EPA/63 0/R-03/005
October 2001

Technical Issue Paper

Age Group Recommendations for Assessing Childhood Exposure
and the Adequacy of Existing Exposure Factors Data for Children

Contract No. 68-C-99-238
Task Order 46

Prepared for:

Risk Assessment Forum
U.S. Environmental Protection Agency
Ariel Rios Building (860ID)
1200 Pennsylvania Ave., NW
Washington, DC 20460

Prepared by:

Versar, Inc.
6850 Versar Center
Springfield, VA 22151


-------
EPA/63 0/R-03/005
October 2001

Technical Issue Paper

Age Group Recommendations for Assessing Childhood Exposure
and the Adequacy of Existing Exposure Factors Data for Children

Risk Assessment Forum
U.S. Environmental Protection Agency
Washington, DC 20460

1


-------
Disclaimer

Although the information in this document has been funded wholly by the U.S.
Environmental Protection Agency under Contract 68-C-99-238to Versar, Inc, it does not
necessarily reflect the views of policies of the EPA, and no official endorsement should be
inferred.

11


-------
TABLE OF CONTENTS

EXECUTIVE SUMMARY	 vii

1.0 INTRODUCTION	 1-1

1.1	BACKGROUND	 1-1

1.2	PURPOSE 	1-6

1.3	ORGANIZATION	1-6

1.4	REFERENCES 	1-7

2.0 BREAST MILK: INTAKE, NURSING DURATION, AND FAT COMPOSITION .2-1

2.1	INTRODUCTION	2-1

2.2	EVALUATION OF EXISTING DATA 	2-3

2.2.1	Studies on Breast Milk Intake	2-3

2.2.2	Studies on Lipid Content of Breast Milk and Fat Intake from Breast
Milk	2-6

2.3	STUDIES SELECTED FOR THE ANALYSIS TO OBTAIN NEW
RECOMMENDATIONS	2-7

2.3.1	Studies Selected for Estimating Breast Milk Intake 	2-7

2.3.2	Studies Selected for Estimating Breast Milk Fat Content	2-8

2.4	RECOMMENDATIONS FOR PROPOSED AGE BINS	2-8

2.4.1	Breast Milk 	2-9

2.4.2	Recommendations for Breast Milk Fat Concentration 	2-11

2.5	RECOMMENDATIONS FOR FURTHER ANALYSIS AND RESEARCH
NEEDS	2-12

2.6	REFERENCES 	2-12

3.0 FOOD INTAKE	3-1

3.1	INTRODUCTION	3-1

3.2	EVALUATION OF EXISTING DATA 	3-1

3.3	ANALYSES USED TO OBTAIN NEW FOOD INTAKE
RECOMMENDATIONS FOR CHILDREN	3-4

3.3.1	Individual Intake Rates 	3-4

3.3.2	Total Diet Analysis 	3-4

3 .4 RECOMMENDATIONS FOR PROPOSED AGE BINS	3-5

3.4.1	Results of Reanalysis	3-5

3.4.2	Uncertainties 	3-6

3.5	RECOMMENDATION FOR FURTHER ANALYSIS AND RESEARCH
NEEDS	3-7

3.6	REFERENCES 	3-7

l


-------
TABLE OF CONTENTS (continued)

4.0 DRINKING WATER AND TOTAL FLUIDS 	4-1

4.1	INTRODUCTION	4-1

4.2	EVALUATION OF EXISTING DATA 	4-1

4.3	STUDIES SELECTED FOR ANALYSIS TO OBTAIN NEW
RECOMMENDATIONS	4-3

4.4	RECOMMENDATIONS FOR PROPOSED AGE BINS	4-8

4.5	RECOMMENDATIONS FOR FURTHER ANALYSIS AND RESEARCH
NEEDS	4-8

4.6	REFERENCES 	4-11

5.0 SOIL INGESTION AND PICA	5-1

5.1	INTRODUCTION	5-1

5.2	EVALUATION OF EXISTING DATA	5-2

5.2.1	Soil Ingestion	5-2

5.2.2	Soil Pica	5-5

5.3	STUDIES SELECTED FOR ANALYSIS TO OBTAIN NEW
RECOMMENDATIONS	5-5

5.3.1	Soil Ingestion	5-5

5.3.2	Pica	5-8

5.4	RECOMMENDATIONS FOR PROPOSED AGE BINS	5-9

5.4.1	Soil Ingestion	5-9

5.4.2	Pica	5-11

5.5	RECOMMENDATIONS FOR FURTHER ANALYSIS AND RESEARCH
NEEDS	5-12

5.6	REFERENCES 	5-13

6.0 NON-DIETARY INGESTION EXPOSURE 	6-1

6.1	INTRODUCTION	6-1

6.2	EVALUATION OF EXISTING DATA 	6-1

6.3	NEW STUDIES	6-4

6.3.1	Modeling Efforts 	6-5

6.3.2	Monitoring Studies	6-6

6.3.3	Other Papers of Interest	6-7

6.4	STUDIES SELECTED FOR ANALYSIS TO OBTAIN NEW
RECOMMENDATIONS	6-7

6.5	RECOMMENDATIONS FOR PROPOSED AGE BINS	6-7

6.6	RECOMMENDATION FOR FURTHER ANALYSIS AND RESEARCH
NEEDS	6-8

6.7	REFERENCES 	6-8

li


-------
TABLE OF CONTENTS (continued)

7.0 EXPOSURE FACTORS FOR INHALATION	7-1

7.1	INTRODUCTION	7-1

7.2	EVALUATION OF EXISTING DATA 	7-1

7.2.1	Activity-Based Estimation of Inhalation Rates 	7-2

7.2.2	Metabolically Based Inhalation Rates	7-3

7.2.3	Inhalation Rates Determined from Heart Rate Measurements and
Activity Data 	7-4

7.3	STUDIES SELECTED FOR THE ANALYSIS TO OBTAIN NEW
RECOMMENDATIONS	7-4

7.3.1	Food Energy Intakes for Children	7-4

7.3.2	Ventilatory Equivalents for Children	7-6

7.4	RECOMMENDATIONS FOR PROPOSED AGE BINS	7-8

7.5	RECOMMENDATIONS FOR FURTHER ANALYSIS AND RESEARCH
NEEDS	7-9

7.6	REFERENCES 	7-12

8.0 EXPOSURE FACTORS FOR THE DERMAL ROUTE 	8-1

8.1	INTRODUCTION	8-1

8.2	EVALUATION OF EXISTING DATA 	8-2

8.2.1	Surface Area Studies 	8-2

8.2.2	Soil Adherence Studies 	8-5

8.3	STUDIES SELECTED FOR THE ANALYSIS TO OBTAIN NEW
RECOMMENDATIONS	8-5

8.4	RECOMMENDATIONS FOR PROPOSED AGE BINS	8-6

8.5	RECOMMENDATIONS FOR FURTHER ANALYSIS AND RESEARCH
NEEDS	8-7

8.6	REFERENCES 	8-7

9.0 CHILD-SPECIFIC ACTIVITY PATTERNS 	9-1

9.1	INTRODUCTION	9-1

9.2	EVALUATION OF EXISTING DATA 	9-2

9.3	STUDIES SELECTED FOR THE ANALYSIS TO OBTAIN NEW
RECOMMENDATIONS	9-6

9.4	RECOMMENDATIONS FOR EACH AGE BIN	9-6

9.5	RECOMMENDATIONS FOR FURTHER ANALYSIS AND RESEARCH
NEEDS	9-10

9.6	REFERENCES 	9-10

iii


-------
TABLE OF CONTENTS (continued)

10.0 BODY WEIGHT	10-1

10.1	INTRODUCTION	10-1

10.2	EVALUATION OF EXISTING DATA 	10-1

10.3	STUDIES SELECTED FOR ANALYSIS TO OBTAIN NEW
RECOMMENDATIONS	10-3

10.4	RECOMMENDATIONS FOR PROPOSED AGE BINS	10-4

10.5	RECOMMENDATIONS FOR FURTHER ANALYSIS AND RESEARCH
NEEDS	10-4

iv


-------
LIST OF TABLES

Table 1. Availability of Recommendations for the Proposed Age Bins	ix

Table 2. Summary of Exposure Factor Recommendations	 x

Table 2-1. Mean Fat Content of Breast Milk Intake Samples for Infants 1-6 Months	2-7

Table 2-2. Recommended Breast Milk Intake and Proposed Age Bins	2-9

Table 2-3. Confidence in Recommendations for Breast Milk Intake and Breast Milk Fat

Content 	2-10

Table 3-1. Weighted and Unweighted Number of Observations used in the 1994-96 CSFII

Analysis, for the Child-Specific Exposure Factors Handbook Age Groups	3-2

Table 3-2. Weighted and Unweighted Number of Observations Used in the Reanalysis of

the 1994-96 CSFII for the Selected Age Bins	3-3

Table 3-3. Per Capita Intake of the Major Food Groups (g/kg/day, as consumed) 	3-8

