodes. The biokinetic model
comparison methodology has the advantage of
reflecting longer term exposures. However, the
methodologies used to estimate soil and dust
ingestion rates and prevalence of soil and dust
ingestion behaviors have certain limitations when
used for the purpose of developing recommended
soil and dust ingestion rates. These limitations may
not have excluded specific studies from use in the
development of recommended ingestion rates, but
have been noted throughout this handbook. This
section describes some of the known limitations,
presents an evaluation of the current state of the
science for U.S. children's soil and dust ingestion
rates, and describes how the limitations affect the
confidence ratings given to the recommendations.
5.4.1. Tracer Element Methodology
This section describes some previously identified
limitations of the tracer element methodology as it
has been implemented by U.S. researchers, as well
as additional potential limitations that have not been
explored. Some of these same limitations would also
apply to the Dutch and Jamaican studies that used a
control group of hospitalized children to account for
dietary and pharmaceutical tracer intakes.
Binder et al. (1986) described some of the major
and obvious limitations of the early U.S. tracer
element methodology as follows:
[T]he algorithm assumes that children
ingest predominantly soil from their own
yards and that concentrations of elements in
composite soil samples from front and back
yards are representative of overall
concentrations in the yards...children
probably eat a combination of soil and dust;
the algorithm used does not distinguish
between soil and dust ingestion....fecal
sample weights...were much lower than
expected...the assumption that aluminum,
silicon and titanium are not absorbed is not
entirely true...dietary intake of aluminum,
silicon and titanium is not negligible when
compared with the potential intake of these
elements from soil...Before accepting these
estimates as true values of soil ingestion in
toddlers, we need a better understanding of
the metabolisms of aluminum, silicon and
titanium in children, and the validity of the
September 2017
Page 5-39

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
assumptions we made in our calculations
should be explored further.
The subsequent U.S. tracer element studies
(Calabrese et al., 1997a, 1989; Barnes, 1990; Davis
et al., 1990; Davis and Mirick, 2006) made some
progress in addressing some of the Binder et al.
(1986) study's stated limitations.
Regarding the issue of nonyard
(community-wide) soil as a source of ingested soil,
one study (Calabrese et al., 1989; Barnes, 1990)
addressed this issue to some extent, by including
samples of children's daycare center soil in the
analysis. Calabrese et al. (1997a) attempted to
address the issue by excluding children in daycare
from the study sample frame. Homogeneity of
community soils' tracer element content would play
a role in whether this issue is an important biasing
factor for the tracer element studies' estimates. Davis
et al. (1990) evaluated community soils' aluminum,
silicon, and titanium content and found little
variation among 101 yards throughout the three-city
area. Stanek et al. (2001a) concluded that there was
"minimal impact" on estimates of soil ingestion due
to mis-specifying a child's play area.
Regarding the issue of soil and dust both
contributing to measured tracer element quantities in
excreta samples, the key U.S. tracer element studies
all attempted to address the issue by including
samples of household dust in the analysis, and in
some cases estimates are presented in the published
articles that adjust soil ingestion estimates on the
basis of the measured tracer elements found in the
household dust. The relationship between soil
ingestion rates and indoor settled dust ingestion rates
has been evaluated in some of the secondary studies
(Calabrese and Stanek, 1992b). An issue similar to
the community-wide soil exposures in the previous
paragraph could also exist with community-wide
indoor dust exposures (such as dust found in schools
and community buildings occupied by study subjects
during or prior to the study period). A portion of the
community-wide indoor dust exposures (due to
occupying daycare facilities) was addressed in the
Calabrese et al. (1989) and Barnes (1990) studies,
but not in the other three key tracer element studies.
In addition, if the key studies' vacuum cleaner
collection method for household and daycare indoor
settled dust samples influenced tracer element
composition of indoor settled dust samples, the dust
sample collection method would be another area of
uncertainty with the key studies' indoor dust-related
estimates. The survey response studies suggest that
some young children may prefer ingesting dust to
ingesting soil. The existing literature on soil versus
dust sources of children's lead exposure may provide
useful information that has not yet been compiled for
use in soil and dust ingestion recommendations.
Regarding the issue of fecal sample weights and
the related issue of missing fecal and urine samples,
the key tracer element studies have varying strengths
and limitations. The Calabrese et al. (1989) article
stated that wipes and toilet paper were not collected
by the researchers, and thus fecal quantities, and soil
and dust ingestion may have been underestimated.
Calabrese et al. (1989) stated that cotton cloth
diapers were supplied for use during the study;
commodes apparently were used to collect both
feces and urine for those children who were not
using diapers. Barnes (1990) described cellulose and
polyester disposable diapers with significant
variability in silicon and titanium content and
suggested that children's urine was not included in
the analysis. Thus, it is unclear to what extent
complete fecal and urine output was obtained for
each study subject. The Calabrese et al. (1997a)
study did not describe missing fecal samples and did
not state whether urinary tracer element quantities
were used in the soil and dust ingestion estimates,
but stated that wipes and toilet paper were not
collected. Missing fecal samples may have resulted
in negative bias in the estimates from both of these
studies. Davis et al. (1990) and Davis and Mirick
(2006) were limited to children who no longer wore
diapers. The authors made adjustments to the soil
and dust ingestion estimates based on assumptions
regarding the quantities of feces and urine in missed
samples. These adjustments may have affected those
studies' estimates, but the direction of the bias is
uncertain. Adjustments for missing fecal and urine
samples could introduce errors sufficient to cause
negative estimates if missed samples were heavier
than the collected samples used in the soil and dust
ingestion estimate calculations.
Regarding the issue of dietary intake, the key
U.S. tracer element studies have all addressed
dietary (and nondietary, nonsoil) intake by
subtracting calculated estimates of these sources of
tracer elements from excreta tracer element
quantities, or by providing study subjects with
personal hygiene products that were low in tracer
element content. Applying the food and nondietary,
nonsoil corrections required subtracting the tracer
element contributions from these nonsoil sources
from the measured fecal/urine tracer element
quantities. To perform this correction required
assumptions to be made regarding the
gastrointestinal transit time, or the time lag between
inputs (food, nondietary nonsoil, and soil) and
outputs (fecal and urine). The gastrointestinal transit
September 2017
Page 5-40

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
time assumption introduced a new potential source
of bias that some authors (Stanek and Calabrese,
1995b) called input/output misalignment or transit
time error. Stanek and Calabrese (1995a) attempted
to correct for this transit time error by using the BTM
and focusing estimates on those tracers that had low
food: soil tracer concentration ratios. The lag time
may also be a function of age. Davis et al. (1990) and
Davis and Mirick (2006) assumed a 24-hour lag time
in contrast to the 28-hour lag times used in Calabrese
et al. (1989), Barnes (1990), and Calabrese et al.
(1997a). ICRP (2002) suggested a lag time of
37 hours for 1-year-old children and 5- to 15-year-
old children. Stanek and Calabrese (1995b) describe
a method designed to reduce bias from this error
source. Lag times that are shorten or longer than
what is assumed may result in misalignment of the
intake of tracers and their output. The direction of
these biases is difficult to predict.
Regarding gastrointestinal absorption, the
authors of three of the studies appeared to agree that
the presence of silicon in urine represented evidence
that silicon was being absorbed from the
gastrointestinal tract (Davis et al., 1990; Calabrese et
al., 1989; Barnes, 1990; Davis and Mirick, 2006).
There was some evidence of aluminum absorption in
Calabrese etal. (1989) and Barnes (1990); Davis and
Mirick (2006) stated that aluminum and titanium did
not appear to have been absorbed, based on low
urinary levels. Davis et al. (1990) stated that silicon
appears to have been absorbed to a greater degree
than aluminum and titanium, based on urine
concentrations. Absorption of these tracer elements
would likely bias the estimated soil and dust
ingestion rates low, because intake of the tracer
elements would not be accounted for in the excretia
if they had been absorbed. More research to better
understand the uptake of these tracer elements may
be warranted.
Aside from the gastrointestinal absorption, lag
time, and missed fecal sample issues, Davis and
Mirick (2006) offered another possible explanation
for the negative soil and dust ingestion rates
estimated for some study participants. Negative
values result when the tracer amount in food and
medicine is greater than that in urine/fecal matter.
Given that some analytical error may occur, any
overestimation of tracer amounts in the food samples
would be greater than an overestimation in
urine/feces because the food samples were many
times heavier than the urine and fecal samples.
Overestimating the amount of tracers ingested via
food and medicines would tend to negatively bias
soil and dust ingestion estimates while
underestimating the amount of tracers ingested via
foods and medicines would tend to positively bias
soil and dust ingestion estimates. Interactions among
elements in the gastrointestinal tract, in particular
fluoride from ingested toothpaste, and how these
may affect the absorption of tracer elements are not
available based on the studies reviewed for this
chapter.
Another limitation on the accuracy of tracer
element-based estimates of soil and dust ingestion
relates to inaccuracies inherent in environmental
sampling and laboratory analytical techniques. The
"percent recovery" of different tracer elements
varies (according to validation of the study
methodology performed with adults who swallowed
gelatin capsules with known quantities of sterilized
soil, as part of the Calabrese et al. [1989, 1997a]
studies). Digestion/extraction efficiencies also vary
by media type (e.g., food, soil, medicine). Estimates
based on a particular tracer element with a lower or
higher recovery than the expected 100% in any of the
study samples would be influenced in either a
positive or negative direction, depending on the
recoveries in the various samples and their degree of
deviation from 100% (Calabrese et al., 1989).
Soil/dust size fractions, and digestion/extraction
methods of sample analysis may be additional
limitations.
Davis et al. (1990) offered an assessment of the
impact of swallowed toothpaste on the tracer-based
estimates by adjusting estimates for those children
whose caregivers reported that they had swallowed
toothpaste. Davis et al. (1990) had supplied study
children with toothpaste that had been preanalyzed
for its tracer element content, but it is not known to
what extent the children actually used or swallowed
the supplied toothpaste. Similarly, Calabrese et al.
(1989, 1997a) supplied children in the Amherst, MA
and Anaconda, MT studies with toothpaste
containing low levels of most tracers, but it is unclear
to what extent those children used or swallowed the
supplied toothpaste. Not accounting for the
consumption of toothpaste or other materials that
contain the tracer elements would tend to bias soil
and dust ingestion rates in the positive direction, but
over-predicting the amount of tracers consumed in
these materials would have the opposite effect.
Other research suggests additional possible
limitations that have not yet been explored. First,
lymph tissue structures in the gastrointestinal tract
might serve as reservoirs for titanium dioxide food
additives and soil particles, which could bias
estimates either upward or downward depending on
the tracers' entrapment within, or release from, these
reservoirs during the study period (ICRP, 2002;
Shepherd et al., 1987; Powell et al., 1996). Second,
September 2017
Page 5-41

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
gastrointestinal uptake of silicon may have occurred,
which could bias those estimates downward. There
is an increasing body of evidence that silicon is an
essential nutrient and that it plays a role in the
initiation of the mineralization process and bone
formation (Chen et al., 2016; Nielsen, 2014; Price et
al., 2012; Jugdaohsingh, 2007; Carlisle, 1980).
However, absorption of silicon in the gastrointestinal
tract is not well understood (Jugdaohsingh et al.,
2002).	Van Dyck et al. (2000) suggests a possible
negative bias in the silicon-based soil ingestion
estimates, depending on the quantities of silicon
absorbed by growing children. Third, regarding the
potential for swallowed toothpaste to bias soil
ingestion estimates upward, commercially available
toothpaste may contain quantities of titanium and
perhaps silicon and aluminum in the range that could
be expected to affect the soil and dust ingestion
estimates. Fourth, for those children who drank
bottled or tap water during the study period, and did
not include those drinking water samples in their
duplicate food samples, slight upward bias may exist
in some of the estimates for those children because
drinking water may contain small, but relevant,
quantities of silicon and potentially other tracer
elements. Fifth, the tracer element studies conducted
to date have not explored the impact of soil
properties' influence on toxicant uptake or excretion
within the gastrointestinal tract. Nutrition
researchers investigating influence of clay geophagy
behavior on human nutrition have begun using in
vitro models of human digestion (Dominy et al.,
2003; Hooda et al., 2004). A recent review (Wilson,
2003)	covers a wide range of geophagy research in
humans and various hypotheses proposed to explain
soil ingestion behaviors, with emphasis on the soil
properties of geophagy materials.
5.4.2. Biokinetic Model Comparison
Methodology
It is possible that the IEUBK biokinetic model
comparison methodology contained sources of both
positive and negative bias, like the tracer element
studies, and that the net impact of the competing
biases is not known. There may be several
significant sources of bias with the biokinetic model
comparison methodology. One source of potential
bias was the possibility that the biokinetic model
cannot account for sources of lead exposure that are
important for certain children due to incomplete
exposure characterizations. For these children, the
model might under-predict blood lead levels
compared to actual measured lead levels. However,
this result may actually mean that the default
assumed lead intake rates via either soil and dust
ingestion, or another lead source that is accounted
for by the model, are too high. Another source of
potential bias includes not accounting for the
variability in the children's blood lead levels with
time. There are also uncertainties with regard to the
representativeness of the media concentrations in the
children's play areas. For example, there is potential
bias when predicting blood lead levels in children
who have not spent a significant amount of time in
the areas characterized as the main sources of
environmental lead exposure. Modeling this
population could result in either upward or
downward biases in predicted blood lead levels.
Comparing upward-biased predictions with actual
measured blood lead levels and finding a relatively
good match could lead to inferences that the model's
default soil and dust ingestion rates are accurate,
when in fact the children's soil and dust ingestion
rates, or some other lead source, were actually higher
than the default assumption. Von Lindern et al.
(2016) attempted to address the issue of
representativeness by assuming different partition
models (i.e., percentage of time spent at various
soil-contact locations).
Additionally, there is uncertainty with the
assumption within the model itself regarding the
biokinetics of absorbed lead, which could result in
either positively or negatively biased predictions and
the same kinds of incorrect inferences as the second
source of potential bias. Another source of bias is the
education and intervention programs implemented at
contaminated sites which can potentially result in
children's temporary reduction in soil/dust ingestion
rates (see Section 5.4.4) (von Lindern et al., 2016).
In addition, there was no extensive sensitivity
analysis. The calibration step used to fix model
parameters limits the degree that most parameters
can reasonably be varied. Second, the IEUBK model
was not designed to predict blood lead levels greater
than 25-30 ng/dL; there are few data to develop
such predictions and less to validate them. If there
are site-specific data that indicate soil ingestion rates
(or other ingestion/intake rates) are higher than the
defaults on average (not for specific children), the
site-specific data should be considered. EPA
considers the default IEUBK value of
30% bioavailability reasonable for most data
sets/sites. Bioavailability has been assayed for soils
similar to those in the calibration step and the
empirical comparison data sets; 30% was used in the
calibration step, and is therefore recommended for
similar sites. The default provides a reasonable
substitute when there are no specific data. Speciation
of lead compounds for a particular exposure scenario
September 2017
Page 5-42

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
could support adjusting bioavailability if they are
known to differ strongly from 30%. The use of the
30% bioavailability was further supported by von
Lindern et al. (2016). In general, EPA supports using
bioavailability rates determined for the particular
soils of interest if available.
5.4.3. Activity Pattern Methodology
The limitations associated with the activity
pattern methodology relate to the availability and
quality of the underlying data used to model soil
ingestion rates. Some examples where data are
limited or lacking include: information on the
activities and environments of very young children
(e.g., hand-to-mouth, object-to-mouth, and pica
behaviors), information on skin adherence and dust
loadings on indoor objects and floors (bare and
carpeted), as well as the transfer of dust from both
bare floors and carpets to hands, soil loadings to
hands while playing outdoors, and information to
evaluate temporal variability. Soil properties that
may also be important include: soil particle size,
organic content, moisture content, and other
properties that affect soil adherence to the skin.
Real-time hand recording, where observations
are made by trained professionals (rather than
parents), may offer the advantage of consistency in
interpreting visible behaviors and may be less
subjective than observations made by someone who
maintains a care-giving relationship to the child. On
the other hand, young children's behavior may be
influenced by the presence of unfamiliar people
(Davis et al., 1995). Groot et al. (1998) indicated that
parent observers perceived that deviating from their
usual care-giving behavior by observing and
recording mouthing behavior appeared to have
influenced the children's behavior. With video-
transcription methodology, an assumption is made
that the presence of the videographer or camera does
not influence the child's behavior. This assumption
may result in minimal biases introduced when
filming newborns or when the camera and
videographer are not visible to the child. However, if
the children being studied are older than newborns
and can see the camera or videographer, biases may
be introduced. Ferguson et al. (2006) described
apprehension caused by videotaping and described
situations where a child's awareness of the
videotaping crew caused "play-acting" to occur, or
parents indicated that the child was behaving
differently during the taping session. Another
possible source of measurement error may be
introduced when children's movements or positions
cause their mouthing not to be captured by the
camera. Data transcription errors can bias results in
either the negative or positive direction (i.e.,
underestimating mouthing behavior may bias results
low while overestimating mouthing behavior may
bias estimates high). Also, measurement error can
occur if situations arise in which caregivers are
absent during videotaping and researchers must stop
videotaping and intervene to prevent risky behaviors
(Zartarian et al., 1995). Finally, the videography
studies have relatively small sample sizes, and the
number of children studied and their method of
selection may not be entirely representative of the
population of the US.
Survey response studies rely on responses to
questions about a child's mouthing behavior posed
to parents or caregivers. Measurement errors from
these studies could occur for a number of different
reasons, including language/dialect differences
between interviewers and respondents, question
wording problems and lack of definitions for terms
used in questions, differences in respondents'
interpretation of questions, and recall/memory
effects.
Other data collection methodologies (in-person
interview, mailed questionnaire, or questions
administered in "test" format in a school setting)
may have had specific limitations. In-person
interviews could result in either positive or negative
response bias due to distractions posed by young
children, especially when interview respondents
simultaneously care for young children and answer
questions. Other limitations include positive or
negative response bias due to respondents'
perceptions of a "correct" answer, question wording
difficulties, lack of understanding of definitions of
terms used, language and dialect differences
between investigators and respondents, respondents'
desires to avoid negative emotions associated with
giving a particular type of answer, and respondent
memory problems ("recall" effects) concerning past
events. Positive biases would tend to occur when
respondents overestimate behaviors that could result
in soil and dust ingestion; negative biases would tend
to occur when respondents underestimate behaviors
that could result in soil and dust ingestion. Mailed
questionnaires have many of the same limitations as
in-person interviews, but may allow respondents to
respond when they are not distracted by childcare
duties. An in-school test format is more problematic
than either interviews or mailed surveys because
respondent bias related to teacher expectations could
influence responses.
One approach to evaluating the degree of bias in
survey response studies may be to make use of a
surrogate biomarker indicator providing suggestive
September 2017
Page 5-43