Table 3-3a. Per Capita Intake of the Major Food Groups (g/day, as consumed)	3-10

Table 3-4. Per Capita Intake of Individual Foods (g/kg/day, as consumed) 	3-12

Table 3-5. Per Capita Intake of USD A Categories of Vegetables and Fruits (g/kg/day, as

consumed)	3-14

Table 3-6. Per Capita Intake of Exposed/Protected Fruit and Vegetable Categories

(g/kg/day, as consumed)	3-16

Table 3-7. Per Capita Intake of Major Food Groups (g/kg/day, as consumed) 	3-18

Table 3-7a. Per Capita Intake of Major Food Groups (g/day, as consumed)	3-20

Table 3-8. Consumer Intake of Major Food Groups (g/kg/day, as consumed) 	3-22

Table 3-8a. Consumer Intake of Major Food Groups (g/day, as consumed)	3-24

Table 3-9. Intake of Total Foods and Major Food Groups, and Percent of Total Food
Intake for Individuals with Low-end, Mid-range, and High-end Total Food

Intake	3-28

Table 3-10. Intake of Total Foods and Major Food Groups, and Percent of Total Food

Intake for Individuals with Low-end, Mid-range, and High-end Total Intake . . 3-30
Table 3-11. Intake of Total Foods and Major Food Groups, and Percent of Total Food
Intake for Individuals with Low-end, Mid-range, and High-end Total Meat

and Dairy Intake 	3-32

Table 3-12. Intake of Total Foods and Major Food Groups, and Percent of Total Food
Intake for Individuals with Low-end, Mid-range, and High-end Total Fish

Intake	3-34

Table 3-13. Per Capita Intake of Total Foods and Major Food Groups, and Percent of
Total Food Intake for Individuals with Low-end, Mid-range, and High-end

Total Fruit & Vegetable Intake	3-36

Table 3-14. Per Capita Intake of Total Foods and Major Food Groups, and Percent of
Total Food Intake for Individuals with Low-end, Mid-range, and High-end

Total Dairy Intake	3-38

Table 3-15. Confidence in Recommendations for Food Intake	3-40

Table 3-16. Number of Children Providing Intake Data in CSFII 1994-96 and CSFII 1998 . 3-41

v


-------
LIST OF TABLES (continued)

Table 4-1. Estimated Direct and Indirect Total Water Ingestion by Source for U.S.

Population	4-3

Table 4-2. Estimate of Total Direct and Indirect Water Ingestion, All Sources, by Broad

Age Category for U.S. Children	4-4

Table 4-3. Estimate of Total Direct and Indirect Water Ingestion, All Sources, by Fine

Age Category for U.S. Children	4-5

Table 4-4. Urinary Volume Rates 	4-6

Table 4-5. Recommended Values for Direct, Indirect, and Both Direct and Indirect Water

Ingestion Excluding Commercial and Bottled Water	4-9

Table 4-6. Confidences in Recommendations for Drinking Water Ingestion 	4-10

Table 5-1. Soil Equivalent Amount in Fecal Samples for Pica Subject, by Week (mg/day) . 5-8

Table 5-2. Recommended Values for Soil Ingestion	5-10

Table 5-3. Confidence in Recommendations for Soil Ingestion and Pica	5-16

Table 6-1. Confidence Evaluation of the Existing Non-Dietary Exposure Studies	6-11

Table 7-1. Summary of the Inhalation Rate Estimates for Selected Age Groups 	7-2

Table 7-2. Comparison of Food Energy Intakes for Children Under 5 Years of Age

Sampled in the 1994-96 and 1998 CSFII and the 1977-78 NFCS 	7-5

Table 7-3. Summary of Energy Expenditures for Children and Adolescents 	7-6

Table 7-4. Summary of Ventilatory Equivalents for Children and Adolescents 	7-8

Table 7-5. Daily Inhalation Rates Estimated for Children and Adolescents 	7-9

Table 7-6. Confidence in Recommendations for Inhalation Rates	7-11

Table 8-1. Evaluation of Existing Surface Area Studies 	8-9

Table 9-1. Number of Person-Days/Individualsa for Children in CHAD Database	9-4

Table 9-2. Number of Hours Per Day Children Spend in Various Microenvironments by

Age 	9-5

Table 9-3. Number of Hours Per Day Children Spend Doing Various Macroactivities

While Indoors at Home by Age 	9-6

Table 9-4. Estimated Number of Hours Per Day Children Spend in Various

Microenvironments by Age Bin	9-7

Table 9-5. Estimated Number of Hours Per Day Children Spend Doing Various

Macroactivities While Indoors at Home by Age Bin 	9-8

Table 9-6. Confidence for Recommendations for Activity Factors	9-9

Table 10-1. Recommended Body Weight Values for Proposed Age Bins (kg)	10-5

Table 10-2. Confidence in Recommendations for Body Weight	10-6

LIST OF FIGURES

Figure 4-1. Changes in human body composition during fetal development and early life.

These data often serve as a reference standard for assessing growth in the

preterm infant	4-7

Figure 8-1. Prevalence of Overweight Among Children and

Adolescents Ages 6-19 Years	8-4

vi


-------
AUTHORS/CONTRIBUTORS

This issue paper was prepared at the direction of the U.S. Environmental Protection
Agency, Risk Assessment Forum. It was prepared by Versar, Inc. under EPA Contract No. 68-
C99-238, Task Order 46. Mr. Steven Knott of the Risk Assessment Forum served as Task Order
Manager providing overall direction, technical assistance, and guidance.

AuthorsContributors

Dr. Paul Brubaker, P.E., DBATVersar, Inc.

Brubaker Associates, Inc.Patricia Wood, Work Assignment Manager
Kathy Kelly, Technical Editor
Dr. Christine ChaissonSusan Perry, Word Processing
Chaisson Scientific Advisors, Inc.

Mr. Robert Fares
Risk Management Initiative, LLC

Dr. David Layton
Lawrence Livermore National Laboratory

Dr. Linda Phillips
Versar, Inc.

Dr. P. Barry Ryan
School of Public Health
Emory University

Dr. Edward Stanek
School of Public Health
University of Massachusetts

vii


-------
EXECUTIVE SUMMARY

In keeping with initiatives such as the Food Quality Protection Act of 1996 and Executive
Order 13045, EPA has been investigating ways to improve the assessment of childhood risks from
exposure to environmental contaminants. In July 2000, a workshop was convened by EPA to
explore how to consider age-related changes in behavior and anatomy when assessing risks to
children. The discussions and findings of the workshop are presented in the document Summary
Report of the Technical Workshop on Issues Associated with Considering Developmental Changes
in Behavior and Anatomy When Assessing Exposure to Children (EPA/630/R-00/005). The current
technical issue paper was prepared to provide additional expert input for issues raised during the July
2000 workshop. In particular, this issue paper explores whether existing exposure factors data can
be used to address childhood age groups derived from the July 2000 discussions.

Typically, Agency assessors have classified individuals under the age of 21 years as youth
or children. However, how to subdivide this group to capture important developmental milestones
has been somewhat elusive. To that end, participants in the July 2000 workshop concluded that,
although development occurs along a continuum, age groups (or bins) can be useful as a guide for
the development of exposure scenarios for children. Because children's behavior, such as crawling
and mouthing of hands and objects, changes during the early life stages, dermal and oral exposures
vary. Similarly, physiological changes affect children's exposures and their susceptibility to certain
health effects. Two workshop subgroups, one addressing behavioral development and the other
addressing anatomical changes, presented recommendations for age groups. The age groups
considered in this issue paper were derived from those recommendations and include:

Less than 1 month
1 through 2 months
3 through 5 months
6 through 11 months
1 through 2 years

3 through 5 years
6 through 10 years
11 through 15 years
16 through 17 years

Prenatal development was outside the scope of the workshop discussions and of this issue paper.

This issue paper explores whether existing child-specific exposure factors data can be fit to
the proposed EPA age groupings. The authors evaluated the exposure factors — breast milk intake,
food intake, drinking water and total fluid intake, soil ingestion and pica, non-dietary ingestion
factors, inhalation factors, skin surface area and soil adherence to skin (dermal), activity factors, and
body weight — against the exposure factors recommended in the draft Child-Specific Exposure

viii


-------
Factors Handbook (CSEFH). Where reevaluation of the underlying data was possible and
productive, the authors recommended new exposure factors for the proposed age groups. The
authors also discussed the quality of the data provided by the key studies used in developing the
CSEFH. Where reevaluation of exposure factors data was not possible or would not support new
recommendations for the proposed age groups, the authors addressed the uncertainties that would
be introduced if the existing data were used. The authors evaluated the data (new and existing) using
the same characterization criteria used in the CSEFH.

In each section the author evaluated existing data, described the studies he or she selected
for analysis to obtain new recommendations, recommended exposure factors data for the proposed
age bins based on new analysis or reanalysis, and recommended additional research and analysis
needs. The availability of recommendations and the recommended values for the proposed age bins
are presented in Tables 1 and 2, respectively.

Chapter 2: Breast Milk Intake, Nursing Duration, and Fat Composition of Human Milk

Breast milk is a potential source of exposure to toxic substances for nursing infants. Lipid-
soluble chemical compounds may accumulate in body fat and may be transferred to breast-fed infants
in the lipid portion of breast milk. Because nursing infants obtain most (if not all) of their dietary
intake from breast milk, they are especially vulnerable to exposures to these compounds. Estimating
the magnitude of the potential dose to infants and the duration of that exposure requires information
on the milk intake rate (quantity of breast milk consumed per day) and the duration over which
breast-feeding occurs. Information about the fat composition of breast milk is also needed to
estimate the possible concentration of the chemical moieties in the milk. The delivered dose of toxic
chemical can then be estimated from breast milk residue concentrations that have been indexed to
lipid content.

The author evaluated several studies that were used in developing the CSEFH to determine
those with data applicable to the EPA proposed age bins. Study designs varied, but they typically
comprised fairly small cohorts. The most typical measurement tool used in the studies was the test
weighing methodology, in which infant weights were recorded prior to and after feeding.

IX


-------
Table 1. Availability of Recommendations for the Proposed Age Bins

Age Bins

Considerations

<1
Month

1-2
Months

3-5
Months

6-11
Months

1-2

Years

3-5

Years

6-10
Years

11-15
Years

16-17
Years

Breast Milk Intake

/a

/

/

/

b









Food Intake0

—

/

/

/

/

/

/

/

/

Drinking Water and
Total Fluids

—

—

—

/

/

/

/

/

/

Soil Ingestion

—

—

—

—

/

/

/

—

—

Pica

—

—

—

—

—

—

—

—

—

Non-Dietary Factors

—

—

—

—

—

—

—

—

—

Inhalation Route

—

—

—

—

—

/

/

/

/

Dermal Route



















Surface Area

—

—

—

—

—

—

—

—

—

Soil Adherence4

/

/

/

/

/

/

/

/

/

Activity Factors

—

—

—

—

/

/

/

—

—

Body Weight

/

/

/

/

/

/

/

/

/

a y = Recommendation is available.
b — = Recommendation not available.

c Note that the sample sizes are small for some of the age groups, most notably <1 month. Recommendations cannot
be made for age groups <1 month. For most food categories, a recommendation cannot be made for age groups 1-2
months and 3-5 months.
d The same factor value is recommended for all age bins.

X


-------
Table 2. Summary of Exposure Factor Recommendations

Breast Milk

Intake (g/day)

<1 month
1 -2 months
3-5 months
6-11 months
12 months

Breast Milk Fat Intake

All age bins

650
680
780
740
410

4% lipid content

Food Intake



Per Capita Intake

Consumer Only

Total diet & major

food groups

Tables 3-3 and 3-3a



Individual foods



Table 3-4



Various USDA food categories

Table 3-5



Exposed/protected fruits/vegetables

Table 3-6



Major food groups



Tables 3-7 and 3-7a

Tables 3-8 and 3-8a

Contribution of Major

Food Group to

Tables 3-9 through 3-14



Total Dietary Intake







Drinking Water



Mean

95th Percentile





Direct Ingestion (mL/person/day)

6-11 months



96

-

1-2 years



184

677

3-5 years



274

880

6-10 years



317

1,030

11-15 years



414

1,531

16-17 years



531

2,618





Indirect Ingestion (mL/person/day)

6-11 months



316

-

1 -2 years



129

432

3-5 years



145

458

6-10 years



136

482

11-15 years



180

629

16-17 years



531

2,618





Direct and Indirect Ingestion (mL/person/day)

6-11 months



412

—

1 -2 years



313

942

3-5 years



420

1,165

6-10 years



453

1,219

11-15 years



594

1,722

16-17 years



760

2,062

XI


-------
Table 2. Summary of Exposure Factor Recommendations (continued)

Soil Ingestion





Ingestion Rate (mg/day)







Mean

Median

90th Percentile

1-2 years



30

24

100

3-5 years



30

20

150

6-10 years



71

37

187

Inhalation



Mean Inhalation Rate (m3/day)





Male

Female

Male and Female

3-5 years



-

-

14

6-10 years



14

12

-

11-15 years



14

13

-

16-17 years



16

12

-

Dermal Soil Adherence









All age bins



Use same values recommended in CSEFH (Table 8-8)

Activity Factors





Time Spent (hrs/day)





Indoors

Outdoors Indoors at

Outdoors In

Microenvironment



at Home

at home School

at Park Vehicle

1-2 years



18.6

1.9 5.5

2.0 1.2

3-5 years



17.2

2.3 5.4

1.8 1.3

6-10 years



15.7

2.3 6.0

2.0 1.2







Watch

Read,







TV or

Write, Think,

Macroactivity in Home



Sleep or Shower

Play Listen to

Home- Relax,

Microenvironment

Eat

Nap or Bathe

Games Radio

work Passive

1-2 years

1.4

11.8 0.5

3.2 2.0

0.6 1.8

3-5 years

1.1

10.9 0.5

2.5 2.5

0.8 1.1

6-10 years

1.0

9.9 0.4

1.9 2.8

1.1 0.8





Male: Mean at 50th

Female: Mean at 50th

Average Male/

Body Weight



Percentile

Percentile

Female Value

0-1 month



4.00

3.80

3.90

1-2 months



4.88

4.54

4.71

3-5 months



6.72

6.15

6.43

6-11 months



9.04

8.28

8.66

1-2 years



11.53

10.80

11.16

3-5 years



16.27

15.83

16.05

6-10 years



25.86

25.95

25.90

11-15 years



45.77

45.41

45.59

16-17 years



62.84

54.54

58.69

Xll


-------
The author selected four studies (cross-sectional or longitudinal) for analysis and made
recommendations for the proposed age bins birth to 1 month, 1 to 2 months, 3 to 5 months, 6 to 11
months, and 12 months on the basis of one or more of the studies. In general, the author expanded
the results of studies of homogeneous populations with small cohorts to a large, diverse U.S.
population, so exact time-weighted averages or statistical boundaries could not be calculated for
those values. The author derived estimates of intake for various age groups by calculating averages
across the findings in different studies or by deriving time-weighted averages across time periods.
The author recommended data for breast milk fat concentrations using two studies with narrowly
defined populations and different analysis methodologies, which made it difficult to extrapolate the
values across the population for all age groups and durations.

Recommendations for further analysis and research needs included the need to (1) consider
the major ethnic groups in the U.S. population to estimate the variability across the population for
milk intake as a function of age and/or infant weight (new studies are needed on black, Asian, and
Hispanic mother/infant groups); and (2) estimate the effect of nutrient status of the mother before
and during lactation on the fat content of the milk. Data are needed on the types of lipids that may
change because of these variables and the mobility of such lipids in the milk during lactation.

Chapter 3: Food Intake

Children's exposure from food ingestion may differ from that of adults because of differences
in the types and amounts of food eaten. Exposure also may differ because the intake per unit body
weight is greater for children than for adults. The author reevaluated the data used in the CSEFH to
provide food intake rates for the proposed age groups of interest to the Risk Assessment Forum. The
reevaluation included intake rates for various food groups and individual foods, as well as for the
total dietary analysis. The chapter did not reevaluate some CSEFH data, including intake rates for
home-produced foods, serving sizes, or fish intake.

The primary source used by the CS EFH for recent information on consumption rates of foods
among children was the U.S. Department of Agriculture's (USDA) 1994-96 Continuing Survey of
Food Intakes by Individuals (CSFII). The 1994-96 CSFII used a stratified statistical sampling
technique designed to ensure that all seasons, geographic regions, and demographic and
socioeconomic groups were represented. The survey included individuals of all ages living in
selected households in the 50 States and Washington, D.C. Individual data was for 2 non-
consecutive days, based on 24-hour recall (76 percent response rate). Approximately 15,000
individuals provided intake data over the 3 survey years. Because of the relatively large sample size
in the 1994-96 CSFII data set, the author was able to reanalyze the CSFII data to generate intake


-------
rates for the proposed age groups. The resulting number of observations in some of the new age
groups were lower than in the original CSEFH analysis.

The author addressed the uncertainties resulting from new exposure data values that have
been generated from short-term data. Although the data are suitable for estimating mean average
daily intake rates representative of both short-term and long-term consumption, the distribution of
average daily intake rates generated using short-term data (e.g., 2-day) does not necessarily reflect
the long-term distribution of average daily intake rates. Other variations occur that affect the
applicability of the data, and the author noted that the small sample sizes used in the reanalysis affect
the confidence ratings for data for children less than 1 year of age.

The author recommended that, although 1994-96 CSFII data were used for this reanalysis,
the 1998 CFSII data that are now available would be a useful supplement in evaluating food intake
for children. A similar analysis to that conducted here for this issue paper would increase the
number of observations on which the child intake rates are based and improve the confidence level
for intake rates generated by such an analysis.

Chapter 4: Drinking Water and Total Fluids

The CSEFH summarized two studies, theUSDA 1994-96 Continuing Survey of Food Intakes
by Individuals and Estimated Per Capita Water Ingestion in the United States, which reported intake
rates for both direct and indirect ingestion of water. Direct intake is defined as direct consumption
of water as a beverage, while indirect intake includes water added during food preparation but not
water intrinsic to purchased foods. Both studies supported EPA's use of 1 L/day as the default
drinking water intake rate for infants (10 kg body mass or less) and children, including drinking
water consumed in the form of juices and other beverages containing tap water (direct intake).
Because the use of total water intake to estimate exposure may overestimate the potential exposure
to toxic substances present only in local water supplies, the two studies emphasized intake of tap
water (community water), rather than total water intake. The two studies were selected for the new
analysis for the proposed age bins.