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
evidence of ingestion of significant quantities of soil
(although quantitative estimates would not be
possible). The biomarker technique measures the
presence of serum antibodies to Toxocara species, a
parasitic roundworm from cat and dog feces. Two
U.S. studies have found associations between
reported soil ingestion and positive serum antibody
tests for Toxocara infection (Marmor et al., 1987;
Glickman et al., 1981); a third (Nelson et al., 1996)
has not, but the authors stated that reliability of
survey responses regarding soil ingestion may have
been an issue. Further refinement of survey response
methodologies, together with recent NHANES data
on U. S. prevalence of positive serum antibody status
regarding infection with Toxocara species, may be
useful.
5.4.4. Environmental and Household
Interventions
Some of the studies discussed in this chapter
were conducted near hazardous waste sites where
soil contamination was present (e.g., smelting and
mining waste containing lead). Environmental and
household educational interventions have occurred
at many of these sites. Several studies have been
published in the literature to evaluate the
effectiveness of educational and environmental
interventions for reducing blood lead levels in
children. It may be reasonable to assume that
environmental interventions (e.g., soil removal)
would not have had an effect on soil and dust
ingestion rates because blood lead levels may come
down as a result of a reduction of lead concentrations
and not a change in behavior. However, awareness
of contamination and educational interventions may
affect behaviors (e.g., increase in frequency of hand
washing and household cleaning); thus, reducing soil
ingestion rates. Researchers have studied the
effectiveness of educational interventions alone in
reducing blood lead levels in children (Brown et al.,
2006; Jordan et al., 2003; Wasserman, 2002; Rhoads
et al., 1999). These studies found a decline in blood
lead levels in the intervention groups compared to
the control groups. However, Rhoads et al. (1999)
and Jordan et al. (2003) concluded that educational
intervention was only partially effective in reducing
or maintaining lower blood lead levels. Yeoh et al.
(2014) conducted a meta-analysis of five studies of
educational interventions (Lanphear et al., 1996;
Lanphear et al., 1999; Brown et al., 2006; Jordan et
al, 2003; and Wasserman, 2002). The meta-analysis
of the log-transformed data from these five studies
showed that there was no statistical significant
reduction in blood lead levels (Yeoh et al, 2014).
However, another factor that may affect the results
observed in these intervention studies is that bone
turnover in children occurs at a higher rate than that
of an adult, resulting in longer half-life of lead in
blood (8 to 11 months for acute exposure and 20 to
38 months for chronic exposures), thus, providing a
continuous source of lead in blood (Yeoh et al.,
2014). With the exception of Jordan et al. (2003),
who conducted quarterly booster sessions with
participants for 2 years after the first year of
educational intervention, the duration of these
intervention studies was 12 months or less.
5.4.5. Key Studies: Representativeness
of the U.S. Population
Limitations regarding the key studies performed
in the United States for estimating soil and dust
ingestion rates in the entire population of U.S.
children ages 0 to <21 years fall into the broad
categories of geographic range and demographics
(age, sex, race/ethnicity, socioeconomic status).
Regarding geographic range, the two most
obvious issues relate to soil types and climate. Soil
properties might influence the soil ingestion
estimates that are based on excreted tracer elements.
The Davis et al. (1990), Calabrese et al. (1989),
Barnes (1990), Calabrese et al. (1997a), and Davis
and Mirick (2006) tracer element studies were
conducted in locations with soils that had sand
content ranging from 21-80%, silt content ranging
from 16-71%, and clay content ranging from 3-20%
by weight, based on data from USDA (2008). The
location of children in the Calabrese et al. (1997b)
study was not specified, but due to the original
survey response study's occurrence in western
Massachusetts, the soil types in the vicinity of the
Calabrese et al. (1997b) study are likely to be similar
to those in the Calabrese et al. (1989) and Barnes
(1990) study.
The Hogan et al. (1998) study included locations
in the central part of the United States (an area along
the Kansas/Missouri border, and an area in western
Illinois) and one in the eastern United States
(Palmerton, PA). The Davis et al. (1990) study was
conducted in Washington State, Von Lindern et al.
(2016) was conducted in Idaho, and Wilson et al.
(2013) was conducted in Canada. The only key study
conducted in the southern part of the United States
was Vermeer and Frate (1979).
Children might be outside and have access to soil
in a very wide range of weather conditions (Wong et
al., 2000). In the parts of the United States that
experience moderate temperatures year-round, soil
ingestion rates may be fairly evenly distributed
September 2017
Page 5-44

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
throughout the year. During conditions of deep snow
cover, extreme cold, or extreme heat, children could
be expected to have minimal contact with outside
soil. All children, regardless of location, could ingest
soils located indoors in plant containers, soil derived
particulates transported into dwellings as ambient
airborne particulates, or outdoor soil tracked inside
buildings by human or animal building occupants.
Davis et al. (1990) did not find a clear or consistent
association between the number of hours spent
indoors per day and soil ingestion, but reported a
consistent association between spending a greater
number of hours outdoors and high (defined as the
uppermost tertile) soil ingestion levels across all
three tracers used.
The key tracer element studies all took place in
northern latitudes. The temperature and precipitation
patterns that occurred during these four studies' data
collection periods were difficult to discern due to no
mention of specific data collection dates in the
published articles. The Calabrese et al. (1989) and
Barnes (1990) study apparently took place in mid to
late September 1987 in and near Amherst, MA;
Calabrese et al. (1997a) apparently took place in late
September and early October 1992, in Anaconda,
MT; Davis et al. (1990) took place in July, August,
and September 1987, in Richland, Kennewick, and
Pasco, WA; and Davis and Mirick (2006) took place
in the same Washington state location in late July,
August, and very early September 1988 (raw data).
Inferring exact data collection dates, a wide range of
temperatures may have occurred during the four
studies' data collection periods (daily lows from
22-60°F and 25-48°F, and daily highs from
53-81°F and 55-88°F in Calabrese et al. [1989] and
Calabrese et al. [1997a], respectively, and daily lows
from 51-72°F and 51-67°F, and daily highs from
69-103°F and 80-102°F in Davis et al. [1990] and
Davis and Mirick [2006], respectively) (NCDC,
2006). Significant amounts of precipitation occurred
during Calabrese et al. (1989) (more than 0.1 inches
per 24-hour period) on several days; somewhat less
precipitation was observed during Calabrese et al.
(1997a); precipitation in Kennewick and Richland
during the data collection periods of Davis et al.
(1990) was almost nonexistent; there was no
recorded precipitation in Kennewick or Richland
during the data collection period for Davis and
Mirick (2006) (NCDC, 2006).
One key biokinetic model comparison study
(Hogan et al., 1998) targeted three locations in more
southerly latitudes (Pennsylvania, southern Illinois,
and southern Kansas/Missouri) than the tracer
element studies. The other key biokinetic model
comparison study was conducted in Idaho. The
biokinetic model comparison methodology had an
advantage over the tracer element studies in that the
study represented long-term environmental
exposures over periods up to several years that
would include a range of seasons and climate
conditions.
A brief review of the representativeness of the
key studies' samples with respect to sex and age
suggested that males and females were represented
roughly equally in those studies for which study
subjects' sex was stated. Children up to age 12 years
were studied in the key studies with an emphasis on
younger children.
A brief review of the representativeness of the
key studies' samples with respect to socioeconomic
status and racial/ethnic identity suggested that there
were some discrepancies between the study subjects
and the current U.S. population of children age 0 to
<21 years. The single survey response study
(Vermeer and Frate, 1979) was specifically targeted
toward a predominantly rural Black population in a
particular county in Mississippi. The tracer element
studies are of predominantly White populations,
apparently with limited representation from other
racial and ethnic groups. The Amherst, MA study
(Calabrese etal., 1989; Barnes, 1990) did not publish
the study participants' socioeconomic status or racial
and ethnic identities. The socioeconomic level of
children studied by Davis et al. (1990) was reported
to be primarily of middle to high income.
Self-reported race and ethnicity of relatives of the
children studied (in most cases, they were the parents
of the children studied) in Davis et al. (1990) were
White (86.5%), Asian (6.7%), Hispanic (4.8%),
Native American (1.0%), and Other (1.0%), and the
91 married or living-as-married respondents
identified their spouses as White (86.8%), Hispanic
(7.7%), Asian (4.4%), and Other (1.1%). Davis and
Mirick (2006) did not state the race and ethnicity of
the follow-up study participants, who were a subset
of the original study participants from Davis et al.
(1990). For the Calabrese et al. (1997a) study in
Anaconda, MT, population demographics were not
presented in the published article. The study sample
appeared to have been drawn from a door-to-door
census of Anaconda residents that identified
642 toilet-trained children who were less than
72 months of age. Of the 414 children participating
in a companion study (out of the 642 eligible
children identified), 271 had complete study data for
that companion study, and of these 271, 97.4% were
identified as White and the remaining 2.6% were
identified as Native American, Black, Asian, and
Hispanic (Hwang et al., 1997). The 64 children in
the Calabrese et al. (1997a) study apparently were a
September 2017
Page 5-45

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
stratified random sample (based on such factors as
behavior during a previous study, the existence of a
disability, or attendance in daycare) drawn from the
642 children identified in the door-to-door census.
Presumably these children identified as similar races
and ethnicities to the Hwang et al. (1997) study
children. The Calabrese et al. (1997b) study
indicated that 11 of the 12 children studied were
White.
In summary, the geographic range of the key
study populations was somewhat limited. Of those
performed in North America, U.S. locations include
Massachusetts, Kansas, Montana, Missouri, Illinois,
Washington, Pennsylvania, and Idaho. Canada is
also represented. The two most obvious issues
regarding geographic range relate to soil types and
climate. Soil types were not always described, so the
representativeness of the key studies related to soil
types and properties is unclear. The key tracer
element studies all took place in northern latitudes.
The only key study conducted in the southern part of
the United States was Vermeer and Frate (1979).
In terms of sex and age, males and females were
represented roughly equally in those studies for
which study subjects' sex was stated, while the
majority of children studied were under the age of
eight. The tracer element studies are of
predominantly White populations, with a single
survey response study (Vermeer and Frate, 1979)
targeted toward a rural Black population. Other
racial and ethnic identities were not well reported
among the key studies, nor was socioeconomic
status.
5.5. DERIVATION OF RECOMMENDED
SOIL AND DUST INGESTION VALUES
Table 5-34 summarizes the soil and dust
ingestion estimates from the key studies for general
population children, by age range, based on the
tracer, biokinetic modeling, and activity patterns
approaches. These three methods were given equal
weight in deriving the recommendation because of
the inherent limitations in all of the methods (see
Section 5.4). Also, there is no supportive evidence
to suggest that one method provides more reliable
estimates than the other.
The mean and upper percentile recommendations
were derived by averaging the values for each age
group for each of the three study methodologies and
then taking the average of the three study types, as
follows:
IRsoil + dust (IRt+IRb+IRa)/ 3 (Eqn 5-15)
Where:
IRsoil + dust age-specific mean (or 95th percentile)
soil + dust ingestion rate (mg/day)
IRt = average of the age-specific mean (or 95th
percentile) soil + dust ingestion rates from the
various tracer studies (mg/day);
IRh = average of the age-specific mean (or 95th
percentile) soil + dust ingestion rates from the
various biokinetic modeling comparison studies
(mg/day);
IRa = average of the age-specific mean (or 95th
percentile) soil + dust ingestion rates from the
various activity pattern modeling studies.
For example, the mean soil + dust ingestion rate
for children 1 to <6 years was estimated as follows
(see Table 5-34 for additional details):
IRsoil + dust (IRt [99 mg/day] + IRh [90 mg/day] +
IRa [65 mg/day]) / 3
IRsoii + dust = 84 mg/day
Where:
IRt = average of 132, 69,66, and 129 mg/day (means
from 4 tracers studies that represent children 1 to <6
years of age) = 99 mg/day;
IRh = average of 113 and 67 mg/day (means from 2
biokinetic modeling studies that represent children 1
to <6 years of age) = 90 mg/day;
IRa = average of 68 and 61 mg/day (means from 2
activity pattern studies that represent children 1 to <6
years) = 65 mg/day.
Using the number of study participants in the
various studies to weight the means and upper
percentile estimates would not change the
recommended values, when rounded to one
significant figure. Also, although there might be
alternatives to averaging upper percentile values to
get an upper percentile value, it does not appear that
other approaches would significantly change the
upper percentile values for these data because the
upper percentile values from all study types are
similar when rounded to one significant figure.
As stated earlier in this chapter, the key studies
were used as the basis for developing the soil and
dust ingestion recommendations shown in Table 5-1.
The following sections describe in more detail how
the recommended soil and dust ingestion, soil pica,
and geophagy values were derived. Appendix B
September 2017
Page 5-46

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
provides a comparison of data from the tracer studies
conducted in Amherst, MA, Washington State,
Anaconda, MT, and western MA. Appendix C
provides a detailed summary of the key studies used
in developing the recommended soil and dust
ingestion rates. Appendix D provides a detailed
summary of the studies on the prevalence of pica.
5.5.1. Central Tendency Soil and Dust
Ingestion Recommendations
In general, "central tendency" recommendations
in this chapter reflect an arithmetic mean (average)
of measured values within a study, across studies
within a methodology, and across the three
methodologies. For some studies, arithmetic means
were not available from the study, and some other
central estimate was used. For example, for some of
the tracer studies, the means represent the average of
the median values for multiple tracers for each child.
As noted, when describing how different estimates
were averaged, there is uncertainty as to what is a
central estimate consumption rate. Also, for some
studies, the age groups evaluated did not match the
age stratifications in EPA's Guidance on Selecting
Age Groups for Monitoring and Assessing
Childhood Exposures to Environmental
Contaminants (U.S. EPA, 2005). However, for the
purpose of providing recommended values for use in
risk assessment, the ages were assumed to represent
the age categories that they most closely matched.
For infants of 0 to <6 months in age, the
recommended central tendency soil + dust ingestion
estimate for use in risk assessments is 40 mg/day
(36 mg/day rounded to one significant figure). This
value is based on a single key activity pattern
modeling study that provided data for infants 0 to
<7 months (Wilson et al., 2013). Based on the
assumption that 45% of the soil + dust ingestion can
be attributed to soil and 55% can be attributed to
dust, as in EPA's IEUBK model for lead in older
children (U.S. EPA, 1994a), the central tendency soil
ingestion rate for infants 0 to <6 months of age is
18 mg/day and the dust ingestion rate is 22 mg/day.
Rounded to one significant figure, both soil and dust
mean ingestion estimates for infants 0 to <6 months
of age are 20 mg/day. Because the Wilson et al.
(2013) study did not include exposures from
object-to-mouth contact, the recommended values
provided here may underestimate soil + dust
ingestion among this age group.
For infants, 6 months to <1 year, data from two
key biokinetic modeling studies were used to
estimate a central tendency soil + dust ingestion rate
of 70 mg/day; no key tracer studies or activity
pattern modeling studies were available for this age
group. This value is based on the average of the
central estimate generated by von Lindern et al.
(2016) and an EPA-adjusted central estimate from
Hoganetal. (1998). Using the IEUBK model, Hogan
et al. (1998) reported a predicted geometric mean
blood lead level that was 1.4-fold higher than the
geometric mean blood lead levels observed for 31
children 6- to 12-months old from a site in Illinois.
The default soil and dust intakes for the 6- to
12-month old infants in the model (U.S. EPA,
1994b) are 38 mg soil/day and 47 mg house dust/day,
for a total soil + dust intake of 85 mg/day for this life
stage (U.S. EPA, 1994a). Assuming all other model
input parameters are roughly accurate, EPA adjusted
the default ingestion rate of 85 mg/day by a factor of
1.4 to estimate a soil + dust ingestion rate of
61 mg/day. Also using the biokinetic modeling
approach, von Lindern et al. (2016) reported a mean
value of 81 mg/day for children 6 months to <1 year
of age, based on the average of the two best-fit model
runs, as described in Section 5.3.3.7. The average of
the estimates from these two biokinetic modeling
studies is 71 mg/day (rounded to 70 mg/day). Based
on the assumption that 45% of the soil + dust
ingestion can be attributed to soil and 55% can be
attributed to dust, as in EPA's IEUBK model for lead
(U.S. EPA, 1994a), the central tendency soil
ingestion rate for infants 6 months to <1 year of age
is 32 mg/day and the dust ingestion rate is
38 mg/day. Rounded to one significant figure, the
mean soil-only estimate is 30 mg/day and the mean
dust-only estimate is 40 mg/day.
For children 1 to <2 years and children 2 to
<6 years, the recommended central tendency
soil + dust ingestion estimate for use in risk
assessments are 90 mg/day (92 mg/day rounded to
one significant figure) and 60 mg/day (62 mg/day
rounded to one significant figure), respectively.
These values are based on a single key biokinetic
modeling study that provided data for these age
groups (von Lindern et al., 2016). The averages of
the two best-fit model runs were used for these age
groups (see Section 5.3.3.7). Based on the
assumption that 45% of the soil + dust ingestion can
be attributed to soil and 55% can be attributed to
dust, as in EPA's IEUBK model (U.S. EPA, 1994a)
for lead, the central tendency soil ingestion rate for
children 1 to <2 years of age is 41 mg/day and the
dust ingestion rate is 49 mg/day. Rounded to one
significant figure, the soil ingestion rate is 40 mg/day
and the dust ingestion rate is 50 mg/day for children
1 to <2 years old. For children 2 to <6 years, the
soil-only estimate is 27 mg/day and the dust-only
value is 33 mg/day. Rounded to one significant
September 2017
Page 5-47