T o estimate per capita ingestion of plain drinking water (direct ingestion) and water ingested
indirectly, the USD A study gathered responses to questionnaires in which respondents averaged
quantities of water they ingested over 2 nonconsecutive days to generate a 2-day average. These
daily averages comprised the empirical distributions from which mean and upper percentile per
capita ingestion rates were developed from 2-day averages for more than 15,000 individuals in 50

xiv


-------
States and the District of Columbia, which were then extrapolated to the population of the entire
United States.

In the second study, EPA used the 1994-96 CSFII data to estimate the per capita drinking
water ingestion rate for subpopulations segregated by (1) gender and age, and (2) by pregnant,
lactating, and childbearing-age women. The study report, Estimated Per Capita Water Ingestion in
the United States, estimated the 2-day average per capita ingestion of community water at 2.016
L/person/day, with estimates for infants less than 1 year of age and for children 1 to 10 years of age
consistent with the standard 1-liter ingestion rate used in risk assessments for a 10 kg child.

The author conducted a new analysis using the CSFII data to obtain recommended values for
the proposed age bins; however, the CSFII data set did not provide enough data for the proposed age
groups of children up to 1 year of age. Also, the author noted that assessing exposure as volume of
intake per unit body mass clearly indicates the greater potential for exposure of the young child.
Fetuses, newborns, and toddlers up to 2 years of age are vulnerable as a result of their different renal
function and fluid intake needs.

The author recommended that more research be conducted to collect data on intake rates for
drinking water within the proposed age bins, noting that specific age data from the CSFII
questionnaires might be used if the sample sizes for those age groups are adequate to extrapolate the
data to the general children's population. Also, a new questionnaire should be developed that
considers independent data validation, such as use of physiological parameters to gauge water
balance and fluid intake values. The author also discussed assessment methods for children,
including the greater potential for exposure of young children because of the ratio of body surface
area to intake rates as well as their different renal function and fluid intake needs. Drinking water
consumption and total fluid intake should be normalized by body surface area as well as by body
weight. Fetuses, infants, and adults have distinct shape-weight relationships, and estimates of BSA
are used frequently in the practice of anesthesiology and critical care medicine to reflect the body's
metabolic functions, such as ventilation rate, fluid requirements, and extracorporeal circulation. A
simple linear relationship between BSA and weight in infants and children weighing between 3 and
30 kg can be developed using linear regression analysis on published human BSA data.

Chapter 5: Soil Ingestion and Pica

To quantify the amount of toxic substance(s) ingested by a child over time, exposure
parameters relevant for risk assessment need to be selected and computed. Although absorption of

xv


-------
toxicants and toxic effects have obvious importance, this chapter limits its discussion to quantifying
ingestion of toxic substances.

Two facts make quantifying ingestion of toxic substances in soil and/or dust difficult. First,
it has not been possible to directly quantify the amount of a toxic substance ingested. Instead,
several indirect strategies for constructing such estimates have been proposed, including a behavioral
strategy, a mass-balance strategy, an estimation based on comparison of average blood lead of
populations between areas with different soil lead concentrations, and a combined approach using
behavioral data to develop a model for soil ingestion with age, and then applying this model to mass-
balance soil ingestion estimates. Only the mass-balance method has been experimentally validated
(among adults); therefore, the authors limited their discussion to mass-balance studies of soil
ingestion using trace elements contained in soil. The second factor that made estimate of soil
ingestion difficult was the design of the existing studies.

Although there is much literature on ingestion of nonfood items (which is also referred to as
pica), the authors found limited information on soil pica, most of which was anecdotal or for special
populations. The studies that do exist fall short of defining the amount of soil ingested in a pica
event, but they provide insights that may help guide the design of studies that can quantify soil pica.

The authors focused on four primary studies (of seven reviewed) of soil ingestion and
numerous manuscripts based on data from those studies. The common features of the studies were
the use of mass-balance methodology, their conduct in the United States, and the use of the trace
elements aluminum, silicon, and titanium. The authors also reviewed reports that directly discussed
the relationship between soil ingestion and age. Because the studies were conducted on children 1-7
years of age, the authors made exposure factor recommendations for soil ingestion for only the age
groups 1-2 years, 3-5 years, and 6-10 years.

According to the authors, interpretations of estimates of soil ingestion from the four studies
have several limitations that affect their generalizability. However, they used three of the four mass-
balance studies to develop recommendations for the proposed age bins. They reported the estimates
first as ranges to indicate the extent of variability of individual studies, then combined the studies'
trace-element-specific estimates to form single estimates. To more closely correspond to estimates
based on a longer study design, they used long-term data from one of the studies, which provided
recommended values with higher confidence levels for some age bins. The authors compared their
recommendations for the mean and 90th percentile with those in the CSEFH and discussed the
differences.

xvi


-------
To highlight the difficulty of framing a recommendation concerning soil pica that has value
for exposure assessment, the authors examined in detail a subject in one of the four studies who
exhibited very high soil ingestion, presumed to be soil pica.

The authors concluded that there are many areas where little is known about soil ingestion.
Even where they could provide recommendations, they said there is no evidence to support
generalizing the estimates to other seasons of the year or to other parts of the country. It is unlikely
that there are sufficient data available to reliably distinguish soil ingestion in different age ranges,
although existing data could be used to improve on the estimates of soil ingestion for age groupings.
They recommended using the multiple estimates of soil ingestion (using the multiple trace elements)
to obtain a single estimate of soil ingestion and characterize the distribution. The main limitations
to making recommendations for all the proposed age bins included the fact that soil ingestion studies
are difficult to conduct and data collection and chemical processing are expensive.

According to the authors, the most critical need at this point is new data, which should be
gathered in the context of a research program. Data are needed that span a broader age range,
perhaps initially expanding the age range to extend from 3 months to 12 years. Data on children in
future soil ingestion studies need to span the range of demographic variables such as geography,
race, and economic status so that results can be more confidently applied to the general U.S.
population. Estimates of soil ingestion need to reflect longer time periods. Seasonal effects and
longitudinal studies (both over seasons and over years) are important to identify tracking that may
lead to a broader or narrower soil ingestion distribution. Finally, soil ingestion studies need to be
integrated with behavioral studies and made efficient. Much has been learned as a result of the
conduct of soil ingestion studies in the past, and this needs to be taken advantage of in the future.

Chapter 6: Non-dietary Ingestion Exposure

For young children, mouthing activities offer one of the most common ways for the child to
explore his or her environment. However, contamination of any object used in mouthing activities
may lead to elevated exposure to a variety of chemical compounds, including metals, pesticides, and
other potentially toxic compounds.

The authors evaluated the data from studies used in the CSEFH, all of which they consider
to be lacking in some fundamental way, and then examined the literature since 1999 to assess the
applicability of any studies not considered for the CSEFH, including several modeling efforts and
monitoring studies. The authors determined that no studies provide definitive data for the age bins
proposed by EPA. There have been no systematic, probability-based studies undertaken that would

xvii


-------
afford a reasonable assessment of such, nor have any studies been designed and implemented that
would determine whether age-bin-specific factors differ from one another. Sample sizes in the
studies outlined above were too small, and study designs were inconsistent and often spanned only
part of the range (e.g., very little work has been done on children ages 10-20 years).

The authors stressed that the most important research need is a systematic method of data
collection. Current meta-analysis is limited by inconsistencies in data and small sample sizes. One
clear recommendation arising from the authors' assessment of existing data was that a new,
comprehensive data collection effort to determine non-dietary ingestion exposure factors should be
designed and undertaken.

Chapter 7: Exposure Factors for Inhalation

Three basic techniques are used for estimating inhalation rates: activity-based approach,
using assigned breathing rates for various activities and calculating daily values; metabolic approach,
which determines breathing rate as a function of oxygen demand; and a hybrid approach, in which
a physiological measure of oxygen consumption, such as heart rate, is used along with activity data
and a personal calibration curve of heart rate to inhalation to estimate how much air is inhaled during
an individual's daily activities.

The studies reviewed by the authors used these approaches, or modifications of them, to
develop inhalation estimates for different age groups. The authors focused on the data and methods
that could best be used to develop recommendations for inhalation exposure factors for the proposed
age groups, based on chronic or daily inhalation rates applicable to the proposed childhood age
groups.

The data from the selected studies could not be reanalyzed to support recommendations for
the new age bins, had a very limited data set on which to determine applicable multipliers for
different age groups, or involved small test cohorts. However, the studies were used as the
foundation for developing new recommendations for age-specific inhalation rates, followed by a
review of more recent information to address data gaps pertaining to the calculation of breathing
rates for children of different ages.

The authors recommended improving particular studies and methodologies that were
discussed and identified several analyses that need to be performed. First, daily energy expenditures
computed using CHAD should be compared with food-energy intakes obtained from USDA food
consumption surveys for children in the different age bins. They also recommended developing

xviii


-------
expanded data sets ofVQ measurements on children using the applied physiology literature together
with new, direct measurements. Reanalysis of the raw data from selected studies could provide age-
dependent distributions of VQ. Results of such analyses could be used to design experimental
studies targeting children in selected age bins. Another related issue concerns the nature of the age-
dependent decline in VQ as well as related gender differences (e.g., at what age should males and
females be treated separately with respect to VQ?). In addition, Linn et al. (1991) noted that
asthmatic children in their study had higher inhalation rates. Given those children's enhanced
susceptibility to ozone, this also should be an area of special research.