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
figure, both soil and dust mean ingestion estimates
for infants 2 to <6 years of age are 30 mg/day.
For children ages 1 to <6 years the recommended
central tendency ingestion rate for use in risk
assessment is 80 mg/day. This value is based on
estimates provided by studies conducted using
tracer, biokinetic modeling, and activity pattern
modeling approaches. Four key tracer studies
(Calabrese et al. [1989] as reanalyzed by Stanek and
Calabrese [1995a]; Davis et al. [1990] as reanalyzed
by Stanek and Calabrese [1995a]; Calabrese et al.
[1997a,b]) estimated central tendency soil and dust
ingestion rates for children ranging in age from
1 to <8 years (N= 241). The estimates ranged from
66 to 132 mg/day, with an average of 99 mg/day. For
the Calabrese et al. (1989) study, as reanalyzed by
Stanek and Calabrese et al. (1995a), the average of
the median of the best four tracers for each child was
used in developing the recommendations for
children 1 to <6 years of age. Stanek and Calabrese
et al. (1995a) suggest that this is the most reliable
estimate for this data set. For the Davis et al. (1990)
study, as reanalyzed by Stanek and Calabrese
(1995a), data for only three tracers were available,
and the average of the median of these three tracers
was used. For the Calabrese et al. (1997a) study, the
average based of the best tracer for each child was
used. This estimate was assumed to be more reliable
than the average of the best four tracers because the
central tendency estimates for some of the four
individual tracers were negative. For Calabrese et
al. (1997b), data were available for aluminum,
silicon, and titanium. However, only the data for
aluminum and silicon were used, given the high
degree of variability in the titanium estimates. The
results of two biokinetic modeling studies were used
(Hogan et al., 1998; von Lindern et al., 2016) in
developing the recommendations for children 1 to <6
years of age. In the Hogan et al. (1998) study, blood
levels for 471 children were similar to those
predicted by the IEUBK model. The IEUBK default
soil + dust ingestion values used in the Hogan et al.
(1998) biokinetic modeling study were 135 mg/day
for 1-, 2-, and 3-year-olds; 100 mg/day for 4-year-
olds; 90 mg/day for 5-year-olds; and 85 mg/day for
6-year-olds (U.S. EPA 1994b, 2007). The
time-averaged daily soil + dust ingestion rate for
these 6 years of life was 113 mg/day. Also, using the
biokinetic modeling approach, von Lindern et al.
(2016) estimated a mean soil ingestion rate of
68 mg/day for children 1 to <6 years of age
(N = 1,075), based on the average of the two best-fit
model runs, as described in Section 5.3.3.6. The
average based on these two biokinetic model studies
is 91 mg/day. The two activity pattern modeling
studies provide somewhat lower soil + dust ingestion
estimates for this life stage. Ozkaynak et al. (2011)
provided an estimate of 68 mg/day for 3- to
<6-year-old children, and Wilson et al. (2013)
provided an estimate of 61 mg/day for children aged
7 months to <5 years. The mean of these two
estimates is 65 mg/day. Averaging the soil + dust
ingestion estimates from the three approaches yields
a soil + dust ingestion estimate of 84 mg/day which
was rounded to 80 mg/day. Based on the assumption
that 45% of the soil + dust ingestion can be
attributed to soil and 55% can be attributed to dust,
the central tendency estimates for soil and dust
ingestion alone are both 40 mg/day, rounded to one
significant figure (i.e., 36 mg/day soil and 44 mg/day
dust) for children 1 to <6 years old. Although
children 1 to <6 years old have been grouped
together for the purposes of deriving this
recommendation, it is important to note that children
ages 1 to <2 years have been reported to have higher
blood lead levels compared to other children (Hogan
et al., 1998). This age group also presents higher
mouthing behavior frequency (i.e., hand-to-mouth
and object-to-mouth) than other age groups based on
the available data (See Chapter 4 of this handbook).
Default soil ingestion rates used in the IEUBK
model are higher for this age group than for older
aged children, and von Lindern et al. (2016) showed
that soil ingestion rate for this age group was the
highest among all of the age groups evaluated up to
<10 years of age, with an estimated value of 93
mg/day. The recommended soil ingestion rate for
1-to <2-year-olds for use in risk assessment is
90 mg/day, based on the von Lindern et al. (2016)
study, as discussed above.
For children 6 to <12 years old, the
recommended central tendency soil + dust ingestion
rate is based on data from one biokinetic modeling
study (von Lindern et al., 2016) and one activity
pattern modeling study (Wilson et al., 2013). Based
on biokinetic modeling, soil + dust ingestion was
estimated to be 56 mg/day for 6- to <10-year-olds,
based on the average of two best-fit model runs
conducted by von Lindern et al. (2016) (see Section
5.3.3.7). The estimate for 5- to <12-year-olds,
predicted by Wilson et al. (2013) using the activity
pattern modeling was also 55 mg/day. The average
of these two estimates is 56 mg/day. Rounded to one
significant figure, the central tendency soil + dust
ingestion rate for this life stage is 60 mg/day. Again,
assuming that 45% of the soil + dust ingestion can
be attributed to soil and 55% can be attributed to
dust, the central tendency estimates for soil and dust
ingestion alone are both 30 mg/day, rounded to one
significant figure, for 6- to <12-year-old children.
September 2017
Page 5-48

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
For older children (i.e., ages 12 to <19 years) and
general population adults (i.e., ages >19 years), the
recommended central tendency soil + dust ingestion
rate is 30 mg/day. This is based on data from one key
tracer study for adults (Davis and Mirick, 2006) and
one activity pattern modeling study (Wilson et al.,
2013). Davis and Mirick (2006) reported a mean
soil + dust ingestion rate of 52 mg/day for 33 adults,
based on average estimates using aluminum and
silicon as tracers. Wilson et al. (2013) used an
activity pattern modeling approach to estimate
soil + dust ingestion among various age groups,
including teens (ages 12 to <20 years), adults (ages
20 to 59 years), and senior (ages 60+ years). The
estimated soil + dust ingestion rates were 3.7 mg/day
for teens, 4.2 mg/day for adults, and 3.8 mg/day for
seniors. Given the similarity in these estimates, the
three ages group were combined into one category
(i.e., 12 years through adults), with a central
tendency soil + dust ingestion estimate of 4 mg/day.
Averaging the estimates from the Davis and Mirick
(2006) tracer study and the Wilson et al. (2013)
activity pattern modeling study results in a central
tendency soil + dust ingestion estimate of 28 mg/day.
Rounded to one significant figure, this value is
30 mg/day. Assuming that 45% of the soil + dust
ingestion can be attributed to soil and 55% can be
attributed to dust, the central tendency estimate for
soil is 13 mg/day and the estimate for dust is
17 mg/day. Rounded to one significant figure, the
central tendency estimates are 10 mg/day and soil
and 20 mg/day dust.
For rural populations following traditional rural
or wilderness lifestyles as described by Doyle et al.
(2012) and Irvine et al. (2014) adult soil ingestion
rates may be somewhat higher than those of the
general population (see Sections 5.3.4.20 and
5.3.4.21). Doyle et al. (2012) conducted a tracer
study in Canada in which the estimated mean
soil + dust was 42 mg/day based on the average
values for aluminum and silicon tracers. Irvine et al.
(2014) also conducted a tracer study in Canada and
estimated a mean soil + dust ingestion rate of
52 mg/day, based on the aluminum and silicon
tracers. The average of these two values is
47 mg/day. Rounded to one significant figure, the
soil + dust ingestion estimate is 50 mg/day
(20 mg/day soil and 30 mg/day dust).
5.5.2. Upper Percentile, Soil Pica, and
Geophagy Recommendations
In general, there is considerably more
uncertainty related to the upper percentile soil and
dust ingestion rate estimates than for the average
estimates. Biases due to the errors (e.g., sampling
errors, measurement errors, analytical errors, etc.)
are more likely to affect the upper percentile
estimates than the average estimates. The use of data
obtained from short-term studies to represent
long-term usual behavior also introduces biases that
may have a more considerable effect on the upper
percentile estimates.
The upper percentile soil + dust ingestion rate for
infants 0 to <6 months of age was estimated to be
100 mg/day. This value is based on a single key
activity pattern modeling study that provided a
soil + dust ingestion rate of 140 mg/day for infants 0
to <7 months (Wilson et al., 2013). Rounded to one
significant figure, this value is 100 mg/day. Based on
the assumption that 45% of the soil + dust ingestion
can be attributed to soil and 55% can be attributed to
dust, as in EPA's IEUBK model for lead in older
children (U.S. EPA, 1994a), the upper percentile soil
ingestion rate for infants 0 to <6 months of age is
45 mg/day and the dust ingestion rate is 55 mg/day.
Rounded to one significant figure, the soil ingestion
rate is assumed to be 50 mg/day for soil and
60 mg/day for dust for infants 0 to <6 months of age.
An uncertainty associated with this estimate is that
the Wilson et al. (2013) study did not include
exposures from object-to-mouth contact. Thus, the
recommended values provided here may
underestimate soil + dust ingestion among this age
group.
The recommended upper percentile soil + dust
ingestion rate for infants 6 months to <1 year is
200 mg/day. This value is based on data from one
biokinetic modeling study. Von Lindern et al. (2016)
provided a 95th percentile value of 208 mg/day,
based on the average of two best-fit model runs (see
Section 5.3.3.6), which has been rounded to
200 mg/day. It is assumed that 90 mg/day represents
soil ingestion and 110 mg/day represents dust
ingestion, based on the assumption that 55% of the
200 mg/day is soil and 45% is dust. Rounded to one
significant figure the recommended upper percentile
ingestion rates for infants 6 months to 1 year of age
are 90 mg/day soil and 100 mg/day dust.
The von Lindern et al. (2016) biokinetic
modeling study was also used as the basis for the
upper percentile soil + dust ingestion rate for
children 1 to <2 years of age and 2 to <6 years of
age. For 1- to <2-year-olds, the upper percentile
soil + dust ingestion rate was 240 mg/day, based on
the average of two best-fit model runs (see Section
5.3.3.6). Rounded to one significant figure, the
soil + dust ingestion recommendation for use in risk
assessment is 200 mg/day (90 mg/day soil; 100
mg/day dust). For children ages 2 to <6 years, von
September 2017
Page 5-49

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
Lindern et al. (2016) reported an upper percentile
soil + dust ingestion rate of 162 mg/day, based on the
average of two best-fit model runs. Rounded to one
significant figure, the soil + dust ingestion
recommendation for use in risk assessment is also
200 mg/day (90 mg/day soil; 100 mg/day dust).
For children 1 to <6 years old, upper percentile
soil + dust ingestion rates were derived from three
tracer studies that used the BTM approach, one
biokinetic study, and two activity pattern studies.
The 95th percentile soil + dust ingestion estimates
from the three tracer studies ranged from 154 to
282 mg/day, with a mean of 227 mg/day (N = 229)
(Calabrese et al. [1989] as reanalyzed by Stanek and
Calabrese [1995a]; Davis et al. [1990]; Calabrese et
al. [1997a]). Von Lindern et al. (2016) estimated a
95th percentile rate of 178 mg/day. Ozkaynak et al.
(2011) and Wilson et al. (2013) reported similar
estimates of the 95th percentile value (i.e.,
224 mg/day and 204 mg/day for age groups 3 to
<6 years and 7 months to <5 years, respectively).
The average of these two values is 214 mg/day.
Averaging the estimates from the three approaches
gives an estimated 95th percentile soil + dust
ingestion rate of 206 mg/day for children ages 1 to
<6 years. Rounding to one significant figure, the
recommended upper percentile estimate of
soil + dust ingestion is 200 mg/day (90 mg/day soil;
100 mg/day dust).
Similar upper percentile estimates were provided
for older children by von Lindern et al. (2016) and
Wilson et al. (2013), and the recommended
soil + dust ingestion rate for risk assessment
purposes is also 200 mg/day. This upper percentile
recommendation is the average of the von Lindern et
al. (2016) estimate of 155 mg/day for children ages
6 to <10 years and the Wilson et al. (2013) estimate
of 185 mg/day for children ages 5 to <12 years. The
average of these two values is 170 mg/day. Rounded
to one significant figure the soil + dust ingestion
recommendation for use in risk assessment is
200 mg/day (90 mg/day soil; 100 mg/day dust).
Data that could be used to develop an upper
percentile soil + dust ingestion rate for ages 12 years
through adult were limited. For example, an upper
percentile rate was not provided in the only adult
tracer study. The only data available were from a
single activity pattern study that provided an upper
percentile rate (i.e., 14 mg/day) for teens, adults, and
seniors that is inconsistent with the recommended
central tendency rate of 30 mg/day. Therefore, an
upper percentile recommendation for 12 years of age
through adults was developed by assuming that the
ratio of the 95th percentile value to the mean value
for adults is the same as the average of the ratios of
95th percentiles to means for all other age groups
(i.e., average ratio of the 95th percentile
recommendations to the mean recommendation for
all ages groups <12 years = 3.2). It should be noted
that this assumes that the variance is the same for
children and adults. Because estimates for adults are
lower than those of children, they are likelier to have
a lower variance. Applying this ratio to the central
tendency estimate of 30 mg/day for the 12 years
through adult age group gives an estimated upper
percentile value of 100 mg/day (i.e., 30 mg/day x
3.2 = 94 mg/day, rounded to one significant figure).
If upper percentile soil- or dust-only values are
needed, the recommended rates are 50 mg/day soil
and 60 mg/day dust. These values are rounded to one
significant figure from 45 mg/day and 55 mg/day,
assuming that 45% of the soil + dust estimate is soil
and 55% is dust.
For rural populations following traditional rural
or wilderness lifestyles the data from tracer studies
conducted in rural Canadian locations and reported
in Doyle et al. (2012) and Irvine et al. (2014) (see
Section 5.3.4.20 and 5.3.4.21) may be used to
estimate an upper percentile soil + dust ingestion rate
for this population. Doyle et al. (2012) reported a
90th percentile soil + dust ingestion rate of
124 mg/day, based on the average of the results using
aluminum and silicon as tracers. Irvine et al. (2014)
reported a 90th percentile value of 220 mg/day based
on the average of these same two tracers. Averaging
the results from these two studies gives an upper
percentile estimate of 172 mg/day. Rounded to one
significant figure, the upper percentile soil + dust
ingestion rate for these populations would be
200 mg/day (90 mg/day soil and 100 mg/day dust).
For the upper percentile soil pica and geophagy
recommendations shown in Table 5-1, two primary
lines of evidence suggest that at least some U.S.
children exhibit soil pica behavior at least once
during childhood. First, the survey response studies
of reported soil ingestion behavior that were
conducted in numerous U.S. locations and of
different populations consistently yield a certain
proportion of respondents who acknowledge soil
ingestion by children. The surveys typically did not
ask explicit and detailed questions about the soil
ingestion incidents reported by the caregivers who
acknowledged soil ingestion in children. Responses
conceivably could fall into three categories:
(1)	responses in which caregivers interpret visible
dirt on children's hands and subsequent
hand-to-mouth behavior as soil ingestion;
(2)	responses in which caregivers interpret
intentional ingestion of clay, "dirt," or soil as soil
ingestion; and (3) responses in which caregivers
September 2017
Page 5-50

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
regard observations of hand-to-mouth behavior of
visible quantities of soil as soil ingestion.
Knowledge of soils' bulk density allows one to infer
that these latter observed hand-to-mouth soil
ingestion incidents are likely to represent a quantity
of soil that meets the quantity part of the definition
of soil pica used in this chapter, or 1,000 mg.
Occasionally, what is not known from survey
response studies is whether the latter type of survey
responses include responses regarding repeated soil
ingestion that meets the definition of soil pica used
in this chapter. The second category probably does
represent ingestion that would satisfy the definition
of soil pica as well as geophagy. The first category
may represent relatively small amounts that appear
to be ingested by many children based on the
biokinetic modeling studies and the tracer element
studies. Second, the U.S. tracer studies report a wide
range of soil ingestion values. Due to averaging
procedures used, for 4-, 7-, or 8-day periods, the
rounded range of these estimates of soil ingestion
behavior that apparently met the definition of soil
pica used in this chapter is from 1,000 to
6,000 mg/day averaged over the study period. Due
to the short-term nature of these studies and the
limited amount of data available for children
exhibiting pica behavior, the lower end of this range
of 1,000 mg/day is recommended for use in risk
assessments involving soil pica for children 1 to
<6 years old. However, it is important to note that
soil ingestion for these children exhibiting soil pica
behavior has been reported as high as 20 to 25 g/day
on a given day (Calabrese et al., 1997b).
Although no tracer element studies or biokinetic
model comparison studies have been performed for
children 15- to <21-year-olds in which soil pica
behavior has been investigated, EPA is aware of one
study documenting pica behavior in a group that
includes children in this age range (Hyman et al.,
1990). The study was not specific regarding whether
soil pica (vs. other pica substances) was observed,
nor did it identify the specific ages of the children
observed to practice pica. In the absence of data that
can be used to develop specific soil pica ingestion
recommendations for children aged 15 years and 16
to <21 years, EPA recommends that risk assessors
who need to assess risks via soil and dust ingestion
to children ages 15 to <21 years use the same soil
ingestion rate as that recommended for younger
children, in the 1 to <6, 6 to <11, and 11 to <16-year
age categories.
Researchers who have studied human geophagy
behavior around the world typically have studied
populations in specific locations, and often include
investigations of soil properties as part of the
research (Wilson, 2003; Aufreiter et al., 1997). Most
studies of geophagy behavior in the United States
were survey response studies of residents in specific
locations who acknowledged eating clays. Typically,
study subjects were from a relatively small area such
as a county, or a group of counties within the same
state. Although geophagy behavior may have been
studied in only a single county in a given state,
documentation of geophagy behavior by some
residents in one or more counties of a given state
may suggest that the same behavior also occurs
elsewhere within that state.
A qualitative description of amounts of soil
ingested by geophagy practitioners was provided by
Vermeer and Frate (1979) with an estimated mean
amount, 50 g/day that apparently was averaged over
32 adults and 18 children. The 18 children whose
caregivers acknowledged geophagy (or more
specifically, eating of clay) were ( Y = 16) ages 1 to
4 and ( Y = 2) ages 5 to 12 years. The definition of
geophagy used included consumption of clay "on a
regular basis over a period of weeks." EPA is
recommending this 50 g/day value for geophagy.
This mean quantity is roughly consistent with a
median quantity reported by Geissler et al. (1998) in
a survey response study of geophagy in primary
school children in Nyanza Province, Kenya
(28 g/day, range 8 to 108 g/day; interquartile range
13 to 42 g/day).
Several studies are available that investigated
pica behavior among pregnant women. Many of
these studies focus on the prevalence of the behavior,
and very few provide data on amounts ingested
(Fawcett et al., 2016; Lin et al., 2015; Lumish et al.,
2014; Klitzman et al., 2002; Simpson et al., 2000;
Smulian et al., 1995; Cooksey, 1995; Bronstein and
Dollar 1974; Ferguson and Keaton, 1950). Studies of
pica among pregnant women in various U.S.
locations (Rainville, 1998; Corbett et al., 2003;
Smulian et al., 1995) suggest that clay geophagy
among pregnant women may include those less than
21 years old (Smulian et al., 1995; Corbett et al.,
2003). Smulian provides a quantitative estimate of
clay consumption of approximately
200-500 g/week, for the very small number of
geophagy practitioners (N = 4) in that study's sample
{N = 125). If consumed on a daily basis, this quantity
(approximately 30 to 70 g/day) is roughly consistent
with the Vermeer and Frate (1979) estimated mean
of 50 g/day.
Johns and Duquette (1991) describe use of clays
in baking bread made from acorn flour, in a ratio of
1 part clay to 10 or 20 parts acorn flour, by volume,
in a Native American population in California and in
Sardinia (~12 grams clay suspended in water added
September 2017
Page 5-51