In summary, until additional data on VQ values are obtained for children ages 5 years and
younger, estimates of inhalation rates for this particular age cohort will have potentially significant
uncertainties. The existing recommendations given in the Child-Specific Exposure Factors
Handbook could underestimate actual inhalation rates by 50 percent or more, depending on the
results of additional measurements of VQ values as well as the development of improved estimates
of age-specific energy intakes and expenditures.

Chapter 8: Exposure Factors for the Dermal Route

Dermal exposure is estimated, in part, by the amount of body surface area available for
contact with contaminated media. The amount of body surface area exposed during an event is
influenced by age-specific behavioral factors. For children, such factors include playing and
crawling on contaminated surfaces, and the amount of clothing worn during play activities. Surface
area of the skin is determined via direct measurement or regression models that consider the
dependence of surface area on such other body dimensions as height and weight. The CSEFH
described various measurement techniques and reviewed pertinent surface area studies as a basis
for recommending body surface areas for children that are representative of the subpopulation under
consideration (i.e., age and sex-dependent).

The author conducted a thorough search of peer-reviewed literature back to 1997; however,
no new studies were identified that have been performed relative to those factors. Also, total surface
area could not be estimated from the NHANESII or III data sets, because height information, which
is equally importantto calculate surface area, was not included. Therefore, no recommendations for
surface area values could be made with respect to the proposed age bins. The lack of NHANES II
height data precluded estimating the recommended total surface area values for children less than
1 month of age, 1-2 months, 3-5 months, 6-11 months, or 1-2 years of age. Estimates of surface area
values for the older age groups would require an extensive statistical reevaluation of the NHANES
data that is beyond the scope of the current paper.

xix


-------
The most important research need is to compile appropriate height and weight data for
children under 2 years of age. Because of their behavior patterns (e.g., playing and crawling on
contaminated surfaces with fewer clothes for protection) and physical factors (i.e., higher surface
area relative to body weight), children in this age group may have potentially higher exposure to
environmental toxicants than other age groups.

With regard to dermal soil adherence, the CSEFH already recommended that more-detailed
studies are necessary because the control experiments and field studies conducted to date were based
on specific situations and a limited number of measurements. The results of those studies showed
that soil adherence generally could be directly correlated with moisture content, inversely correlated
with particle size, and independent of clay content or organic carbon. Additionally, the rate of soil
adherence is higher for hands than for other parts of the body. Therefore, because soil adherence is
more activity-specific than age-specific, the values recommended in the CSEFH can be used for any
age group, depending on the activity considered.

Chapter 9: Activity Patterns

Chemical exposures of children are influenced by the types of activities in which they are
engaged as well as the locations of the activities and the level of participation in those activities.
Consequently, exposure to chemicals in the environment can vary among children of similar
developmental stages because of the variability associated with their behavior. Additionally,
seasonal and geographic differences among children of similar developmental stages influence the
variability of exposure.

The CSEFH described various measurement techniques and reviewed pertinent activity
studies as a basis for recommending activity factors for children that are representative of the
subpopulations (i.e., age and sex-dependent) under consideration. No new pertinent studies have
been performed relative to those factors since the development of that document.

The author reviewed five studies (four that were used in the CSEFH) and determined that,
presently, no recommendations can be made with respect to activity factor values for children less
than 1 month old, 1 to 2 months, 3 to 5 months, 6 to 11 months, 11 to 15 years, or 16 to 17 years.
As stated in the CSEFH, the current database on children's macroactivities is sparse and data are
insufficient to adequately assess exposures to environmental contaminants. However, data are
sufficient to estimate values for time spent in various microenvironments and participation in certain
macroactivities for children in age bins of 1-2 years, 3-5 years, and 6-10 years (see Tables 9-1
through 9-3). The author presents the recommended values for time spent in microenvironments and

xx


-------
in macroactivities for children in age bins of 1-2 years, 3-5 years, and 6-10 years, as well as the level
of confidence for the recommended activity factor values.

Overall, the present state of knowledge regarding children's exposures and activities are
inadequate to assess exposures to environmental contaminants sufficiently. Research needs to be
conducted in three specific areas in order to improve the database that is currently available to assess
children's exposures. Methods for monitoring children's activities and exposures need to be
improved. Additionally, physical activity data for children, but especially young children less than
4 years of age and in age bins 11-15 years and 16-17 years, need to be collected in order to assess
exposure by all routes. In order to accomplish this, population-based data are required to improve
the characterization of children's activities and exposures as a function of age, gender, environmental
setting (residence, school, daycare), socioeconomic status, race/ethnicity, location (urban, suburban,
rural), region, and season.

Chapter 10: Body Weight

Exposure and risk assessments are frequently expressed as a function of dose normalized to
the average body weight of the exposed population. Body weight is one of the parameters in the
calculation of the body mass index, by which overall fitness is categorized and body fat content
estimated. It also can serve as one parameter in estimating body surface area, which is a key factor
in some exposure and risk scenarios.

Creating an average growth reference of the relationship between weight and age requires
a database that is representative of the population under consideration, contains accurate
measurements from the sample subjects, and uses a statistical process that appropriately fits smooth
percentile curves to the data. The authors' evaluation of the data within the proposed age bins used
those criteria.

TheNHANES III database, distributed byNCHS, contains measured physical parameters on
a representative population of more than 30,000 individuals between the ages of 2 months and 90
years, collected between 1988 and 1994. The NHANES IE data support classification not only by
age, but also by sex, race, and ethnicity, and reflects actual measurements under consistent
conditions (as opposed to self-reported values), includes data on a large number of individuals
collected as a representative sample of the U.S. population, and contains demographic data that have
been confirmed by in-person interviews with survey respondents.

xxi


-------
The authors selected NHANES III to develop recommendations for the proposed age bins
because it provides data for constructing age-based bins of growth parameters, including age-to-
weight, as well as provides data for which age-to-height measurements and age-to-body mass indices
can be constructed. Therefore, the data sets and graphs are convenient for estimating exposure
factors data for the proposed age bins. In their analysis, the authors selected the 50th percentile
values for each age grouping. For time periods over several months, they summed the values and
computed an average (one value for males and one for females for each age bin).

Although the age-to-weight bins in the CDC growth charts are adequate for estimations of
exposure and risk, normalized by averages of the population weights, that approach is only
minimally adequate because many exposure and risk assessments use other body metrics as key
components. Dermal exposure assessments use surface area factors and are usually related to some
age groups and gender/age subpopulalions. Increasingly, risk assessment considers pharmakokinetic
and pharmacodynamic relationships. The NHANES III survey provides data for all of these
situations. The NHANES III survey could be improved by presenting the age-to-weight estimates
as gender-specific values, particularly values representing ages greater than 2 years. NHANES III
data also could be used to conduct separate analyses for selected ethnic groups ethnic-specific
exposure and risk assessments.

XXll


-------
1.0 INTRODUCTION

1.1 BACKGROUND

The 1993 National Academy of Sciences (NAS) report "Pesticides in the Diets of Infants and
Children" highlighted important differences between children and adults with respect to risks posed
by pesticides. Some of the principles in the NAS report provided the foundation for the Food
Quality Protection Act of 1996 (FQPA) and the President's Executive Order 13045, "Protection of
Children from Environmental Health Risks and Safety Risk." The FQPA requires that children's
aggregate exposure be considered when establishing pesticide tolerances (legal limits for residues
in food). Executive Order 13045 broadens consideration of impacts on children by stating that "each
Federal agency: shall ensure that its policies, programs, activities, and standards address
disproportionate risks to children that result from environmental health risks or safety risks." Many
of the comments the EPA received on the Proposed Guidelines for Carcinogen Risk Assessment
relate to the implementation of Executive Order 13045. In response to these comments and
regulatory initiatives, EPA has been investigating ways to improve Agency risk assessments for
children.

An Agency workgroup convened under the auspices of the Risk Assessment Forum has been
exploring children's exposure assessment issues. This workgroup has concluded that a major issue
facing the Agency is how to consider age-related changes in behavior and physiology when assessing
early lifestage exposure. This issue is critical for scientists involved in preparing exposure
assessments applicable to children or for use in evaluating integrated lifetime exposure. Typically,
Agency assessors have classified individuals under the age of 21 years as youth or children.
However, how to subdivide this group in a consistent and scientifically supported manner has been
somewhat elusive. Children's behavior changes over time in ways that can have an important impact
on exposure. For example, crawling and mouthing of hands and objects during the toddler stage of
life can lead to dermal and oral exposures that are appreciably higher than those of adults. Further,
children's physiology changes over time in ways that can affect both their exposures and their
susceptibility to certain health effects. The key issue is how to capture those changes in an
assessment of risks from exposure to environmental contaminants.