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
to 100 grams acorn). Either preparation method
would add several grams of clay to the final prepared
food; daily ingestion of the food would amount to
several grams of clay.
5.6. REFERENCES FOR CHAPTER 5
ASTM (American Society for Testing and Materials)
(2017) Standard specification for woven
wire test sieve cloth and test sieves.
Designation: El 1-17. ASTM International,
West Conshohocken, PA.
ATSDR (Agency for Toxic Substances and Disease
Registry). (2001) Summary report for the
ATSDR soil-pica workshop. ATSDR, U.S.
Department of Health and Human Services,
Atlanta, GA. March 20, 2001.
Aufreiter, S; Hancock, RGV; Mahaney, WC;
Stambolic-Robb, A; Sanmugadas, K.
(1997) Geochemistry and mineralogy of
soils eaten by humans. Int J Food Sci Nutr
48(5):293-305.
Barltrop, D. (1966) The prevalence of pica. Am J
Dis Child 112(2): 116-123.
Barnes, R. (1990) Childhood soil ingestion: how
much dirt do kids eat? Anal Chem
62(19): 1023A-1033 A.
Beamer, PI; Elish, CA; Roe, DJ; Loh, M; Layton,
DW. (2012) Differences in metal
concentration by particle size in house dust
and soil. J Environ Monit 14(3):839-844.
Bergstrom, C; Shira, J; Kissel, J. (2011) Particle size
distributions, size concentration
relationships, and adherence to hands of
selected geologic media derived from
mining, smelting, and quarrying activities.
Sci Total Environ 409(20):4247-4256.
Binder, S; Sokal, D; Maughan, D. (1986) Estimating
soil ingestion: the use of tracer elements in
estimating the amount of soil ingested by
young children. Arch Environ Health
41(6):341-345.
Black, K; Shalat, SL; Freeman, NC; Jimenez, M;
Donnelly, KC; Calvin, JA. (2005)
Children's mouthing and food-handling
behavior in an agricultural community on
the US/Mexico border. J Expo Anal
Environ Epidemiol 15(3):244-251.
Bronstein, ES; Dollar, J. (1974) Pica in pregnancy.
J Med Assoc Ga 63(8):332-335.
Brown, MJ; McLaine, P; Dixon, S; Simon, P. (2006).
A randomized, community-based trial of
home visiting to reduce blood lead levels in
children. Pediatrics 117(1): 147-53.
Bruhn, CM; Pangborn, RM. (1971) Reported
incidence of pica among migrant families.
J Am Diet Assoc 58(5):417-420.
Calabrese, EJ; Stanek, EJ. (1992a) Distinguishing
outdoor soil ingestion from indoor dust
ingestion in a soil pica child. Regul Toxicol
Pharmacol 15:83-85.
Calabrese, EJ; Stanek, EJ. (1992b) What proportion
of household dust is derived from outdoor
soil? J Soil Contam l(3):253-263.
Calabrese, EJ; Stanek, EJ. (1993) Soil pica: Not a
rare event. J Environ Sci Health A
28(2):373-384.
Calabrese, EJ; Stanek, EJ. (1995) Resolving
intertracer inconsistencies in soil ingestion
estimation. Environ Health Perspect
103(5):454-456
Calabrese, EJ; Barnes, R; Stanek, EJ 3rd; Pastides,
H.; Gilbert, CE; Veneman, P; Wang, X;
Lasztity, A; Kostecki, PT. (1989) How
much soil do young children ingest: An
epidemiologic study. Regul Toxicol Pharm
10(2): 123—137.
Calabrese, EJ; Stanek, EJ; Gilbert, CE; Barnes, R.
(1990) Preliminary adult soil ingestion
estimates: results of a pilot study. Regul
Toxicol Pharmacol 12:88-95.
Calabrese, EJ; Stanek, EJ; Gilbert, CE. (1991)
Evidence of soil-pica behavior and
quantification of soil ingested. Hum Exp
Toxicol 10(4):245-249.
Calabrese, EJ; Stanek, EJ; Barnes, R; Burmaster,
DE; Callahan, BG; Heath, JS;
Paustenbach, D; Abraham, J; Gephart, LA.
(1996) Methodology to estimate the
amount and particle size of soil ingested
by children: implications for exposure
assessment at waste sites. Regul Toxicol
Pharmacol 24(3):264-268.
Calabrese, EJ; Stanek, EJ 3rd; Pekow, P; Barnes,
RM. (1997a) Soil ingestion estimates for
children residing on a superfund site.
Ecotoxicol Environ Saf 36(3):258-268.
Calabrese, EJ; Stanek, EJ; Barnes, RM. (1997b) Soil
ingestion rates in children identified by
parental observation as likely high soil
ingesters. J Soil Contam 6(3):271-279.
Cao, ZG; Yu, G; Chen, YS; Cao, QM; Fiedler, H;
Deng, SB; Huang, J; Wang, B. (2012)
Particle size: A missing factor in risk
assessment of human exposure to toxic
chemicals in settled indoor dust. Environ
Int 49:24-30.
Carlisle, EM. (1980) Biochemical and
morphological changes associated with
long bone abnormalities in silicon
September 2017
Page 5-52

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
deficiency. J Nutr 110(5): 1046-1055.
Chen, Z; Klein, T; Murray, RZ; Crawford, R;
Chang, J; Wu, C; Xiao, Y. (2016)
Osteoimmodulation for the development
of advanced bone biomaterials. Mater
Today 19(6):304-321.
Chien, LC; Tsou, MC; Hsi, HC; Beamer, P;
Bradham, K; Hseu, ZY; Jien, SH; Jiang,
CB; Dang, W; Ozkaynak, H. (2015) Soil
ingestion rates for children under 3 years
old in Taiwan. J Expo Sci Environ
Epidemiol 27(2):33-40.
Choate, LM; Ranville, JF; Bunge, AL; Macalady,
DL. (2006) Dermally adhered soil: 1.
Amount and particle size distribution. Int
Environ Assess Manag 2(4):375-384.
Clausing, P; Brunekreef, B; Van Wijnen, JH. (1987)
A method for estimating soil ingestion by
children. Int Arch Occup Environ Health
59(l):73-82.
Cooksey, NR. (1995) Pica and olfactory craving of
pregnancy: How deep are the secrets? Birth
22(3):129-137.
Cooper, M. (1957) Pica: A survey of the historical
literature as well as reports from the fields
of veterinary medicine and anthropology,
the present study of pica in young children,
and a discussion of its pediatric and
psychological implications. Springfield,
IL: Charles C. Thomas.
Corbett, RW; Ryan, C; Weinrich, SP. (2003) Pica in
pregnancy: Does it affect pregnancy
outcome? MCN Am J Matern Child Nurs
25(3):183—189.
Danford, DE. (1982) Pica and nutrition. Annu Rev
Nutr 2:303-322.
Danford, DE. (1983) Pica and zinc. In: Prasad, AS;
Cavdar, AO; Brewer, GJ; Aggett, PJ; eds.
Zinc deficiency in human subjects. New
York: AlanR. Liss, Inc., pp. 185-195.
Davis, S; Mirick, D. (2006) Soil ingestion in children
and adults in the same family. J Exp Anal
Environ Epidem 16:63-75.
Davis, S; Waller, P; Buschbom, R; Ballou, J; White,
P. (1990) Quantitative estimates of soil
ingestion in normal children between the
ages of 2 and 7 years: Population based
estimates using aluminum, silicon, and
titanium as soil tracer elements. Arch
Environ Health 45(2): 112-122.
Davis, S; Myers, PA; Kohler, E; Wiggins, C. (1995)
Soil ingestion in children with pica: Final
report. U.S. EPA Cooperative Agreement
CR 816334-01. Fred Hutchinson Cancer
Research Center, Seattle, WA.
Dickins, D; Ford, RN. (1942) Geophagy (dirt eating)
among Mississippi Negro school children.
Amer Soc Rev 7:59-65.
Dominy, NJ; Davoust, E; Minekus, M. (2003)
Adaptive function of soil consumption: an
in vitro study modeling the human stomach
and small intestine. J Exp Biol
207:319-324.
Doyle, JR; Blais, JM; Holmes, RD; White, PA.
(2012) A soil ingestion pilot study of a
population following a traditional lifestyle
typical of rural and wilderness areas. Sci
Total Environ 424:110-120.
Fawcett, EJ; Fawcett, JM; Mazmanian, D. (2016) A
meta-analysis of the worldwide prevalence
of pica during pregnancy and the
postpartum period. Int J Gynecol Obstet
133(3):277-283.
Ferguson, JH; Keaton, A. (1950) Studies of the diets
of pregnant women in Mississippi: II Diet
patterns. New Orleans Med Surg J
103(2):81-87.
Ferguson, AC; Canales, RA; Beamer, P; Auyeung,
W; Key, M; Munninghoff, A; Lee, KT;
Robertson, A; Leckie, JO. (2006) Video
methods in the quantification of children's
exposures. J Expo Sci Environ Epidemiol
16(3):287-298.
Gavrelis, N; Sertkaya, A; Bertelsen, L; Cuthbertson,
B; Phillips, L; Moya, J. (2011) An analysis
of the proportion of the U.S. population that
ingests soil or other non-food substances.
HumEcol Risk Assess 17(4):996-1012.
Geissler, PW; Shulman, CE; Prince, RJ; Mutemi, W;
Mnazi, C; Friis, H; Lowe, B. (1998)
Geophagy, iron status and anaemia among
pregnant women on the coast of Kenya.
Trans Royal Soc Trop Med Hyg
92(5):549-553.
Glickman, LT; Chaudry, IU; Costantino, J; Clack,
FB; Cypess. RH; Winslow, L. (1981) Pica
patterns, toxocariasis, and elevated blood
lead in children. Am J Trop Med Hyg
30(l):77-80.
Grigsby, RK; Thyer, BA; Waller, RJ; Johnston, GA
Jr. (1999) Chalk eating in middle Georgia:
A culture-bound syndrome of pica? South
Med J 92(2): 190-192.
Groot, ME; Lekkerkerk, MC; Steenbekkers, LPA.
(1998) Mouthing behavior of young
children: An observational study.
Wageningen, the Netherlands: Agricultural
University. Available online at
http://edepot.wur.nl/384470.
Harris, SG; Harper, BL. (1997) A Native American
September 2017
Page 5-53

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
exposure scenario.	Risk Anal
17(6):789-795.
Hawley, JK. (1985) Assessment of health risk from
exposure to contaminated soil. Risk Anal
5(4):289-302.
Hogan, K; Marcus, A; Smith, R; White, R (1998)
Integrated exposure uptake biokinetic
model for lead in children: empirical
comparisons with epidemiologic data.
Environ Health Perspect 106 (Supp
6):1557-1567.
Hooda, PS; Henry, CJ; Seyoum, TA; Armstrong,
LD; Fowler, MB. (2004) The potential
impact of soil ingestion on human mineral
nutrition.	Sci Total Environ
333(l-3):74-87.
Hook, EB. (1978) Dietary cravings and aversions
during pregnancy. Am J Clini Nutri
31:1355-1362.
Hwang, YH; Bornschein, RL; Grote, J; Menrath, W;
Roda, S. (1997) Environmental arsenic
exposure of children around a former
copper smelter site. EnvironRes 72:72-81.
Hyman, SL; Fisher, W; Mercugliano, M; Cataldo,
MF. (1990) Children with self-injurious
behavior. Pediatrics 85:437-441.
ICRP (International Commission on Radiological
Protection). (2002) Basic anatomical and
physiological data for use in radiological
protection: reference values. A report of
age- and gender-related differences in the
anatomical and physiological
characteristics of reference individuals.
ICRP Publication 89. Ann ICRP 32
(3-4):5-265.
Ikegami, M; Yoneda, M; Tsuji, T; Bannai, O;
Morisawa, S. (2014) Effect of particle size
on risk assessment of direct soil ingestion
and metals adhered to children's hands at
playgrounds. Risk Anal 34(9): 1677-1687.
Irvine, G; Doyle, JR; White, PA; Blais, JM. (2014)
Soil ingestion rate determination in a rural
population in Alberta, Canada practicing a
wilderness lifestyle. Sci Tot Environ. 470-
471:138-146.
Jang, JY; Kim, SY; Kim, SJ; Lee, KE; Cheong, HK;
Kim, EH; Choi, KH; Kim, YH. (2014)
General factors of the Korean exposure
factors handbook. J Prev Med Pub Health
47:7-17.
Johns, T; Duquette, M. (1991) Detoxification and
mineral supplementation as functions of
geophagy. Am J ClinNutr 53(2):448-456.
Jordan, CM; Yust, BL; Robison, LL; Hannan, P;
Deinard, AS. (2003) A randomized trial of
education to prevent lead burden in
children at high risk for lead exposure:
Efficacy as measured by blood lead
monitoring. Environ Health Perspect
111(16): 1947—51.
Jugdaohsingh, R. (2007) Silicon and bone health. J
Nutr Health Aging 11(2):99-110.
Jugdaohsingh, R; Anderson, SHC; Tucker, KL;
Elliott, H; Kiel, DP; Thompson, RPH;
Powell, JJ. (2002) Dietary silicon intake
and absorption. Am J Clin Nutr
75(5):887-893.
Klitzman, S; Sharma, A; Nicaj, L; Vitkevich, R;
Leighton, J. (2002) Lead poisoning among
pregnant women in New York City: Risk
factors and screening practices. J Urban
Health 79(2):225-237.
Kinnell, HG. (1985) Pica as a feature of autism. Br
J Psychiatry 147:80-82.
Kissel, JC; Shirai, JH; Richter, KY; Fenske, RA.
(1998a) Empirical investigation of hand-to-
mouth transfer of soil. Bull Environ
Contam Toxicol 60(3):379-86.
Kissel, JC; Shirai, JH; Richter, KY; Fenske, RA.
(1998b) Investigation of dermal contact
with soil in controlled trials. J Soil Contam
7(6):737-752.
Korman, SH. (1990) Pica as a presenting symptom
in childhood celiac disease. Am J Clin Nutr
51(2): 139-14l.Lanphear, BP; Winter, NL;
Apetz, L; Eberly, S; WeitzmanM. (1996).
A randomized trial of the effect of dust
control on children's blood lead levels.
Pediatrics 98(l):35-40.
Lanphear, BP; Winter, NL; Apetz, L; Eberly, S;
Weitzman M. (1996) A randomized trial of
the effect of dust control on children's
blood lead levels. Pediatrics 98(l):35-40.
Lanphear, BP; Howard, C; Eberly, S; Auinger, P;
Kolassa, J; Weitzman, M; Schaffer, SJ;
Alexander, K. (1999) Primary prevention
of childhood lead exposure: A randomized
trial of dust control. Pediatrics 103(4 Pt
l):772-777.
Layton DW; Beamer, P (2009) Migration of
contaminated soil and airborne particulates
to indoor dust. Environ Sci Technol. 43
(21), 8199-8205.
Lasztity, A; Wang, X; Viczian, M; Israel, Y; Barnes,
R. (1989) Inductively coupled plasma
spectrometry in the study of childhood soil
ingestion. Part 2. Recovery. J Anal Atomic
Spectrom 4(8):737-742.
Lin, JW; Temple, L; Trujillo, C; Mejia-Rodriquez, F;
Rosas, LG; Fernald, L; Young, SL. (2015)
September 2017
Page 5-54

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
Pica during pregnancy among Mexican-
born women: a formative study. Matern
Child Nutr 11:550-558.
Lumish, RA; Young, SL; Lee, S; Cooper, E;
Pressman, Eva; Guillet, R; O'Brien, KO.
(2014) Gestational iron deficiency is
associated with pica behaviors in
adolescents. J Nutr 144(10): 1533-1539.
Ma, J; Pan, LB; Wang, Q; Lin, CY; Duan, XL; Hou,
H. (2016) Estimation of the daily soil/dust
(SD) ingestion rate of children from Gansu
Province, China via hand-to-mouth contact
using tracer elements. Environ Geochem
Health. DOI 10.1007/sl0653-016-9906-l.
Manton, WI; Angle, CR; Stanek, KL; Reese, YR;
Kuehnemann, TJ. (2000) Acquisition and
retention of lead by young children.
Environ Res 82:60-80.
Marmor, M; Glickman, L; Shofer, F; Faich, LA;
Rosenberg, C; Cornblatt, B; Friedman, S.
(1987) Toxocara canis infection of
children:	epidemiologic	and
neuropsychologic findings. Am J Public
Health 77(5):554-559.
Moya, J; Phillips, L. (2014) Areview of soil and dust
ingestion studies for children. J Expo Sci
Environ Epidemiol 24:545-554.
NCDC (National Climatic Data Center). (2006) Data
documentation for data set TD3200: (DSI-
3200). Surface land daily cooperative
summary of the day. Asheville, NC:
National Climatic Data Center. Available
online	at
https://rda.ucar. edu/datasets/ds510.0/docs/
2006jun. td3200.html.
Nelson, S; Greene, T; Ernhart, CB. (1996) Toxocara
canis infection in preschool age children:
risk factors and the cognitive development
of preschool children. Neurotoxicol Teratol
18(2):167-174.
Nielsen, FH. (2014) Update on the possible
nutritional importance of silicon. J Trace
ElemMed Biol 28(4): 379-382.
Obialo, CI; Crowell, AK; Wen, XJ; Conner, AC;
Simmons, EL. (2001) Clay pica has no
hematologic or metabolic correlate in
chronic hemodialysis patients. J Ren Nutr
11(1):32—36.
Ozkaynak, H; Xue, J; Zartarian, VG; Glen, G; Smith,
L. (2011) Modeled estimates of soil and
dust ingestion rates for children. Risk Anal
31(4):592-608.
Powell, JJ; Ainley, CC; Harvey, RS; Mason, IM;
Kendall, MD; Sankey, EA; Dhillon, AP;
Thompson, RP. (1996) Characterisation of
inorganic microparticles in pigment cells of
human gut associated lymphoid tissue. Gut
38(3):390-395.
Price, CT; Langford, JR; Liporace, FA (2012)
Essential nutrients for bone health and a
review of their availability in the average
North American diet. Open Orthop J
6:143-149.
Rainville, AJ. (1998) Pica practices of pregnant
women are associated with lower maternal
hemoglobin level at delivery. J Am Diet
Assoc 98(3):293-296.
Rhoads, GG; Ettinger, AS; Weisel, CP; Rhoads, GG;
Ettinger, AS; Weisel, CP; Buckley, TJ;
Goldman, KD; Adgate, J; Lioy, PJ. (1999).
The effect of dust lead control on blood lead
in toddlers: a randomized trial. Pediatrics
103(3):551-555.
Robischon, P. (1971) Pica practice and other hand-
mouth behavior and children's
developmental level. Nurs Res 20(1):4-16.
Shepherd, NA; Crocker, PR; Smith, AP; Levison,
DA. (1987) Exogenous pigment inPeyer's
patches. Hum Pathol 18(l):50-54.
Simpson, E; Mull, JD; Longley, E; East, J. (2000)
Pica during pregnancy in low-income
women born in Mexico. West J Med
173(l):20-24.
Smulian, JC; Motiwala, S; Sigman, RK. (1995) Pica
in a rural obstetric population. South Med
J88(12):1236-1240.
Stanek, EJ III; Calabrese, EJ. (1995a) Soil ingestion
estimates for use in site evaluations based
on the best tracer method. Hum Ecol Risk
Assess 1(3): 133-156.
Stanek, EJ III; Calabrese, EJ. (1995b) Daily
estimates of soil ingestion in children.
Environ Health Perspect 103(3):276-285.
Stanek, EJ III; Calabrese, EJ. (2000) Daily soil
ingestion estimates for children at a
Superfund site. Risk Anal 20(5):627-635.
Stanek, EJ III; Calabrese, EJ; Mundt, K; Pekow, P;
Yeatts, KB. (1998) Prevalence of soil
mouthing/ingestion among healthy
children aged 1 to 6. J Soil Contam
7(2):227-242.
Stanek, EJ III; Calabrese, EJ; Barnes, RM. (1999)
Soil ingestion estimates for children in
Anaconda using trace element
concentrations in different particle size
fractions. Hum Ecol Risk Assess
5(3):547-558.
Stanek, EJ III; Calabrese, EJ; Zorn, M. (2001a)
Biasing factors for simple soil ingestion
estimates in mass balance studies of soil
September 2017
Page 5-55