In July 2000, a workshop was convened by EPA to explore how to consider age-related
changes in anatomy and behavior when assessing risk to children. A summary of the workshop
discussions is provided in the document Summary Report of the Technical Workshop on Issues
Associated with Considering Developmental Changes in Behavior and Anatomy When Assessing
Exposure to Children (U.S. EPA, 2000). Although viewing development as a continuum is

1-1


-------
considered the "ideal" approach, workshop participants concluded that age groupings (or bins) can
be useful as a guide for the development of exposure scenarios. To that end, workshop participants
offered some preliminary advice on possible age groups related to developmental change. The
workshop subgroup addressing behavioral development recommended dividing the first year of life
into three groups: 0 to 2 months, 3 to 5 months, and 6 to 11 months. After the first year of life, they
recommended the following groups: 1 to 2 years, 2 to 5 years, 6 to 10 years, 11 to 15 years, and 16
to 20 years. The subgroup addressing anatomy/physiology changes recommended the following
groupings: 0 to 1 month, 1 to 3 months, 3 to 6 months, 6 to 12 months, 1 to 3 years, 3 to 8 (female)
or 9 (male) years, and 8 or 9 years to 16 (female) or 18 (male) years. Prenatal development was
outside the scope of the workshop discussions, but both subgroups stressed the importance of
including this lifestage in exposure and risk assessments.

On the basis of the workshop discussions, EPA is considering developing a minimum set of
childhood age groups for exposure and risk assessment, as follows:

Prenatal

less than 1 month
1 through 2 months
3 through 5 months
6 through 11 months

I	through 2 years
3 through 5 years
6 through 10 years

II	through 15 years
16 through 17 years

At the workshop, in addition to defining the proposed age bins, factors that influence
children's exposure were discussed and techniques currently used to assess their exposures were
addressed. To that end, highlights of the Hubal et al. (2000) paper were discussed and are briefly
summarized to follow. A child's exposure is greatly influenced by where the child is and what the
child is doing. Characteristics of children that influence exposure are physical, behavioral, physical
activities, diet and eating habits, gender, and demographic characteristics. For each exposure route
(inhalation, dermal, ingestion), an exposure algorithm mathematically expresses exposure as a
function of (1) chemical concentration in the exposure medium, (2) contact rate, (3) rate of transfer
from the exposure medium to the portal of entry, and (4) the exposure duration. There are several
key algorithms used when estimating exposure. Key terms used to develop these algorithms are as
follows:

•	Microenvironment—The location a child occupies for a specified period of time, such
as indoors at home in the kitchen.

•	Macroactivitv — Activities that are part of what a child is doing over a specified period
of time, such as watching TV, sleeping, or crawling.

1-2


-------
• Microactivitv — Detailed actions that occur within a general activity, such as hand-to-
surface and hand-to-mouth behavior.

Microenvironmental data have been used for years to evaluate inhalation; however, in recent
years, it became obvious that general activity descriptions do not provide enough information on
specific contact with exposure media within a microenvironment to estimate dermal and non-dietary
ingestion exposures. Activity pattern data requirements are demonstrated in the context of
algorithms for inhalation, dermal contact, and ingestion. These are described in the following
paragraphs (as presented in Hubal et al., 2000).

Inhalation exposure is estimated for each of the microenvironments where a child spends
time and each macroactivity that would result in a different inhalation rate while engaging in that
activity. Exposure over the 24-hour period is then the sum of all of the microenvironmental/
macroactivity (me/ma) exposures. For each individual me/ma, inhalation exposure over the 24-hour
period (Eime/ma) is defined as:

Ejme/ma ~~ Tme/ma X Came X IRma	(1)

where:

"'ime/ma

me/ma

c

ame

IRma

inhalation exposure over the 24-hour period for a single
mi cr oenvir onment al/macro activity (mg/day)
the time spent in that me/ma over the 24-hour period (h/24h)
the air concentration measured in the microenvironment (mg/m3)
the child's respiration rate representing his activity level for that
macroactivity (m3/h)

Two main approaches are currently used to assess dermal and non-dietary ingestion exposure.
To estimate dermal exposure using the macroactivity approach, microenvironments are defined by
location and surface type (e.g., indoors at home on carpet). The dermal exposure associated with a
given macroactivity (e.g., actively playing in the yard) is measured and used to develop an activity-
and microenvironment-specific transfer coefficient. Exposure can then be estimated individually
for each of the microenvironments where a child spends time and each macroactivity that the child
conducts within that microenvironment. Exposure over the 24-hour period is the sum of all of the
microenvironment/macroactivity (me/ma) exposures. For each me/ma, dermal exposure over the
24-hour period (Edme/ma) is defined as:

^dme/ma ^surf x TCder x ED	(2)

1-3


-------
where:

dme/ma
Csurf

TCder
ED

dermal exposure over the 24-hour period for each

mi croenvironmental/macro activity (mg/day)

total contaminant loading on surface (mg/cm2)

dermal transfer coefficient for the me/ma (cm2/hr)

exposure duration that represents the time spent in the me/ma (hr/day)

To assess dermal exposure and non-dietary ingestion using the microactivity approach,
exposure is estimated individually for each of the microactivities or events (e.g., each time a child
touches a given object) from which dermal contact or non-dietary ingestion occurs. Exposure over
the 24-hour period is then the sum of all of the individual exposures. For each microactivity, dermal
exposure over the 24-hour period (EderAni) can be defined as:

Eder/mi = CsurfxTExSAxEF

(3)

where:

der/mi
Csurf

TE
SA
EF

dermal exposure for a given microactivity over a 24-hour period
(mg/day)

total contaminant loading on surface (mg/cm2)

transfer efficiency, fraction transferred from surface to skin (unitless)

area of surface that is contacted (cm2/event)

frequency of contact event over a 24-hour period (events/day)

For each microactivity resulting in non-dietary ingestion, exposure over the 24-hour period
(Endintfmi) can be defined as:

"'nding/mi

i = cx X TExm X SAX X EF

(4)

where:

nding/mi

X

Cx

TExm

SA,

EF

non-dietary ingestion exposure for a given microactivity over a 24-

hour period (mg/day)

hand or object that is mouthed

total contaminant loading on hand or object (mg/cm2)

transfer efficiency, fraction transferred from object or hand to mouth

(unitless)

area of object or hand that is mouthed (cm2/event)
frequency of mouthing event over a 24-hour period (events/day)

1-4


-------
Dietary exposures of young children are difficult to accurately assess or measure. Young
children do not consume foods in a structured manner. While eating, their foods contact surfaces
(hands, floors, eating surfaces, etc.) that may be contaminated. To assess dietary ingestion, exposure
is estimated individually for each item of food consumed by the child. Total dietary exposure is then
the sum of exposures for all food items consumed over a 24-hour period. The intake of a
contaminant associated with one food item, the specific eating activities resulting in that food item's
contact with contaminated surfaces ( z), and specific activities resulting in the food item's contact
with the child's hands before it is eaten (j) can be described as follows:

g _ Cfood WT ^ £j[Csurf TEs/f SAs/f EFs/p] ^ EjfChimj TEn/p SA^, EF^,]
diet Term 1	Term 2	Term 3

where:

Ediet	= Total dietary exposure to the environmental contaminant for one food

eaten (mg/food item)

Cfood = Contaminant concentration of food item after preparation for

consumption (jag/g food)

WT	= Total amount of the individual food consumed (g food/food item)

Csurf = Contaminant loading on a contacted surface (|j,g/cm2)

TEs/f = Surface to food contaminant transfer efficiency (where transfer
efficiency is a function of duration of contact, surface type, moisture,
etc.) (unitless)

SAs/f = Area of contaminated surface that is contacted by the food item
(cm2/event)

EFs/f = Frequency of surface to food contact events that occur during

consumption of the food item (events/food item)

Chand = Contaminant loading on child's hand (|j,g/cm2)

THE/f	= Hand to food contaminant transfer efficiency (unitless

SAh/f = Area of the contaminated hand that is contacted by the food
(cm2/event)

EFh/f = Frequency of hand to food contact events that occur during
consumption of the food item (events/food item)

Hubal et al. (2000) concluded that, currently, data on children's exposures and activities are
insufficient to adequately assess exposures to environmental contaminants. As a result, regulators
use a series of default assumptions and exposures factors when conducting exposure assessments,
the more uncertain the assumptions and exposure factors used, the more conservative they must be
to protect children's health. Data to reduce uncertainty in the assumptions and exposure estimates
are needed to ensure that chemicals are regulated appropriately. To improve the database available
to assess children's exposures, three areas of research are required: identification of appropriate
age/development benchmarks for categorizing children in exposure assessments, development and

1-5


-------
improvement of methods for monitoring children's exposures and activities, and collection of
physical activity data for children (especially young children) required to assess exposure by all
routes. Therefore, this issue paper addresses some of the concerns expressed in the conclusions of
Hubal et al. (2000).

1.2	PURPOSE

The purpose of this issue paper is to explore whether existing scientific data support
defensible child-specific exposure factor recommendations for the EPA proposed age groupings
derived from the July 2000 workshop. To accomplish that task, a panel of experts was selected to
present their individual views for particular exposure factors. The authors evaluated exposure
factors for breast milk intake, food intake, drinking water and total fluid intake, soil ingestion and
pica, non-dietary ingestion factors, inhalation factors, skin surface area, soil adherence to skin, body
weight, and activity factors. In the preparation of this document, the authors reevaluated the data
underlying the exposure factor recommendations in the draft Child-Specific Exposure Factors
Handbook (CSEFH) (U.S. EPA, 2001). Where reevaluation of the underlying data was possible and
productive, the authors recommended exposure factors for the proposed age groups. As part of the
reevaluation, the authors discussed quality assurance issues, including the considerations used in
selecting key studies for the CSEFH. These considerations included the level of peer review,
accessibility, reproducibility, focus on the exposure factor of interest, data pertinent to the United
States, primary data, current information, adequacy of the data collection period, validity of
approach, representativeness of the population, variability in the population, minimal or defined bias
in study design, and minimal or defined uncertainty in the data. In each area of exposure in which
reevaluation of exposure factors data was not possible or would not support the development of
recommendations for the proposed children's age groups, the authors addressed the uncertainties
introduced by using the existing data to assess exposures for children.

1.3	ORGANIZATION

The exposure factors discussed in this issue paper are organized in the same sequence as in
the CSEFH. In addition, the authors evaluated the data (new and existing) using the same
characterization criteria used in the CSEFH. New recommendations for the exposure factors, when
available, were provided for the proposed age bins. In addition, a table is provided in each chapter
that describes the authors' confidence in their conclusions for providing new recommendations. The
criteria used for describing the confidence are the same as those used in Table 1-1 of the CSEFH.

1-6


-------
1.4 REFERENCES

Cohen-Hubal, E.; Sheldon, L.; Burke, J.; McCurty, T.; Berry, M.; Rigas, M.; Zartarian, V.; Freeman,
N. (2000) Children's exposure assessment: a review of factors influencing children's
exposure, and the data available to characterize that exposure. Environmental Health
Perspectives 8(6):475.

U.S. EPA. (1996) Proposed guidelines for carcinogen risk assessment. U.S. EPA, Washington, DC,
EPA/600/P-92/003C. Federal Register 61(79): 17960-18011.

U.S. EPA. (2000) Summary report of the technical workshop on issues associated with considering
developmental changes in behavior and anatomy when assessing exposure to children. Risk
Assessment Forum, Washington, DC; EPA/63 0/R-00/005.

U.S. EPA. (2001) Child-specific exposure factors handbook. Prepared by Versar, Inc., for the Office
of Research and Development, National Center for Environmental Assessment, under EPA
contract no. 68-W-99-041.

1-7


-------
2.0 BREAST MILK: INTAKE, NURSING DURATION, AND FAT COMPOSITION

2.1 INTRODUCTION

Breast milk is a potential source of exposure to toxic substances for nursing infants. Lipid-
soluble chemical compounds may accumulate in body fat and may be transferred to breast-fed infants
in the lipid portion of breast milk. Because nursing infants obtain most (if not all) of their dietary
intake from breast milk, theyare especially vulnerable to exposures to these compounds. Estimating
the magnitude of the potential dose to infants and the duration of that exposure requires information
on the milk intake rate (quantity of breast milk consumed per day) and the duration over which
breast-feeding occurs. Information about the fat composition of breast milk is also needed to
estimate the possible concentration of the chemical moieties in the milk. The delivered dose of toxic
chemical can then be estimated from breast milk residue concentrations that have been indexed to
lipid content.

Several studies have provided data applicable to the EPA proposed age bins. Breast milk
intake has been studied in both cross-sectional and longitudinal studies with mother/infant pairs.
Study designs varied, but typically comprised fairly small cohorts, given the sensitivity and technical
complexities of such data collection. In order to minimize confounding factors in such studies, the
participants were confined to very homogeneous demographic factors and anthropometrics. While
these designs enhance the focus on the variable under study, it is difficult to know how the results
can be extrapolated across a population that is far more diverse than the universe of test subjects.
In the cross-sectional studies, a given period or a few periods in time postpartum were chosen as the
points of data collection. This minimized variation introduced by changes in infant size and dietary
choices, and permitted more samples over the shorter viewing period. Longitudinal studies may
follow a smaller cohort across longer periods of time, but must account for the variables introduced
over those time changes. In addition, studies may measure intake as a function of individual feeding
events, as a summation of 24-hour periods, or as a mean of multiple 24-hour periods (approximately
contiguous days).

The most typical measurement tool used in the studies was the test weighing methodology
in which the freshly diapered infant was weighed before and after each feeding, with no diaper
change between weighings. Corrections on this calculated weight were made to adjust for the
possible loss of water during the period of feeding (insensible weight loss), and to convert to volume
of milk intake using the density factor of the breast milk. The first factor was typically about 2 g/kg
body weight per hour, and the second factor typically fell between 1.01 and 1.03 g/mL breast milk.
The data were reported with one or both of these factors and in some cases included estimations of

2-1


-------
those factors from previously reported studies, or the data were independently determined within the
study. The insensible weight loss was calculated using estimates of feeding durations or from daily
diary records of the study participants. The resulting weight or volume measurements of breast milk
intake was reported as discreet feeding events or summed over a 24-hour period. Means of multiple
daily intakes may minimize the intra-individual variability.

The test weighing procedures employed by the study may be validated within the study
design. To do this, the procedure is applied to bottle-fed infants and the milk weight is directly
measured before and after feeding. The known weight of the consumed milk is compared with the
measured change in the infant's weight between commencement and conclusion of feeding. This
methodology has been shown to quite accurately measure the breast milk intake. Neville et al.
(1988) estimated that the error was only about 2.5 grams per feeding. Note that this is not very
different from the average value of the predicted insensible weight loss of 2.8 grams per feeding.

Results of these intake studies among "privileged infants," born to well-educated, healthy,
nonsmoking parents in the middle to upper income range, showed a general intake level of 600 to
900 g/day. These studies reported rather significant inter-individual differences among the infants
(10 to 20 percent), even with these homogeneous cohorts. This was true even when the cohort group
began with infants with weights at birth or 1 month that are considered to be in the normal range for
full-term, healthy infants. Intra-individual variation (day-to-day) also was high in most studies,
though generally less than the inter-individual variation. Exploring the reasons for these differences
was the objective of the Davis Area Research on Lactation, Infant Nutrition, and Growth
(DARLING) study by Dewey et al. (1991 a, 199 lb). Results from this study suggest that we may be
able to apply the findings from homogeneous populations of mothers and infants to a broader section
of the population, where demographic, anthropometric, and physiological differences are surely
greater than in the study cohorts. These results also suggest that older studies may be adequately
representative of the current population, even though maternal characteristics may have changed with
time.

Descriptions of several key studies on breast milk intake are presented in Section 2.2.1.
Section 2.2.2 discusses the research on breast milk composition, particularly estimation of fat
content.

2-2


-------
2.2

EVALUATION OF EXISTING DATA

2.2.1 Studies on Breast Milk Intake

In Milk Intakes and Feeding Patterns of Breast-fed Infants, Pao et al. (1980) conducted a
study of 22 healthy breast-fed infants to estimate breast milk intake rates. Infants were categorized
as completely breast-fed or partially breast-fed. The goal of the study was to enroll infants as close
to 1 month of age as possible and to obtain records near 1, 3, 6, and 9 months of age. Data were
collected for these 22 infants using the test weighing method. Records were collected for three
consecutive 24-hour periods at each test interval. The weighing methodology was tested for
accuracy and determined to be accurate to 95 percent of reference tests using bottle-fed infants.
Measurements were not corrected for insensible water loss. The weight of breast milk was converted
to volume by assuming a density of 1.03 g/mL for all feeding periods and for both completely breast-
fed (CBF) infants and partially breast-fed (PBF) infants. Daily intake rates were calculated for each
infant based on the mean of the three 2-hour periods. Mean daily breast milk intake rates for the
infants surveyed at each time interval are presented in Table 2-1 of the Child-Specific Exposure
Factors Handbook (2001 CSEFH).

The study presented valuable information over multiple time periods, minimizing daily
variability with the use of 3 contiguous days of measurement at each age interval. However, several
issues compromise the quantitative conclusions. The utility of these data as nationally representative
age-related breast milk intake values is questionable. The data are over 34 years old, raising issues
about the possible changes in infant weights and trends of intake over the generations.

I n Milk and Nutrient Intakes of Breast-fed Infants from 1 to 6 months, Dewey and Lonnerdal
(1983) monitored the dietary intake of 20 breast-fed infants between the ages of 1 and 6 months.
Most were completely breast-fed, five had been given some formula, and several were fed small
amounts of solid foods after 3 months of age. A second objective of the study was to estimate
nutrient intake and examine nutrient concentration and milk volume. This second objective provided
data useful for the consideration of intake calculations and comparison with calculations made in
other studies.

Dewey and Lonnerdal noted the very wide range of breast milk intake rates, even among this
very homogeneous group for which reporting precision would be expected to be excellent (see Table
2-2 in the 2001 CSEFH). Variation over time was particularly pronounced; at 1 month, intake
showed an almost threefold variation and at 6 months was twofold. The energy needs of the infants

2-3


-------
seemed to vary greatly, but no consistent relationship was found between breast milk intake volume
and the concentrations of nutrients in the milk during the first 6 months.