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
ingestion. Hum Ecol Risk Assess
7(2):329-355.
Stanek, EJ III; Calabrese, EJ; Zorn, M. (2001b) Soil
ingestion distributions for Monte Carlo risk
assessment in children. Hum Ecol Risk
Assess 7(2):357-368.
Stanek, EJ III; Calabrese, EJ; Xu, B. (2012a) Meta-
analysis of mass balance studies of soil
ingestion in children. Risk Anal 32(3):
433-447.
Stanek, EJ III; Xu, B; Calabrese, EJ. (2012b)
Equation reliability of soil ingestion
estimates in mass-balance soil ingestion
studies. Risk Anal 32(3):448-463.
Starks, PT; Slabach, BL. (2012) The scoop on eating
dirt. Sci Am 306(6):30-32.
USDA (U.S. Department of Agriculture). (1999)
Soil taxonomy: a basic system of soil
classification for making and interpreting
soil surveys. National Resources
Conservation Service, Agriculture
Handbook No. 436. Washington, DC: U.S.
Government Printing Office. Available
online	at
https://www.nrcs.usda.gov/Internet/FSE_D
OCUMENTS/nrcsl42p2_051232.pdf.
USDA (United States Department of Agriculture).
(2008) Web soil survey. U.S. Department of
Agriculture, Natural Resources
Conservation Service. Available online at
http://websoilsurvey.nrcs.usda.gov/app/Ho
mePage.htm.
U.S. EPA (Environmental Protection Agency).
(1994a) Guidance manual for the integrated
exposure uptake biokinetic model for lead
in children. Office of Solid Waste and
Emergency Response Washington, DC;
EPA	540/R-93/081.
nepis.epa.gov/Exe/ZyPURL.cgi?Dockey=
2000WN4R.TXT.
U.S. EPA (Environmental Protection Agency).
(1994b) Technical support document:
parameters and equations used in the
integrated exposure uptake biokinetic
(IEUBK) model for lead in children
(v 0.99d). Technical Review Workgroup
for Lead, Washington, DC and the
Environmental Criteria and Assessment
Office, Research Triangle Park, NC;
EPA/540/R-94/040. Available online at
https://ntrl.ntis.gov/NTRL/dashboard/searc
hResults/titleDetail/PB94963505.xhtml.
U.S. EPA. (Environmental Protection Agency).
(2005) Guidance on selecting age groups
for monitoring and assessing childhood
exposures to environmental contaminants.
Risk Assessment Forum, Washington, DC.
EPA/630/P-03/003F. Available online at
https://www.epa.gov/sites/production/files/
2013 -09/documents/agegroups.pdf.
U.S. EPA. (Environmental Protection Agency).
(2007)	User's guide for the integrated
exposure uptake biokinetic model for lead
in children (IEUBK) Windows®. Office of
Superfund Remediation and Technology
Innovation. Washington, DC. EPA 9285.7-
42/540-K-01-005
U.S. EPA (Environmental Protection Agency).
(2008)	Child-specific exposure factors
handbook. National Center for
Environmental Assessment, Washington,
DC. Available online at
https ://c!pub. epa.gov/ncea/risk/recordispla
y.cfm?deid= 199243.
U.S. EPA (Environmental Protection Agency).
(2010)	The stochastic human exposure and
dose simulation (SHEDS) model. National
Exposure Research Laboratory,
Washington, DC. Available on-line at
https://www.epa.gov/chemical-
research/stochastic-human-exposure-and-
dose-simulation-sheds-estimate-human-
exposure.
U.S. EPA (Environmental Protection Agency)
(2011)	Exposure factors handbook: 2011
Edition. Office of Research and
Development, Washington, DC.
EPA/600//R-09/052F. Available online at
https ://c!pub. epa.gov/ncea/risk/recordispla
y.cfm?deid=236252.
U.S. OSHA (Occupational Safety & Health
Administration). (1987) Dust and its
control. In: Dust control handbook for
minerals processing, Chapter 1. Bureau of
Mines, Department of the Interior,
Washington, DC. Contract No. J0235005.
Van Dyck, K; Robberecht, H; Van Cauwenburgh, R;
Van Vlaslaer, V; Deelstra, H. (2000)
Indication of silicon essentiality in humans:
serum concentrations in Belgian children
and adults, including pregnant women.
Biol Trace Elem Res 77(l):25-32.
Van Wijnen, JH; Clausing, P; Brunekreff, B. (1990)
Estimated soil ingestion by children.
Environ Res 51(2): 147-162.
Vermeer, DE; Frate, DA. (1979) Geophagia in rural
Mississippi: Environmental and cultural
contexts and nutritional implications. Am J
ClinNutr 32(10):2129-2135.
Von Lindern, I; Spalinger, S; Petroysan, V; von
September 2017
Page 5-56

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
Braun, M. (2003) Assessing remedial
effectiveness through the blood
lead: soil/dust lead relationship at the
Bunker Hill Superfund Site in the Silver
Valley of Idaho. Sci Total Environ
303(1-2):139-170.
VonLindern, I; Spalinger, S; Stifelman, ML; Stanek,
LW; Bartrem, C. (2016) Estimating
children's soil/dust ingestion rates through
retrospective analyses of blood lead
biomonitoring from the Bunker Hill
Superfund Site in Idaho. Environ Health
Perspect 124(9): 1462-1470. DOI:
10.1289/ehp.l510144.
Wang, S; Ma, J; Pan, L; Lin, C; Wang, B; Duan, X.
(2015) Quantification of soil/dust (SD) on
the hands of children from Hubei Province,
China using hand wipes. Ecotoxicol
Environ Saf 120:193-197.
Ward, P; Kutner, NG. (1999) Reported pica behavior
in a sample of incident dialysis patients. J
Ren Nutr 9(1): 14-20.
Wasserman, LR. (2002) The effects of a family-
based educational intervention on the
prevention of lead poisoning in children.
Doctorate of Education Dissertation.
Florida International University.
White, PD; VanLeeuwen, P; Davis, BD; Maddaloni,
M; Hogan, KA; Marcus, AH; Elias, RW.
(1998) The conceptual structure of the
integrated exposure uptake biokinetic
model for lead in children. Environ Health
Perspect 106(Suppl 6): 1513-1530.
Wilson, MJ. (2003) Clay mineralogical and related
characteristics of geophagic materials. J
Chem Ecol 29(7): 1525-1547.
Wilson, R; Jones-Otazo, H; Petrovic, S; Mitchell, I;
Bonvalot, Y; Williams, D; Richardson,
GM. (2013) Revisiting soil and dust
ingestion rates based on hand-to-mouth
transfer. Human Ecol Risk Assess
19:158-188.
Wilson, R; Jones-Otazob, H; Petrovic, S;
Roushorned, M; Smith-Munoze, L;
Williams, D; Mitchell, I. (2015) Estimation
of sediment ingestion rates based on hand-
to-mouth contact and incidental surface
water ingestion. Hum Ecol Risk Assess
21(6):1700-1713,	DOI:
10.1080/10807039.2014.974498
Wilson, R; Mitchell, I; Richardson, GM. (2016)
Estimation of dust ingestion rates in units
of surface area per day using a mechanistic
hand-to-mouth model. Hum Ecol Risk
Assess 22(4):874-881, DOI:
10.1080/10807039.2015.1115956.
Wong, MS. (1988) The role of environmental and
host behavioural factors in determining
exposure to infection with Ascaris
lumbricoldes and Trichuris trichiura.
Ph.D. Thesis, Faculty of Natural Sciences,
University of the West Indies.
Wong, EY;, Shirai, JH; Garlock, TJ; Kissel, JC.
(2000) Adult proxy responses to a survey of
children's dermal soil contact activities. J
Exp Anal Environ Epidem
10(6 Pt):509-517.
Yamamoto, N; Takahashi, Y; Yoshinaga, J; Tanaka,
A; Shibata, Y. (2006) Size distributions of
soil particles adhered to children's hands.
Arch Environ Contam Toxicol
51(2): 157—163.
Yeoh, B; Woolfenden, S; Lanphear, B; Ridley, GF;
Livingstone, N; Jorgensen, E. (2014).
Household interventions for preventing
domestic lead exposure in children.
Cochrane Database Syst Rev Dec 15 2014,
Issue 12. Art. No.: CD006047. DOI:
10.1002/14651858.CD006047.pub4.
Young, SL. (2010) Pica in pregnancy: New ideas
about an old condition. Ann Rev Nutr
30:403-422.
Zartarian, VG; Streicker, J; Rivera, A; Cornejo, CS;
Molina, S; Valadez, OF; Leckie, JO. (1995)
A pilot study to collect micro-activity data
of two- to four-year-old farm labor children
in Salinas Valley, California. J Expo Anal
Environ Epidemiol 5(1):21—34.
September 2017
Page 5-57

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
Table 5-3. Soil, Dust, and Soil + Dust Ingestion Estimates for Amherst, MA Study



Children (Ages 1 to <4 Years)






Ingestion (mg/day)

Tracer Element
N
Mean
Median
SD
95th Percentile
Maximum
Aluminum






soil
64
153
29
852
223
6,837
dust
64
317
31
1,272
506
8,462
soil/dust combined
64
154
30
629
478
4,929
Barium






soil
64
32
-37
1,002
283
6,773
dust
64
31
-18
860
337
5,480
soil/dust combined
64
29
-19
868
331
5,626
Manganese






soil
64
-294
-261
1,266
788
7,281
dust
64
-1,289
-340
9,087
2,916
20,575
soil/dust combined
64
^196
-340
1,974
3,174
4,189
Silicon






soil
64
154
40
693
276
5,549
dust
64
964
49
6,848
692
54,870
soil/dust combined
64
483
49
3,105
653
24,900
Vanadium






soil
62
459
96
1,037
1,903
5,676
dust
64
453
127
1,005
1,918
6,782
soil//dust combined
62
456
123
1,013
1,783
6,736
Yttrium






soil
62
85
9
890
106
6,736
dust
64
62
15
687
169
5,096
soil/dust combined
62
65
11
717
159
5,269
Zirconium






soil
62
21
16
209
110
1,391
dust
64
27
12
133
160
789
soil/dust combined
62
23
11
138
159
838
Titanium






soil
64
218
55
1,150
1,432
6,707
dust
64
163
28
659
1,266
3,354
soil/dust combined
64
170
30
691
1,059
3,597
N = Number of subjects






SD = Standard deviation.






Source: Calabrese et al. (1989).





September 2017
Page 5-58

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
Table 5-4. Amherst, MA Soil Pica Child's Daily Ingestion Estimates by Tracer and by Week
(mg/day)
Estimated Soil Ingestion (mg/day)
Tracer Element
Week 1
Week 2
Aluminum
74
13,600
Barium
458
12,088
Manganese
2,221
12,341
Silicon
142
10,955
Titanium
1,543
11,870
Vanadium
1,269
10,071
Yttrium
147
13,325
Zirconium
86
2,695
Source: Calabrese et al. (1991).
Table 5-5. Estimated Soil Ingestion for Sample of Washington State Children (2-7 years;
N = 101)a
Mean
Element (mg/day)
Median
(mg/day)
Standard Error of the
Mean
(mg/day)
Range
(mg/day )b
Aluminum 38.9
25.3
14.4
279.0 to 904.5
Silicon 82.4
59.4
12.2
-404.0 to 534.6
Titanium 245.5
81.3
119.7
-5,820.8 to 6,182.2
Minimum 38.9
25.3
12.2
-5,820.8
Maximum 245.5
81.3
119.7
6,182.2
a Excludes three children who did not provide any samples.
b Negative values occurred as a result of correction for nonsoil sources of the tracer elements. For aluminum, lower
end of range published as 279.0 mg/day in article appears to be a typographical error that omitted the negative sign.
Source: Adapted from Davis et al. (1990).



September 2017
Page 5-59

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
Table 5-6. Soil Ingestion Estimates for 64 Anaconda Children (Ages 1—4 Years)
Estimated Soil Ingestion (mg/day)



Percentile





Tracer
pt
50th
75th
90th
95th
Max
Mean
SD
Aluminum
-202.8
-3.3
17.7
66.6
94.3
461.1
2.7
95.8
Cerium
-219.8
44.9
164.6
424.7
455.8
862.2
116.9
186.1
Lanthanum
-10,673
84.5
247.9
460.8
639.0
1,089.7
8.6
1,377.2
Neodymium
-387.2
220.1
410.5
812.6
875.2
993.5
269.6
304.8
Silicon
-128.8
-18.2
1.4
36.9
68.9
262.3
-16.5
57.3
Titanium
-15,736
11.9
398.2
1,237.9
1,377.8
4,066.6
-544.4
2,509.0
Yttrium
—441.3
32.1
85.0
200.6
242.6
299.3
42.3
113.7
Zirconium
-298.3
-30.8
17.7
94.6
122.8
376.1
-19.6
92.5
BTM; median of
best 4 tracers
BTM; best tracer-
soil
NA
NA
-2.4
20.1
26.8
68.9
73.1
223.6
159.8
282.4
380.2
609.9
6.8
65.5
74.5
120.3
BTM; median of
best 4 tracers-
dust
BTM; best tracer-
dust
NA
NA
-5.5
26.8
62.8
198.1
209.2
558.6
353.0
613.6
683.9
1,499.4
16.5
127.2
160.9
299.1
BTM	= Best tracer methodology.
NA	= Not available.
SD	= Standard deviation.
Note:	Negative values are a result of limitations in the methodology.
Source:	Calabrese et al. (1997a).
Table 5-7. Soil Ingestion Estimates for Massachusetts Child Displaying Soil-Pica Behavior
(mg/day)
Study Day
Aluminum-Based Estimate
Silicon-Based Estimate
Titanium-Based Estimate
1
53
9
153
2
7,253
2,704
5,437
3
2,755
1,827
2,007
4
725
534
801
5
5
-10
21
6
1,452
1,373
794
7
238
76
84
Note: Negative values are a result of limitations in the methodology.
Source: Calabrese et al. (1997b).
September 2017
Page 5-60

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
Table 5-8. Average Daily Soil and Dust Ingestion Estimate for
Children Ages 1—3 Years (mg/day)
Type of
Estimate

Soil Ingestion

Dust Ingestion

Aluminum
Silicon Titanium
Aluminum
Silicon
Titanium
Mean
168
89 448
260
297
415
Median
7
0 32
13
2
66
SD
510
270 1,056
759
907
1,032
Range
-15 to 1,783
-46 to 931 -47 to 3,581
-39 to 2,652
-351 to 3,145
-98 to 3,632
SD = Standard deviation.
Note: N= 12. Negative values
are a result of limitations in the methodology.


Source: Calabrese et al. (1997b).




Table 5-9. Mean and Median Soil Ingestion (mg/day) by Family Members



Estimated Soil Ingestion3 (mg/day)

Participant Tracer Element
Mean
Median
SD
Maximum
Children (ages 3—7 Aluminum
36.7
33.3
35.4
107.9
years)b
Silicon
38.1
26.4
31.4
95.0

Titanium
206.9
46.7
277.5
808.3
Mother0
Aluminum
92.1
0
218.3
813.6

Silicon
23.2
5.2
37.0
138.1

Titanium
359.0
259.5
421.5
1,394.3
Fatherd
Aluminum
68.4
23.2
129.9
537.4

Silicon
26.1
0.2
49.0
196.8

Titanium
624.9
198.7
835.0
2,899.1
a
For some study participants, estimated soil ingestion resulted in a negative value
These estimates have been set

to 0 mg/day for tabulation and analysis.



b
Results based on 12 children with complete food, excreta, and soil data.


c
Results based on 16 mothers with complete food, excreta, and soil data.


d
Results based on 17 fathers with complete food, excreta, and soil data.


SD
= Standard deviation.




Source:
Davis and Mirick (2006).




September 2017
Page 5-61

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
Table 5-10. Positive/Negative Error (Bias) in Soil Ingestion Estimates in Calabrese et al.
(1989) Study: Effect on Mean Soil Ingestion Estimate (mg/day)a
Negative Error
Tracer
Lack of Fecal
Sample on Final
Study Day
Other Causeb
Total
Negative
Error
Total Positive
Error
Net Error
Original
Mean
Adjusted
Mean
Aluminum
14
11
25
43
+18
153
136
Silicon
15
6
21
41
+20
154
133
Titanium
82
187
269
282
+13
218
208
Vanadium
66
55
121
432
+311
459
148
Yttrium
8
26
34
22
-12
85
97
Zirconium
6
91
97
5
-92
21
113
a	How to read table: for example, aluminum as a soil tracer displayed both negative and positive error. The cumulative
total negative error is estimated to bias the mean estimate by 25 mg/day downward. However, aluminum has positive
error biasing the original mean upward by 43 mg/day. The net bias in the original mean was 18 mg/day positive bias.
Thus, the original 153 mg/day mean for aluminum should be corrected downward to 136 mg/day.
b	Values indicate impact on mean of 128 subject-weeks (64 children ages 1—4 years) in milligrams of soil ingested per
day.
Source: Calabrese and Stanek (1995).
September 2017
Page 5-62

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
Table 5-11. Comparison of Soil Ingestion Estimates (mg/day) from Two Sites

Min




Percentile




Max
Mean
SD

Pt
5th
10th
25th
50th
75th
90th
95th
99th
Amherst, MA; 64 children 1 to <4 years3













Median of Aluminum, Silicon, Titanium
-169
-79
-19
-11
6
30
72
188
253
435
11,874
147
1,048
Median of best 4 tracers
-97
-32
-13
-6
9
33
72
110
154
226
11,415
132
1,006
Best tracer
-12
-10
-3
1
10
34
58
100
217
2,782
11,874
176
1,083
Washington State; 101 children 2—7 yearsb
Median of Aluminum, Silicon, Titanium
Best tracer
^104
-131
-242
-59
-94
-22
-52
3
15
26
44
68
116
177
210
531
246
1,320
535
2,846
905
6,182
69
274
146
750
a Based on data from Calabrese et al. (1989).












b Based on data from Davis et al. (1990).













Max = Maximum.













Min = Minimum.













SD = Standard deviation.













Source: Stanek and Calabrese (1995a).













1
September 2017
Page 5-63

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
Table 5-12. Predicted Soil and Dust Ingestion Rates for Children Age 3 to <6 Years
(mg/day)






Percentile


Scenario
N
Mean
5
25
50
75
95
100
Dust ingestion/hand-to-mouth
1,000
19.8
0.6
3.4
8.4
21.3
73.7
649.3
Dust ingestion/object-to-mouth
1,000
6.9
0.1
0.7
2.4
7.4
27.2
252.7
Total dust ingestion3
1,000
27
ND
ND
13
ND
109
360
Soil ingestion/hand-to-mouth
1,000
41.0
0.2
5.3
15.3
44.9
175.6
1,367.4
Total ingestion
1,000
67.6
4.9
16.8
37.8
83.2
224.0
1,369.7
a Email from Haluk Ozkaynak (NERL, U.S. EPA) to Jacqueline Moya (NCEA, U.S. EPA) dated 3/8/11.
N = Number of model runs.
ND = data not provided.

Source: Ozkaynak et al. (2011).








September 2017
Page 5-64

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
Table 5-13. Age-Dependent Probability Density Functions Used to Estimate Dust and Soil Ingestion Rates via the Activity Pattern
Modeling Approach



Age Groups



Infants
Toddlers
Children
Teens
Adults
Seniors
Parameters
0-6 Months
7 Months-4 Years
5-11 Years
12-19 Years
20-59 Years
60+ Years
DSLhard (mg/cm2)
NA
AM 0.052 ± 0.065, LN
AM 0.052 ± 0.065, LN
AM 0.052 ± 0.065, LN
AM 0.052 ± 0.065, LN
AM 0.052 ± 0.065, LN
DSLsoft (mg/cm2)
AM 0.139 ± 0.305, LN
AM 0.139 ± 0.305, LN
AM 0.139 ± 0.305, LN
AM 0.139 ± 0.305, LN
AM 0.139 ± 0.305, LN
AM 0.139 ± 0.305, LN
ET (hr/d)
24 hr/d-ST
24 hr/d-ST-TO
24 hr/d-ST-TO
24 hr/d-ST-TOa
24 hr/d-ST-TOb
24 hr/d-ST-TOc
ST (hr/d)
12; 13; 15, TRI
10.5 ± 2.78, LN
9.9 ±2.6, LN
9.1 ±2.4, LN
8.4 ± 2.2, LN
8.5 ±2.2, LN
TO (hr/d)
NA
0; 1.2; 3.0, TRI
0; 2.2; 4.0, TRI
1.4 ± 1.2, LN
1.4 ± 1.3, LN
1.3 ± 1.4, LN
FQ (events/hr)
28 ± 22, LN
16 ± 9.9, LN
9.1 ±6.8, LN
1.0 ± 0.50, LN
1.0 ± 0.50, LN
1.0 ± 0.50, LN
FSAfingers (unitless)
0.05; 0.08; 0.10, TRI
0.04; 0.07; 0.10, TRI
0.04; 0.07; 0.10, TRI
0.04; 0.05; 0.06, TRI
0.04; 0.05; 0.06, TRI
0.04; 0.05; 0.06, TRI
FTSShard (unitless)
NA
0.7 ± 0.1, LN
0.7 ± 0.1, LN
0.4 ±0.1, LN
0.4 ±0.1, LN
0.4 ±0.1, LN
FTSSsoft (unitless)
0.14 ± 0.02, LN
0.14 ± 0.02, LN
0.14 ± 0.02, LN
0.08 ± 0.02, LN
0.08 ± 0.02, LN
0.08 ± 0.02, LN
SAhand (cm2)
160 ± 15, LN
215 ± 25, LN
295 ± 40, LN
400 ± 50, LN
445 ± 55, LN
450 ± 55, LN
SE (unitless)
0; 0.5; 1.0, TRI
0; 0.5; 1.0, TRI
0; 0.5; 1.0, TRI
0; 0.5; 1.0, TRI
0; 0.5; 1.0, TRI
0; 0.5; 1.0, TRI
SLhands (mg/cm2)
GM0.1 ± 1.8, LN
GM 0.1 ± 1.8, LN
GM0.1 ± 1.8, LN
GM0.1 ± 1.8,LN
GM 0.1 ± 1.8, LN
GM0.1 ± 1.8,LN
a	93.3% of teens were assumed to spend time outdoors and 6.7% were assumed to spend no time outdoors.
b	89.5% of adults were assumed to spend time outdoors and 10.5% were assumed to spend no time outdoors.
c	71.8% of seniors were assumed to spend time outdoors and 28.2% were assumed to spend no time outdoors.
AM	= Arithmetic mean.
DSL	= Dust surface loading.
ET	= Exposure time.
FQ	= Frequency of hand to mouth events.
FSA	= Fraction of surface area of hands.
FTSS	= Fraction of dust transferred from surfaces to skin.
GM	= Geometric mean.
LN	= Lognormal distribution.
NA	= Not applicable.
SA	= Surface area of the hand.
SE	= Saliva extraction fraction.
SL	= Soil loading.
ST	= Sleep time.
TO	= Time outdoors.
TRI	= Triangular distribution.
Source:	Wilson et al. (2013).
September 2017
Page 5-65

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
Table 5-14. Soil and Dust Ingestion Rates, Estimated Using a Probabilistic Activity Pattern
Modeling Approach3

mg/day, Mean ± Standard Deviation (95th
Percentile)
Age Group
Soil
Dustb
Soil + Dust
Infant (0 to 6 months)
NA
36 ± 130 (140)
36 (140)
Toddler (7 months to 4 years)
20 ± 26 (64)
41 ±71 (140)
61 (204)
Child (5 to 11 years)
23 ± 32 (75)
32 ±59 (110)
55(185)
Teenager (12 to 19 years)
1.5 ±2.6 (5.3)
2.2 ±3.6 (7.3)
3.7(12.6)
Adult (20 to 59 years)
1.6 ±2.9 (5.9)
2.6 ±4.2 (8.6)
4.2(14.5)
Senior (60+ years)
1.2 ±2.7 (4.8)
2.6 ±4.2 (8.7)
3.8(13.5)
a Based on 200,000 trials.
b Dust ingestion rate assuming 50% hard and 50% soft surfaces; except infants who were assumed to spend all their
indoor and outdoor awake time in contact with soft surfaces only.
Source: Wilson et al. (2013).



Table 5-15. Age-Specific Central Tendency Soil/Dust Ingestion Rates for Four Scenarios
That Best Predict Observed Blood Lead Levels (mg/day)a
Age (Years)
55/45b'c
40/30/30d
40/30/30e
50/25/10/15f
Average All
0.5 to <1
92
82
76
86
84
1 to <2
100
89
90
94
93
2 to <3
72
64
66
67
67
3 to <4
65
58
62
63
62
4 to <5
69
62
63
67
65
5 to <6
54
49
50
52
51
6 to <7
54
49
54
55
53
7 to <8
51
47
50
51
50
8 to <9
57
53
61
63
59
9 to <10
58
54
57
59
57
a	Atotal of 2,176 records of blood/soil/dust/lead concentrations were available for the analysis.
b	Geometric mean ingestion rate. The IEUBK model default of 55% dust and 45% soil (55/45)
c	Dust/yard soil.
d	Dust/yard/community soil. The original Bunker Hill Superfund Site model of 40% dust, 30% yard soil, and 30%)
community soil (40/30/30GM); geometric mean.
e	Dust/yard/community soil. The original Bunker Hill Superfund Site model of 40%o dust, 30%o yard soil, and 30%o
community soil (40/30/30AM); Arithmetic mean.
f	Dust/yard/neighborhood/community soil. The SEM using 50% dust, 25%o yard soil, 10%o neighborhood soil, and
15%o community soil (50/25/10/15); arithmetic mean.
Source: von Lindern et al. (2016).
September 2017
Page 5-66

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
Table 5-16. Age-Specific Distributions of Soil and Dust Intake Rates for the Four Partition
Scenarios (mg/day)a
Partition
Age
(years)




Percentile



N
5
10
25
50
75
90
95
55/45b
0.5 to <1
60
21
34
49
98
163
265
370

1 to <2
190
17
24
56
106
209
331
493

2 to <3
226
14
22
38
80
139
236
313

3 to <4
225
13
18
32
67
135
219
305

4 to <5
208
11
15
39
75
142
249
307

5 to <6
226
10
16
29
63
107
171
224

6 to <7
229
9
14
27
56
109
184
284

7 to <8
239
9
13
27
53
107
169
233

8 to <9
270
4
15
29
68
132
234
305

9 to <10
255
8
18
32
63
111
201
303
40/30/30GMc
0.5 to <1
60
22
34
46
89
138
210
298

1 to <2
190
18
30
56
91
159
262
323

2 to <3
226
14
22
37
66
113
190
229

3 to <4
225
13
18
35
60
111
160
206

4 to <5
208
12
19
37
66
118
178
240

5 to <6
226
11
15
26
55
94
140
166

6 to <7
229
9
15
26
56
93
149
217

7 to <8
239
9
14
27
51
88
132
185

8 to <9
270
3
19
30
61
110
185
231

9 to <10
255
9
20
32
61
98
169
212
40/30/30AMc,d
0.5 to <1
60
16
24
36
58
88
173
195

1 to <2
190
16
23
40
67
110
196
229

2 to <3
226
11
17
27
50
80
145
171

3 to <4
225
9
13
26
46
79
123
160

4 to <5
208
10
14
30
51
80
120
197

5 to <6
226
9
11
20
38
73
103
128

6 to <7
229
7
11
20
40
68
112
151

7 to <8
239
7
12
19
38
66
98
129

8 to <9
270
2
14
25
42
85
131
170

9 to <10
255
7
17
25
43
79
119
159
September 2017
Page 5-67

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
Table 5-16. Age-Specific Distributions of Soil and Dust Intake Rates for the Four Partition
Scenarios (mg/day) (Continued)
Percentile
Age		
Partition (years) N 5 10 25 50	75	90	95
50/25/10/15^ 0.5 to <1 54 17 27 38 72	94	165	221
1	to <2 174 16 22 42 69	123	188	250
2	to <3 202 10 19 28 53	82	140	178
3	to <4 209 10 14 26 47	76	130	156
4	to <5 192 11 15 32 53	86	122	182
5	to <6 208 10 12 23 41	74	102	126
6	to <7 218 7 11 21 41	68	116	171
7	to <8 228 7 12 21 41	68	105	120
8	to <9 258 2 14 25 44	80	134	170
9	to <10 245 8 17 25 43	80	116	171
a	A total of 2,176 records of blood/soil/dust/lead concentrations were available for analysis.
b	dust/yard soil.
c	dust/yard soil/community soil.
d	Partition scenarios that best fit the blood lead levels predicted by the IEUBK model to the observed blood lead levels.
e	dust/yard soil/neighborhood soil/community soil.
GM	= Geometric mean.
AM	= Arithmetic mean.
N	= Number of observations.
Source:	von Lindern et al. (2016).
Table 5-17. Estimated Daily Soil Ingestion for East Helena, MT Children Ages 1-
(N= 59)
-3 years
Estimation
Method
Mean
(mg/day)
Median
(mg/day)
Standard
Deviation
(mg/day)
Range
(mg/day)
95th Percentile Geometric Mean
(mg/day) (mg/day)
Aluminum
181
121
203
25-1,324
584
128
Silicon
184
136
175
31-799
578
130
Titanium
1,834
618
3,091
4-17,076
9,590
401
Minimum
108
88
121
4-708
386
65
Source: Binder et al. (1986).
September 2017
Page 5-68

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
Table 5-18. Estimated Soil Ingestion for Sample of Dutch Nursery School Children, Ages
2-4 Years

Sample
Child Number
Soil Ingestion as
Calculated from
Titanium
(mg/day)
Soil Ingestion as
Calculated from
Aluminum
(mg/day)
Soil Ingestion as
Calculated from AIR
(mg/day)
Limiting Tracer
(mg/day)
1
L3
L14
L25
103
154
130
300
211
23
107
172
103
154
23
2
L5
L13
L27
131
184
142
103
81
71
82
84
71
82
81
3
L2
L17
124
670
42
566
84
174
42
174
4
L4
Lll
246
2,990
62
65
145
139
62
65
5
L8
L21
293
313
—
108
152
108
152
6
L12
L16
1,110
176
693
362
145
362
145
7
L18
L22
11,620
11,320
77
120
120
77
8
LI
3,060
82
96
82
9
L6
624
979
111
111
10
L7
600
200
124
124
11
L9
133
—
95
95
12
L10
354
195
106
106
13
L15
2,400
—
48
48
14
L19
124
71
93
71
15
L20
269
212
274
212
16
L23
1,130
51
84
51
17
L24
64
566
—
64
18
L26
184
56
—
56
Arithmetic Mean
1,431
232
129
105
AIR
= No data.
= Acid insoluble residue.




Source
Adapted from Clausing et al. (1987).



September 2017
Page 5-69

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
Table 5-19. Estimated Soil Ingestion for Sample of Dutch Hospitalized, Bedridden

Children, Ages 2-
-4 Years


Soil Ingestion as Calculated
Soil Ingestion as Calculated


from Titanium
from Aluminum
Limiting Tracer
Child Sample
(mg/day)
(mg/day)
(mg/day)
1 G5
3,290
57
57
G6
4,790
71
71
2 G1
28
26
26
3 G2
6,570
94
84
G8
2,480
57
57
4 G3
28
77
28
5 G4
1,100
30
30
6 G7
58
38
38
Arithmetic Mean
2,293
56
49
Source: Adapted from Clausing et al. (1987).
Table 5-20. Van Wijnen et al. (1990) Limiting Tracer Method (LTM) Soil Ingestion
Estimates for Sample of Dutch Children
Age (years)
Sex

Daycare Center


Campground

N
GM LTM
(mg/day)
GSD LTM
(mg/day)
N
GM LTM
(mg/day)
GSD LTM
(mg/day)
Birth to <1
Girls
3
81
1.09
NA
NA
NA

Boys
1
75

NA
NA
NA
1 to <2
Girls
20
124
1.87
3
207
1.99

Boys
17
114
1.47
5
312
2.58
2 to <3
Girls
34
118
1.74
4
367
2.44

Boys
17
96
1.53
8
232
2.15
3 to <4
Girls
26
111
1.57
6
164
1.27

Boys
29
110
1.32
8
148
1.42
4 to <5
Girls
1
180

19
164
1.48

Boys
4
99
1.62
18
136
1.30
All girls

86
117
1.70
36
179
1.67
All boys

72
104
1.46
42
169
1.79
Total

162a
111
1.60
78b
174
1.73
a	Age and/or sex not registered for eight children; one untransformed value = 0.
b	Age not registered for seven children; geometric mean LTM value = 140.
GM	= Geometric mean.
GSD	= Geometric standard deviation.
LTM	= Limiting tracer method.
N	= Number of subjects.
NA	= Not available.
Source:	Adapted from Van Wijnen et al. (1990).
September 2017
Page 5-70

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
Table 5-21. Estimated Geometric Mean Limiting Tracer Method (LTM) Soil Ingestion
Values of Children Attending Daycare Centers According to Age, Weather Category,


and Sampling Period





First Sampling Period
Second Sampling Period



Estimated Geometric

Estimated Geometric



Mean

Mean



LTM Value

LTM Value
Weather Category
Age (years)
N
(mg/day)
N
(mg/day)
Bad
<1
3
94
3
67
(>4 days/week precipitation)
1 to <2
18
103
33
80

2 to <3
33
109
48
91

4 to <5
5
124
6
109
Reasonable
<1


1
61
(2-3 days/week precipitation)
1 to <2


10
96

2 to <3


13
99

3 to <4


19
94

4 to <5


1
61
Good
<1
4
102


(<2 days/week precipitation)
1 to <2
42
229



2 to <3
65
166



3 to <4
67
138



4 to <5
10
132


LTM = Limiting tracer method





N = Number of subjects.





Source: Van Wijnenetal. (1990).





September 2017
Page 5-71

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
Table 5-22. Items Ingested by Low-Income Mexican-Born Women
(Ages 18—42 Years) Who Practiced Pica during Pregnancy While in the
United States (N = 46)
Item Ingested

Number (%) Ingesting Items
Dirt

11 (24)
Bean stones3

17 (37)
Magnesium carbonate

8(17)
Ashes

5(11)
Clay

4(9)
Ice

18(39)
Otherb

17 (37)
a Little clods of dirt found among unwashed beans.
b Including eggshells, starch, paper, lipstick, pieces of clay pot, and adobe.
N = Number of individuals reporting pica behavior.
Source: Simpson et al. (2000).


Table 5-23. Soil Ingestion Rates for the Four Most Reliable Tracers (Aluminum, Silicon,
Lanthanum, and Cerium), Aluminum and Silicon Combined, and All Four Tracers
Combined (mg/day)
Tracer N
Mean
SD
50th Percentile
75th Percentile 90th Percentile
Aluminum 43
36.9
51.9
31
61
110
Cerium 43
72.2
179.5
51
142
217
Lanthanum 43
132.6
158.6
104
211
343
Silicon 30
49.4
73.7
40
124
145
Aluminum and 73
silicon
42.0
61.6
32
89
124
All 4 tracers 159
74.7
119.5
50
130
211
iV = Number of fecal samples from the
SD = Standard deviation.
seven adult subjects.



Source: Doyle et al. (2012).





September 2017
Page 5-72

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
Table 5-24. Soil Ingestion Rates for Aluminum, Silicon, and the Four Most
Reliable Tracers (Aluminum, Silicon, Lanthanum, and Cerium) Combined
(mg/day)
Tracer
N
Mean
SD
50th Percentile 90th Percentile
Aluminum
87
36
117
7
161
Silicon
87
68
152
37
231
Aluminum and silicon
87
52
119
37
220
All 4 tracers
87
32
88
18
152
N = Number of fecal samples from the nine
SD = Standard deviation.
adult subjects.


Source: Irvine et al. (2014).





Table 5-25. Estimated Soil Ingestion for Six Jamaican

Children Displaying Soil Pica"

Child
Month
Estimated Soil Ingestion (mg/day)
11
1

55

2

1,447

3

22

4

40
12
1

0

2

0

3

7,924

4

192
14
1

1,016

2

464

3

2,690

4

898
18
1

30

2

10,343

3

4,222

4

1,404
22
1

0

2

—

3

5,341

4

0
27
1

48,314

2

60,692

3

51,422

4

3,782
a Based on study of children 0.3 to 14 years.


— = No data.



Source: Calabrese and Stanek (1993).


September 2017
Page 5-73

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
Table 5-26. Distribution of Average (Mean) Daily Soil Ingestion Estimates per Child for 64
Children, Ages 1 to <4 Years3 (mg/day)
Type of
Estimate Overall Aluminum Barium Manganese Silicon Titanium Vanadium Yttrium Zirconium
Number of
samples
64
64
33
19
63
56
52
61
62
Mean
179
122
655
1,053
139
271
112
165
23
25th Percentile
10
10
28
35
5
8
8
0
0
50th Percentile
45
19
65
121
32
31
47
15
15
75th Percentile
88
73
260
319
94
93
177
47
41
90th Percentile
186
131
470
478
206
154
340
105
87
95th Percentile
208
254
518
17,374
224
279
398
144
117
Maximum
7,703
4,692
17,991
17,374
4,975
12,055
845
8,976
208
a	For each child, estimates of soil ingestion were formed on days 4-8, and the mean of these estimates was then
evaluated for each child. The values in the column "overall" correspond to percentiles of the distribution of these
means over the 64 children. When specific trace elements were not excluded via the relative standard deviation
criteria, estimates of soil ingestion based on the specific trace element were formed for 108 days for each subject.
The mean soil ingestion estimate was again evaluated. The distribution of these means for specific trace elements is
shown.
Source: Stanek and Calabrese (1995b).
Table 5-27. Estimated Distribution of Individual Mean Daily Soil
Ingestion Based on Data for 64 Subjects (Ages 1 to <4 Years)
Projected over 365 Days3
Range

1-2,268 mg/dayb
50th Percentile (median)

75 mg/day
90th Percentile

1,190 mg/day
95th Percentile

1,751 mg/day
a Based on fitting a log-normal distribution to model daily soil ingestion values.
b Subj ect with pica excluded.
Source: Stanek and Calabrese (1995b).

September 2017
Page 5-74

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
Table 5-28. Distribution of Daily Soil Ingestion (mg/day) Over 7 Days,
64 Children (Ages 1-4 years) from Anaconda, MTa
„ , Over elements withm a , ,
Over days , Mean
day
SD


Percentile


Max
25th
50th
75th
90th
95th
Median
Medianb 13
49
14
8
30
82
107
136
Mean
Medianb 31
56
-3
17
53
111
141
219
Median
Mean0 14
59
-14
4
26
120
128
151
Mean
Mean0 36
72
-7
16
72
151
160
283
a
b
c
Based on 7 of 8 tracer elements (excluding titanium), and 28-hour lag time.
Estimates correspond to the median from the 7 trace elements for each subject day.
Estimates correspond to the mean.



Source:
Stanek and Calabrese (2000).







September 2017
Page 5-75

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
Table 5-29. Prevalence of Nonfood Consumption by Substance from NHANES I and
NHANES II
Substance
NHANES I (age 1-74 years)
N (sample size) = 20,724 (unweighted);
	193,716,939 (weighted)
NHANES II (age 6 months-74 years)
N (sample size) = 25,271 (unweighted);
	203,432,944 (weighted)
N
Unweighted
(Weighted)
Prevalence3
95%
Confidence
Interval
N
Unweighted
(Weighted)
Prevalence3
95%
Confidence
Interval
Any nonfood
substance
Clay
Starch
Paint and plaster
Dirt
Dirt and clay
Other
732
(4,900,370)
131
(582,101)
39
(195,764)
385
(2,466,210)
190
(1,488,327)
2.5%
0.3%
0.5%b
1.3%
0.8%
2.2-2.9%
0.2-0.4%
0.3-0.7%
1.1-1.5%
0.6-0.9%
480
(2,237,993)
46
(223,361)
61
(450,915)
55
(213,588)
216
(772,714)
218
(1,008,476)
1.1%
0.1%
0.2%
0.6%c
2. l%d
0.5%
1.0-1.2%
0.1-0.2%
0.1-0.3%
0.4-0.8%
1.7-2.5%
0.4-0.6%
a	Prevalence = Frequency (n) (weighted) Sample Size (N) (weighted).
b	NHANES I sample size (<12 years): 4,968 (unweighted); 40,463,951 (weighted).
c	NHANES II sample size (<12 years): 6,834 (unweighted); 37,697,059 (weighted).
d	For those aged <12 years only; question not prompted for those >12 years.
Unweighted = raw counts.
Weighted = adjusted to account for the unequal selection probabilities caused by the cluster design, item nonresponse,
planned oversampling of certain subgroups, and adjustments to ensure data are representative of the civilian
noninstitutionalized census population in the coterminous United States.
Source:
Gavrelis et al. (2011).
September 2017
Page 5-76

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
Table 5-30. Results of Meta-Analysis on Soil Ingestion (mg/day)
Population Group
Number
of Studies
Number
of
Subjects
Number
of
Subject-
Weeks
Meana
SE
95%
UCL
50th
Percentile
95th
Percentile
Adjusted
Meanb
Subjects
4
216
266
25.5
15.5
73.1
32.6
79.4
31.3
Male
4
114
137
27.7
15.5
74.7
27.1
88.8
31.2
Female
4
102
129
22.2
18.4
78.7
33.5
66.7
31.3
Age 1 to <2 years
2
39
58
3.8
11.8
154
9.2
18.1
6.8
Age 2 to <3 years
3
55
76
20.6
24.0
115
21.9
94.5
28.3
Age 3 to <4 years
4
47
57
32.2
22.3
104
57.0
71.3
46.3
Age 4 to <5 years
3
75
75
40.9
17.0
257
36.2
104
41.3
Excluding Anaconda
data0
3
156
206
43.4
4.5
52.3
40.8
90.0
43.7
All subjects except pica
childd
4
240
303
36.5
17.1
88.7
39.4
114
43.6
a Calculation of the mean includes negative values for some subjects.
b Estimates of soil ingestion less than zero are set equal to zero.
c Excludes data from the Anaconda study (Calabrese et al., 1997a) because the children lived near a Superfund site and it could
be assumed that soil ingestion at that site could be different from at other sites because additional effort may be taken to limit
soil ingestion at this site.
d Excludes subject exhibiting pica behavior.
Source: Stanek et al. (2012a).








September 2017
Page 5-77

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion

Table 5-31. Age-Dependent Probability Density Functions Used to Estimate Sediment Ingestion Rates Using an Activity Pattern


Modeling Approach3




Age Groups


Toddlers
Children
Teens

Parameters 7 months-4 years
5-11 years
12-19 years.
Adults Seniors
FQ (events/hr) 16 ±9.9, LN
9.1 ±6.8, LN
3.0 ± 1.5, LN
3.0 ± 1.5, LN 3.0 ± 1.5, LN
FSAflngers
(unitless) 0.04; 0.07; 0.10, TRI
0.04; 0.07; 0.10, TRI
0.04; 0.05; 0.06, TRI
0.04; 0.05; 0.06, TRI 0.04; 0.05; 0.06, TRI
SAhand (cm2) 215 ± 25, LN
295 ± 40, LN
400 ± 50, LN
445 ± 55, LN 450 ± 55, LN
SE (unitless) 0; 0.5; 1.0, TRI
0; 0.5; 1.0, TRI
0; 0.5; 1.0, TRI
0; 0.5; 1.0, TRI 0; 0.5; 1.0, TRI
All Ages
SLhands (mg/cm2)
GMOi
58 (95% CI = 0.35 to 2.2), LN

SS (mg/L)

39; 847; 5,146, TRI

SWIR (L/hr)
AM 0.0037 (95th percentile 0.0112), LN

a
Based on 200,000 trials.



AM
= Arithmetic mean.



CI
= Confidence interval.



FQ
= Frequency of events.



FSA
= Fraction of surface area of hands.



GM
= Geometric mean.



LN
= Lognormal distribution.



SA
= Surface area.



SE
= Saliva extraction.



SLhands
= Sediment adherence factor.



SS
= Suspended sediment concentration.



SWIR
= Surface water ingestion rate.



TRI
= Triangular distribution.



Source:
Wilson et al. (2015).



September 2017
Page 5-78

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
Table 5-32. Estimated Sediment Ingestion Rates (mg/hour) Using an Activity Pattern
Modeling Approach
Age Group
Deterministic Estimate Probabilistic Estimate3
Arithmetic Mean Arithmetic Mean ± SD (95th percentile)
Sediment Ingestion Due to Hand-to-Mouth Contact
Toddlers (7 months^ years)
59
72 ± 89 (300)
Children (5-11 years)
46
57 ± 78 (250)
Teens (12-19 years)
15
18 ± 20 (70)
Adults (20-59 years)
16
20 ±23 (80)
Seniors (60+ years)
17
20 ±23 (80)
Sediment Ingestion Due to Surface Water Intake
All age groups
3.1
7.7 ± 89 (44)
a Based on 200,000 trials.
SD = Standard deviation.


Source: Wilson et al. (2015).


Table 5-33. Dust Ingestion Rates at Residential Settings" Based on an Activity Pattern
Modeling Approach

Mean ± SD (95th percentile) Dust Ingestion Rates (m2/d)
Age Group
100% Hard Surfaces
100% Soft Surfaces
50% Hard/50%) Soft Surfaces
Infant (0-6 months)
	b
0.025 ±0.024 (0.088)
0.025 ±0.024 (0.088)b
Toddler (7 months-4 years)
0.10 ±0.092 (0.34)
0.020 ±0.018 (0.067)
0.061 ±0.055 (0.20)
Child (5-11 years)
0.078 ±0.081 (0.29)
0.016 ±0.016 (0.059)
0.047 ±0.048 (0.18)
Teen (12-19 years)
0.0053 ±0.0042 (0.016)
0.0011 ±0.00084(0.0032)
0.0032 ± 0.0025 (0.0094)
Adult (20-59 years)
0.0062 ±0.0050 (0.019)
0.0012 ±0.00098 (0.0038)
0.0037 ±0.0029 (0.0093)
Senior (60+ years)
0.0063 ±0.0051 (0.019)
0.0013 ±0.0010 (0.0039)
0.0038 ±0.0030 (0.012)
a Based on probabilistic approach (200,000 trials).
b Infants were assumed to contact soft surfaces only.


Source: Wilson et al. (2016).



September 2017
Page 5-79

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
Table 5-34. Summary of Estimates of Soil and Dust Ingestion by General Population Children and Adults from Key Studies Using
the Tracer Study, Biokinetic Modeling, and Activity Pattern Methodologies (mg/day)a

Sample





Population
Size
Age
Mean
p50
p95
Reference
Tracer Studies
Amherst, MA
64
1 to <4 years
132b
33b
154b
Calabrese et al. (1989) reanalyzed by






Stanek and Calabrese (1995a)
Southeastern, WA
101
2 to 7 years
69°
44c
246c
Davis et al. (1990) reanalyzed by Stanek






and Calabrese (1995a)
Anaconda, MT
64
1 to 4 years
66d
20d
282d
Calabrese et al. (1997a)
Western MA
12
1 to 3 years
129e
4
NR
Calabrese et al. (1997b)
Southeastern, WA
33
Adults
52
7
NR
Davis and Mirick (2006)
Tracer studies

1 to <8 years
99f

227§

Age-specific averages

Adults
52

NR

Biokinetic Model Comparison Studies
Lead smelting site: Illinois
31
0.5 to <1 year
61h
NR
NR
Hogan et al. (1998)
Lead smelting sites: Pennsylvania, Illinois,
440
1 to <7 years
113
NR
NR
Hogan et al. (1998)
Kansas/Missouri






Bunker Hill site, Idaho
60
0.5 to <1 year
811
65
2081
von Lindern et al. (2016)
Bunker Hill site, Idaho
190
1 to <2 years
921
68
2401
von Lindern et al. (2016)
Bunker Hill site, Idaho
885
2 to <6 years
611
47
1621
von Lindern et al. (2016)
Bunker Hill site, Idaho
1,075
1 to <6 years
67«
52
1781J
von Lindern et al. (2016)
Bunker Hill site, Idaho
993
6 to <10 years
561
42
1551
von Lindern et al. (2016)
Biokinetic model comparison studies

0.5 to <1 year
71

208

Age-specific averages

1 to <2 years
92

240



2 to <6 years
62

163



1 to <6 years
91k

178



6 to <10 years
57

155

September 2017
Page 5-80

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
Table 5-34. Summary of Estimates of Soil and Dust Ingestion by General Population Children and Adults from Key Studies Using
the Tracer Study, Biokinetic Modeling, and Activity Pattern Methodologies (mg/day)a
(Continued)
Activity Pattern Modeling Studies
Simulated population 1,000'
3 to <6 years
68
38
224
Ozkaynak et al. (2011)
Canada —m
0 to <7 months
36n
NR
140
Wilson etal. (2013)
Canada —m
7 months to <5 years
61n
NR
204
Wilson etal. (2013)
Canada —m
5 to <12 years
55n
NR
185
Wilson etal. (2013)
Canada —m
12 years through adults
4n
NR
14
Wilson etal. (2013)
Activity pattern modeling studies
0 to <7 months
36

140

Age-specific averages
7 months to <5 years
65°

214p


5 to <12 years
55

185


12 years through adults
4

13

All study types
0 to <7 months
36

140
Wilson et al. (2013)
Age-specific averages
0.5 to <1 year
71

208
Hogan et al. (1998) (mean only); von





Lindern et al. (2016)

1 to <2 years
92

240
von Lindern et al. (2016)

2 to <6 years
61

162
von Lindern et al. (2016)

1 to <6 years
84

206
Calabrese et al. (1989) reanalyzed by





Stanek and Calabrese (1995a); Davis





et al. (1990) reanalyzed by Stanek and





Calabrese (1995a) (mean only);





Calabrese et al. (1997a); and Calabrese





et al. (1997b); Hogan et al. (1998);





Ozkaynak et al. (2011); von Lindern et





al. (2016); Wilson et al. (2013)

6 to <12 years
56

170
Wilson et al. (2013); von Lindern et al.





(2016)

12 years through adults
28

—
Davis and Mirick (2006); Wilson et al.





(2013)
Recommended values
0 to <6 months
40

100

(rounded to one significant figure)
6 months to <1 year
70

200


1 to <2 years
90

200


2 to <6 years
60

200


1 to <6 years
80

200


6 to <12 years
60

200


12 years through
30

100r


adults'




September 2017
Page 5-81

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
Table 5-34. Summary of Estimates of Soil and Dust Ingestion by General Population Children and Adults from Key Studies Using
the Tracer Study, Biokinetic Modeling, and Activity Pattern Methodologies (mg/day)a (Continued)
a
See Appendix B for additional details.
b
Estimates adjusted by Stanek and Calabrese (1995a) based on data from Calabrese et al. (1989) using BTM (average of the median of the four best tracers for each

child).
c
Estimates from Davis et al. (1990) were adjusted by Stanek and Calabrese (1995a) using the BTM (average of the median of three tracers for each child).
d
Estimates based on BTM (average of the best tracer for each child).
e
Estimates based on aluminum and silicon.
f
Average of the means.
g
Average of the 95th percentiles.
h
Adjusted from model default of 85 mg/day under the assumption that the geometric mean model predicted blood lead level was higher than the geometric mean blood

lead by a factor of 1.4 due only to individual soil + dust ingestion rates.
1
Average of two best-fit models from Table 5-15.
j
Average of ages 1 to <2, 2 to <3, 3 to <4,4 to <5, 5 to <6 years.
k
Average of 113 and 68 mg/day.
1
Simulations.
m
Wilson et al. (2013) data based on 200,000 trials.
n
Does not include object-to-mouth exposure to soil and dust.
0
Average of 68 and 61 mg/day.
P
Average of 224 and 204 mg/day.
q
Soil + dust ingestion rates may be higher for adults following a traditional rural or wilderness lifestyle (see Sections 5.3.4.20 and 5.3.4.21). Based on Doyle et al. (2012)

and Irvine et al. (2014) the central tendency adult soil + dust ingestion rates is 50 mg/day (20 mg/day soil and 30 mg/day dust) and the upper percentile rate is 200

mg/day (90 mg/day soil and 100 mg/day dust).
r
Upper percentile value for adults estimated by multiplying the average of the ratios of 95th percentiles to means for all other age groups times the adult central tendency

estimate (i.e., 30 mg/day x 3.2 = 100 mg/day).
NR
= Not reported.
P
= Percentile.
September 2017
Page 5-82

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
30%
25%
20%
15%
10%
5%
0%
CM
V
V
9
v*
V
2
2
2
2
v-
CO
€0
CD



CM
V
2
CO
Ag© Category
CO
v
a
csi
fc RSSfc*?
i I I I i I
$
a
to
5
A
NHANES I
fHANES II
Figure 5-1. Prevalence of Nonfood Substance Consumption by Age, NHANES I and
NHANES II.
Source: Gavrelis et al. (2011).
September 2017
Page 5-83

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
8.0%
7.0%
6.0%
2 5,0%
| 4.0%
>
£ 3.0%
a.
2.0%
1.0%
0.0%
~ White
[~1 Black
NHANESI
NHANES I!
Figure 5-2. Prevalence of Nonfood Substance Consumption by Race, NHANES I and
NHANES II.
Source: Gavrelis et al. (2011).
September 2017
Page 5-84

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
s*
4.0%
3.5%
3.0% -

2,0% -
1.5%
0.5% -
0,0%

NHANESi
NHANES II
Figure 5-3. Prevalence of Nonfood Substance Consumption by Income, NHANES I
(1971-1975) and NHANES II (1976-1980).
Source: Gavrelis et al. (2011).
September 2017
Page 5-85

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
APPENDIX A
Table A-l. Terms Used in Literature Searches
Soil ingestion
Dust ingestion
Soil/dust ingestion prevalence
Pica
Soil pica
Geophagy
Indoor settled dust
Outdoor settled dust
Tracer element methodology
Biokinetic soil/dust model/methodology
Activity pattern soil/dust model/methodology
Chalk/dirt/starch/kaolin/magnesium carbonate/pottery/plaster/paint chip eating/ingestion
Ingestion of nonfood substances
Vermeer DE
Frate DA
Davis S
Mirick D
Calabrese EJ
Stanek EJ
HoganK
September 2017
Page A-l

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
APPENDIX B
Table B-l. Distributions of Soil Ingestion Rates (mg/day) Based on Different Methods of Analyzing Data from the
Tracer Studies
Summary
Statistic
Amherst, MA (N = 64; children 1 to <4 years)
Calabrese et al. (1989)
Washington State
(N= 101; children 2-
years) Davis et al.
(1990)
Anaconda, MT (N= 64; children 1-4 years)
Calabrese et al. (1997a)
Western MA (N= 12;
children 1-3 years)
Calabrese et al.
(1997b)
Average of
Median
Al, Si, Tia
Average of
Median
of Best 4
Tracers3
Average
of Best
Tracer8
Daily
Averageb
Average of
Median
Al, Si, Tic
Average
of Best
Tracer0
Average of
Median
of Best 4
Tracers'1
Average
of Best
Tracer4
Daily
Average6
Average of Mean
values for
Al and Sif
Min
<0
<0
<0
—
<0
<0
<0
0
<0
—
Max
11,874
11,415
11,874
7,703
905
6,182
380
610
219
—
Mean
147
132
176
179
69
274
7
66
31
129
SD
1,048
1,006
1,083
—
146
750
75
120
56
510 (Al); 270 (Si)
Percentile










5th
<0
<0
<0
—
<0
<0
<0
<0
<0

10th
<0
<0
1
10
<0
3
<0
<0
<0

25th
6
9
10
45
15
26
<0
2
<0

50th
30
33
34
88
44
68
<0
20
17
7 (Al); 0 (Si)
75th
72
72
58
—
116
177
27
69
53

90th
188
110
100
186
210
530
73
224
111

95th
253
154
217
208
246
1,320
160
282
141

Based on a reanalysis of the Calabrese et al. (1989) data using theBTM. Median ofAl, Si, andTi; median of best 4 of 8 tracers (i.e., 4 lowest F/S ratios);
or best tracer (lowest F/S ratio) (see Table 6, Stanek and Calabrese, 1995a). The average of the best 4 tracers for each child was used in developing the
recommended soil and dust ingestion rates.
Based on average (mean) daily soil ingestion estimates (mg/day) per child (Table 5, Stanek and Calabrese, 1995b).
Best Tracer Method; median of Al, Si, and Ti or best tracer (Table 9, Stanek and Calabrese, 1995a). The average of the 3 tracers for each child was used
in developing the recommended soil and dust ingestion rates.
Best Tracer Method; median of best 4 of 5 tracers (i.e., lowest F/S ratios) or best tracer for a given subject-week (Table 12, Calabrese et al., 1997a). The
average of the best tracer for each child was used in developing the recommended soil and dust ingestion rates.
Based on the mean of daily soil ingestion estimates based on the median values for 7 trace elements (Table 2, Stanek and Calabrese, 2000).
Average of mean soil ingestion values (Table 3, Calabrese et al., 1997b).
September 2017
Page B-l

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
APPENDIX C
Table C-l. Key Soil Ingestion Studies Used in Developing Soil + Dust Ingestion Recommendation for Use in Risk Assessment
Tracer Studies
Reference
Location
Population
Tracers
Study Design
Results
Calabrese et al.
(1989);	Barnes
(1990)
Amherst MA
64 children
(1 to <4 years)
Davis et al.
(1990)
3-city area in
southeastern
Washington
101 children
(2 to 7 years)
Calabrese et al.
(1997a)
Anaconda, MT
aluminum,
barium,
manganese,
silicon,
titanium,
vanadium,
yttrium,
zirconium
aluminum,
silicon, titanium
64 children (1 to 4
years) at a Superfund
site
aluminum,
cerium,
lanthanum,
neodymium,
silicon,
titanium,
yttrium,
zirconium
Duplicate samples of food,
beverages, medicines, vitamins,
excreta collected over 2-week
period; soil/dust samples from
children's homes/play areas;
participants supplied with
toothpaste, baby cornstarch, diaper
rash cream, and soap with low
levels of most tracer elements;
fecal/urine samples collected.
Collected soil/house dust,
duplicate food, dietary
supplements/medications, and
mouthwash samples over 7 days;
urine/feces collected over 4 days;
toothpaste with known tracer
element content was supplied;
information on dietary habits and
demographics collected.
Duplicate samples of
meals/beverages and over-the-
counter medicines/vitamins
collected; feces collected over 7
days; soil/and dust collected from
the children's homes/play areas;
toothpaste containing
nondetectable tracer levels (except
silica) provided; infants provided
with baby cornstarch, diaper rash
cream, and soap with low levels of
tracers.
Mean soil + dust ingestion ranged from —496 mg/day based on
manganese to 483 mg/day based on silicon. The 95th percentiles
range from 159 for both yttrium and zirconium to 3,174 for
manganese.
Stanek and Calabrese (1995a) reanalyzed the data using the best
tracer method (BTM) and the lowest four food: soil ratios for each
child, calculated on a per-week ("subject-week") basis. Based on the
median of soil ingestion estimates from the best four tracers, the
mean soil ingestion rate for children was 132 mg/day and the median
was 33 mg/day. The 95th percentile value was 154 mg/day.
Mean soil ingestion rates were 39 mg/day for aluminum, 82 mg/day
for silicon, and 245 mg/day for titanium; median values were
25 mg/day for aluminum, 59 mg/day for silicon, and 81 mg/day for
titanium. Adjusted mean soil + dust ingestion: 65 mg/day for
aluminum, 160 mg/day for silicon, 268 mg/day for titanium; median
values were: 52 mg/day for aluminum, 112 mg/day for silicon,
117 mg/day for titanium.
Stanek and Calabrese (1995a) reanalyzed the data using the BTM
and the lowest four food: soil ratios for each child, calculated on a
per-week ("subject-week") basis. The soil + dust ingestion values
were 69 mg/day (mean), 44 mg/day (median), and 246 mg/day
(95th percentile).
Mean ranged from —544 mg/day based on titanium to 270 mg/day
based on neodymium; 95th percentile estimates ranged from
69 mg/day based on silicon to 1,378 mg/day based on titanium.
Using the BTM (average of best tracer for each child), the mean
value for soil was 66 mg/day; the median was 20 mg/day; and the
95th percentile value was 282 mg/day.
September 2017
Page C-l

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
Table C-l. Key Soil Ingestion Studies Used in Developing Soil + Dust Ingestion Recommendation for Use in Risk Assessment (Continued)
Calabrese et al.
(1997b)
Western
Massachusetts
Davis and
Mirick (2006)
3-city area in
southeastern
Washington
12 children 1 to 3
years old observed to
have frequent soil
ingestion in previous
study.
Subset of Davis et al.
(1990): 33 adults
aluminum,
silicon, titanium
Mass balance tracer study with
duplicate food sampling; both soil
and dust samples collected.
aluminum,
silicon, titanium
Duplicate samples of
food/medications; feces collected
for 11 consecutive days; urine
samples collected; soil/house dust
samples collected.
Estimates calculated based on soil tracer element concentrations only
for the 12 subjects ranged from -15 to+1,783 mg/day based on
aluminum, -46 to +931 mg/day based on silicon, and -47
to +3,581 mg/day based on titanium. Mean soil estimates: 168
mg/day based on aluminum, 89 mg/day based on silicon, 448 mg/day
based on titanium. Mean dust ingestion estimates: 260 mg/day based
on aluminum, 297 mg/day based on silicon, 415 mg/day based on
titanium. Based on the average of aluminum and silicon, the mean
estimate is 129 mg/day. One child exhibited pica behavior.
Mean for the three tracers ranged from 23 to 625 mg/day; calculated
by setting negative estimates to zero. Based on the average of
aluminum and silicon, the mean and median values are 52 mg/day
and 7 mg/day respectively.
Biokinetic Modeling Comparison Studies
Reference
Location
Population
Studied
Study Design
Results
Hogan et al.
(1998)
Historic lead
smelting
communities:
Palmerton, PA;
southeastern
Kansas and
southwestern
Missouri;
Madison Co.,
IL.
478 children ages 0.5
to <7 years
Pennsylvania,
Illinois, and
Kansas/Missouri with
blood lead
measurements and
related soil and dust
lead levels; 31 of
these children from
the Illinois site were
0.5 to <1 year old
Compared IEUBK-predicted blood lead levels with
observed blood lead levels using observed house
dust/soil lead levels, and default soil and dust intake
rates, and other model parameters.
The default IEUBK model mean soil + dust soil intake rates
averaged over ages 1 to <7 years was approximately 135 mg/day.
The geometric mean blood lead levels at one site were slightly
over-predicted by the model; blood lead levels were slightly
under-predicted at a second site, and the blood lead levels predicted
by the model were roughly accurate at the third site. The default
IEUBK model mean soil + dust soil intake rates averaged over ages
0.5 to <1 year was approximately 85 mg/day. For children 6 to 12
months old in the Illinois site only, Hogan et al. (1998) reported a
predicted geometric mean blood lead level 1.4-fold higher than the
geometric mean blood lead levels observed.
September 2017
Page C-2

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
Table C-l. Key Soil Ingestion Studies Used in Developing Soil + Dust Ingestion Recommendation for Use in Risk Assessment (Continued)
Von Lindern et
al. (2016)
Northern Idaho;
site of
community-wide
soil remediation
3,203 children (ages
0 to <10 years) with
blood lead
measurements and
related lead levels in
yard, neighborhood
and community soil,
and house dust.
Compared IEUBK-predicted blood lead levels with
observed blood lead levels; the soil and dust
ingestion values used in the model were developed
using a statistical model that apportioned the
contributions of community soils, yard soils of the
residence, and house dust to lead intake; soil + dust
ingestion was estimated based on four partition
scenarios that used different combinations of the
proportions of dust + soil attributed to different
sources (e.g., dust, yard soil, neighborhood soil).
All four partition scenarios produced similar central tendency intake
rates. For children ages 0.5 to <10 years, the mean soil and dust
ingestion rates ranged from 47 mg/day to 100 mg/day. Among the
age groups evaluated, children ages 1 to <2 years had the highest
mean soil and dust ingestion rates (i.e., 89-100 mg/day).
The 95th percentile soil and dust intake rates ranged from 120 mg/day
to 493 mg/day for ages 0.5 to <10 years of age. Based on the
average of all four scenarios, the mean soil + dust ingestion rates
ranged from 50 mg/day for 7- to <8-year-olds to 93 mg/day for 1- to
<2-year-olds.
Activity Pattern Modeling Studies
Reference
Location
Population
Studied
Study Design
Results
Ozkaynak et al.
(2011)
United States Simulated population Used EPA's SHEDS-Multimedia model to estimate Mean total soil and dust ingestion: 68 mg/day; approximately 60%
of children ages 3 to
<6 years.
Wilson et al.
(2013)
Canada
Simulated
populations of infants
aged 0 to <7 months;
toddlers 7 months to
<5 years; children 5
years to <12 years;
teens 12 years to <20
years; adults 20 years
to <60 years; seniors
60+ years.
soil and dust ingestion rates using distributions of
exposure factor values for hand-to-mouth activities;
assumed soil and dust adhered to hands and
remained until washed off or ingested by mouthing;
object-to-mouth pathway for soil/dust ingestion was
also addressed; outdoor matter was designated as
"soil" and indoor matter as "dust."
Modeling approaches (deterministic and
probabilistic) used to estimate soil + dust via
hand-to-mouth contact; object-to-mouth exposures
were not considered. The models used measures of
particle loading to indoor surfaces, the fraction
transferred to the hands, hand surface areas, the
fraction of hand surface area that may be mouthed or
contact food, the frequency of hand-to-mouth
contacts, the amount dissolved in saliva, and the
exposure time. Model parameters used were
representative of the Canadian population. Contact
was assumed to occur with hard surfaces 50% of the
time and with soft surfaces 50% of the time, except
for infants for whom contact was assumed to occur
with soft surfaces only.
originating from soil ingestion, 30% from dust on hands, and 10%)
from dust on objects; 95th percentile: 224 mg/day. The predicted soil
and dust ingestion values fit a log-normal distribution.
Mean soil + dust ingestion rates ranged from 4 mg/day for teens,
adults, and seniors to 61 mg/day for toddlers (7 months to <5 years
of age). The 95th percentile soil + dust ingestion rates ranged from
14 mg/day for teens, adults, and seniors to 204 mg/day for toddlers
(7 months to <5 years of age). Infants (0 to <7 months in age) were
assumed to consume dust only at mean rate of 36 mg/day
(95thpercentile: 140 mg/day).
Source: Adapted from Moya and Phillips (2014).
September 2017	page C-3

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
APPENDIX D
Table D-l. Studies on the Prevalence of Ingesting Soil, Dust, or Other Nonfood Substances
Reference
Location
Population
Results
Dickins and Ford Oktibbeha
(1942)
Ferguson and
Keaton (1950)
Cooper (1957)
County, MS
Mississippi
Baltimore, MD
207 rural Black school children (M4
grade)
361 pregnant women; primarily Black,
low economic and educational level
784 children (>7 months) referred to a
mental hygiene clinic
Baltrop (1966) Boston, MA 439 children (1-6 years): interview
Bruhn and
Pangborn (1971)
California
Robischon (1971) Unspecified
Location
Bronstein and
Dollar (1974)
Hook (1978)
Georgia
New York
Vermeer and Frate Holmes
(1979)
County, MS
Cooksey (1995) Midwest
Smulian et al.
(1995)
Muscogee
County, GA
Stanek et al. (1998) Western
Massachusetts
52 of the children ate dirt in the previous 10 to 16 days; clay was predominant type of soil eaten.
27% of the Black women reported eating clay and 41% reported eating starch. 1% of the White
women reported eating clay and 10% reported eating starch.
Parents/caretakers of 86 children responded positively to "Does your child have a habit, or did he
ever have a habit, of eating dirt, plaster, ashes, etc.?"
19 children ingested dirt (defined as yard dirt, house dust, plant-pot soil, pebbles, ashes, cigarette
ash, glass fragments, lint, and hair combings) in the preceding 14 days.
277 children (1-6 years): mail survey 39 children ingested dirt in the 14 days prior to the survey.
91 Mexican and "Anglo" low-income
families of migrant agricultural workers
130 children (19-24 months) from
urban well-child clinic
410 pregnant, low-income women:
urban N = 201; rural N = 209
250 who had delivered live infants
50 households (229 people; 140
children or adolescents)
350 postpartum women; majority
Black, low income 13 to 42 years old
125 pregnant women ages 12 to 37
years; 73 Black, 47 White, 4 Hispanic,
and 1 Asian.
528 children (ages 1-7 years) at well
medical clinics
12 of 65 Mexican and 11 of 26 "Anglo" respondents indicated consumption of "dirt" among their
family members.
48 "ate nonedibles more than once a week"; substances eaten for 30 of the children were: ashes
(17), "earth" (5), dust (3), fuzz from rugs (2), clay (1), and pebbles/stones (1).
65 (16%o) of the women practiced pica; a variety of substances were ingested, with laundry starch
being the most common; there was no significant difference in the practice of pica between rural
and urban women.
Nonfood items reportedly ingested during pregnancy were ice, reported by three women, and
chalk from a river clay bank, reported by one woman.
Geophagy (regular consumption of clay over a period of weeks) in 16%o of children under 13 years
of age; average daily amount of clay consumed estimated at 50 g for both adults and children.
194 (65%) ingested one or more pica substances during their pregnancy, and the majority (78%)
ate ice/freezer frost alone or in addition to other pica substances.
14.4%) (18 of 125 women) practiced pica; pica prevalence was highest among Black women
(17.8%o). The most common form of pica was ice eating (pagophagia), reported by 44.4%o of the
patients.
Daily mouthing or ingestion: 6%o for sand and stones; 4%o for soil and dirt; 1% for dust, lint, and
dustballs. More than weekly mouthing or ingestion: 16% for sand and stones; 10% for soil and
dirt; 3%o for dust, lint, and dustballs. More than monthly mouthing or ingestion: 27%o for sand and
stones; 18% for soil and dirt; 6%o for dust, lint, and dustballs.
September 2017
PageD-l

-------
Update for Chapter 5 of the Exposure Factors Handbook
Chapter 5—Soil and Dust Ingestion
Table D-l. Studies on the Prevalence of Ingesting Soil, Dust, or Other Nonfood Substances (Continued)
Reference
Location
Population
Results
Grigsby et al.
(1999)
Ward and Kutner
(1999)
Middle Georgia 21 Black individuals (20 of whom were Reasons for eating kaolin included liking the taste, being pregnant, craving it, and to gain weight,
women) who had reported eating kaolin
Metropolitan
Atlanta, GA
Simpson etal.	Santa Ana,
(2000)	Bakersfield,
and East Los
Angeles, CA
Obialo et al. (2001) Atlanta, GA
Klitzman et al. New York City
(2002)
Gavrelis et al.
(2011)
Lumish et al.
(2014)
Lin etal. (2015)
Central
California and
Mexico
226 dialysis patients
150 Mexican-born, low-income,
pregnant or postpartum women
37 (16%) reported pica behavior; Black patients and women made up 86% and 84% of those
reporting pica, respectively. Pica items reportedly consumed included ice, starch, dirt, flour, or
aspirin.
46 (3P/o) of the women interviewed in the United States reported pica behavior; pica substances
included dirt, bean stones, magnesium carbonate, ashes, clay, ice, and other substances; ice was
the most consumed substance.
138 Black dialysis patients; 37-78 years 30 (22%) reported some form of pica behavior; 13 (9.4%o) reported clay pica,
old
United States
nationwide
Rochester, NY
33 pregnant women whose blood lead
levels were >20 |ig/dL; majority were
foreign born; 15—43 years of age
-21,000 individuals (ages 1-74 years)
-25,000 individuals, (ages 0.5-74
years)
158 pregnant adolescents (<18 years of
age); mostly African-American and
~25%o Hispanic
76 Mexican-born pregnant or <2-year
postpartum women
13 women (3 9%o) reported pica behavior during their current pregnancies; 10 reported eating soil,
dirt or clay, 2 reported pulverizing and eating pottery, and 1 reported eating soap.
Prevalence of consuming nonfood substances was 22.1% for the 1- to <3-year age group based on
NHANES I (1971-75) and 12% based onNHANES II (1976-80).
Prevalence estimates for the >21-year age group was 0.1% and 0.4%o for NHANES I (1971-75) and
NHANES 11(1976-80).
46%o reported ingesting one or more items, including raw starches (flour and cornstarch); powder
(dust, vacuum powder from vacuum cleaner bags, and baby powder); soap (soap, bar soap,
laundry soap, and powdered cleansers); plastic/foam (stuffing from pillows/sofas and sponges);
paper (writing paper, toilet paper, and tissues); baking soda/powder; and other (dirt and chalk).
Ice was the nonfood item most often consumed (37%o of all the pregnant adolescents), while only
1.3%o of the teens reported ingestion of dirt/chalk.
In the California, 10 (43%o) had engaged in pica; 6 during pregnancy. In the Mexican group, 18
(34%o) had engaged in pica; 16 during pregnancy. Commonly eaten items were earth, bean stones,
and adobe. In the California group, 5 (22%) had eaten earth; in the Mexican group, 6 (ll%o) had
eaten earth.
Source: Adapted from Moya and Phillips (2014).
September 2017
PageD-2

-------