Study of the energy intake parameters revealed interesting correlations. Weight gain was
positively correlated with energy intake per kilogram body weight at 1 -2 months. But by 5-6 months
the correlation reversed. The fattest infants consumed less per unit weight at 5-6 months than the
lean infants, but such correlations were not statistically significant. Mean intake rates ranged from
673 mL/day for a 1-month old infant to 896 mL/day for a 6-month old infant. The intake findings
confirm the findings of previous studies, which placed breast milk consumption between 600 and
800 mL/day, with great variation between different infants. Breast milk intake was relatively
constant between 2 and 5 months.

Application of these measurements to contemporary infants in the general U.S. population
is compromised by the size of the measured cohort, lack of underweight or overweight infants, lack
of premature births in the cohort, age of the study (18 years), and lack of multiple ethnic groups.

In Human Milk Intake and Growth in Exclusively Breast-Fed Infants, Butte et al. (1984)
studied breast milk intake in exclusively breast-fed infants during the first 4 months of life. Breast-
feeding mothers were recruited through the Baylor Milk Bank Program in Texas. These mothers
were 18-36 years of age, predominantly white (41 white, 2 Hispanics, 1 Asian, and 1 West Indian),
very well educated, healthy nonsmokers, and all were professionals or technicians. Their infants
were full-term, the first or second child, healthy, and of normal weight. Infant growth progressed
satisfactorily over the course of the study. The amount of milk ingested over a 24-hour period was
determined using the test weighing procedure. The procedure was validated with bottle-fed infants
for this study, with a 3.2- gram difference noted. For most participants, test weighing was conducted
over single 2-hour periods monthly; however, in an attempt to capture information about intra-
individual variation, Butte et al. weighed participants in 48- to 96-hour continuous test weighings.
This study provided at least 37 infants per monthly measurement period.

Breast milk intake was relatively constant across time periods in this study (see Table 2-3 in
the 2001 CSEFH). The mean intake was 733 g/day (712 mL/day) using a density factor of 1.03
g/mL. The range was small, with 723 g/day at 3 months to 751 g/day at 1 month for the means of
the group at each time period. Inter-individual variation was found to be 17 percent, even within this
homogeneous cohort. Intra-individual variation in daily intake was estimated to be 7.9 ± 3.6 percent.
The conclusions of this study are consistent with the findings in the literature that report variability
in intakes of infants of the same age to be 11 to 29 percent. Also, it agrees with literature that reports
breast milk intake throughout the first 4 months of life at 600-900 g/day.

2-4


-------
In Studies on Human Lactation, Neville et al. (1988) studied breast milk intake among 13
infants during the first year of life in a longitudinal study designed to characterize the temporal
course of lactation. Such understanding could improve the use of breast milk intake data from
mother/infant pairs in cross-sectional studies. The study design, as well as selection of subjects from
a very homogeneous cohort, maximizes the opportunity to understand sources of intra-subject
variation and possible relationship of intake to milk production and transfer parameters.

Dailymilk intake was estimated by the test weighing method. Data were collected daily from
birth to 14 days, weekly from weeks 3 through 8, and monthly until the study period ended at 1 year.
Results from this study were analyzed for correlations between mean daily intake and birth weights,
infant weight at 1 month, infant weight gain 1 month postpartum, and total milk intake at 1 month
postpartum. There was little or no correlation with birth weight over the 5 months of intake
measurement, but there was consistent correlation with infant weight gain at 1 month postpartum.
The data also suggest that total milk intake through 5 months of age can be related to the total milk
intake at 1 month postpartum. Thus, high-intake infants seem to remain high-intake infants, and vice
versa.

This study is valuable for setting nationally representative breast milk intake values at various
time periods, with some strong qualifying considerations. The measurement of breast milk intake
during the first month postpartum is rare in a study design, and even with the limited number of
subjects, one can see the rapid onset of significant intake and a steadily increasing slope of intake
rate. The longitudinal study design suggests that, although there is the same inter-individual
variability as we saw in all the cross-sectional studies, individuals tended to stay constant in terms
of their position within that range over time. Mean intake quickly reached 600 g/day, with a
maximum intake near 800 g/day by 5 months (see Table 2-4 in the 2001 CSEFH). This information
is consistent with data from cross-sectional studies on similar types of subjects. As in other studies,
the virtues of minimizing the variability during study observations by maintaining a strictly
homogeneous cohort create problems when considering how to apply such information to the general
population with vastly more opportunity for variation.

The DARLING Study was conducted by Dewey et al. in 1986 to evaluate growth patterns,
nutrient intake, morbidity, and activity levels in infants who were exclusively breast-fed for at least
the first 12 months of life (Dewey et al., 1991a, 1991b). The study used a 4-day test weighing
procedure for 73 infants age 3 months. The study was designed to assess the factors that influence
the very wide range in intake among normal infants and measured total volume of breast milk
extracted.

2-5


-------
The mean intake was estimated at 812 g/day, 769 g/day, 646 g/day, and 448 g/day for infants
ages 3 months, 6 months, 9 months, and 12 months, respectively (see Table 2-5 in the 2001 CSEFH).
Variability in residual milk volume was high. The average day-to-day coefficient of variation for
all 73 subjects was 8.9 ± 5.4 percent, and inter-individual coefficient ofvariation was 16.3 percent.
The mean intake for these infants was within the range of the 600 to 900 g/day reported in the
literature. Intake measured from the same cohort at other intervals up to 12 months of age also
showed patterns consistent with those described in other studies (Dewey et al., 1991b). Because
there were more mother/infant pairs included in this study than in most similar studies reported in
the literature, and because the study assessed contributors to the intake variability, the study is
valuable for the determination of breast milk intake in the general U.S. population. Most studies
employ a very homogeneous cohort in order to minimize confounding factors among the small
number of participants. Results of this study also suggest that differences in many demographic and
personal traits may not play an important role in the frequently observed large differences in breast
milk intake by infants in any age group. Thus, the findings in these studies maybe applicable to the
overall U.S. population.

2.2.2 Studies on Lipid Content of Breast Milk and Fat Intake from Breast Milk

In Human Milk Intake and Growth in Exclusively Breast-fed Infants, Butte et al. (1984)
studied breast milk intake in exclusively breast-fed infants during the first 4 months of life. Breast-
feeding mothers were recruited through the Baylor Milk Bank Program in Texas. These mothers
were 18-36 years of age, predominantly white (41 white, 2 Hispanics, 1 Asian, and 1 West Indian),
very well educated, healthy nonsmokers, and all were professionals or technicians. Their infants
were full-term, the first or second child, healthy, and of normal weight. Infant growth progressed
satisfactorily over the course of the study. The amount of milk ingested over a 24-hour period was
determined using the test weighing procedure. This study provided at least 37 infants per monthly
measurement period.

Milk was collected for compositional studies within 3 days after the test weighing procedure
for milk intake estimation. The breast milk was harvested with an Engnell electrical breast pump,
and fat content was determined gravimetrically after methylene chloride extraction (modification of
Roese-Gottlieb method). The fat concentration (mg fat/g milk) of the breast milk at each of the four
monthly testing periods was 36.2, 34.4, 32.2, and 34.8 for infants ages 1 month, 2 months, 3 months,
and 4 months, respectively (see Table 2-6 in the 2001 CSEFH). Fat concentration remained constant
across these 4 months. Since intake of milk was constant in the study, intake changed only as a
function of infant body weight. Because all of the women in the study were well nourished and had
no compromising health problems or dietary restrictions, we could not tell from this study if it is

2-6


-------
reasonable to extrapolate these conclusions to populations of women who are nutritionally deprived
for any reason.

In Milk and Nutrient Intakes of Breast-fed Infants from 1 to 6 months, Dewey and Lonnerdal
(1983) monitored the intake of 20 breast-fed infants between the ages of 1 and 6 months. A key
objective of the study was to estimate nutrient intake and examine nutrient concentration and milk
volume. The participants were very well educated, well-nourished mothers 21 to 36 years of age
who were recruited from Lamaze childbirth classes in California. Some were taking nutrient
supplements. All but one infant remained well within the normal ranges of weight and growth
during the study, and none was obese.

Milk samples were collected from each mother at the second feeding of the morning on the
day after the two 2-hour weighing records. The second feeding was chosen in an attempt to
minimize diurnal differences in fat concentration. Fat was measured as total lipids. Fat
concentration (g/100 mL milk) remained constant across the 6 months of testing, with very little
variation around the mean values. The slight decline in total lipid concentrations between 1 and 6
months was not statistically significant. These data are presented below in Table 2-1.

Table 2-1. Mean Fat Content of Breast Milk Intake Samples for Infants 1-6 Months

Months

1

2

3

4

5

6

g/100 mL 4.92 (1.05)

4.58(0.97)

4.58 (1.65)

4.62 ( 1.86)

4.36 ( 1.67)

4.30 (1.96)

mg/gma 47.4

44.1

44.1

44.6

42.0

41.5

a Assumes milk density of 1.037.

Source: Dewey and Lonnerdal (1983).

Again, data from the cohort used in this study do not permit extrapolation to the general population
without concern as to the possible influence of maternal nutritional status.

2.3 STUDIES SELECTED FOR THE ANALYSIS TO OBTAIN NEW
RECOMMENDATIONS

2.3.1 Studies Selected for Estimating Breast Milk Intake

The studies ofNevilleetal. (1988), Pao et al. (1980), Dewey et al. (1991a, 1991b), and Butte
et al. (1984) provide data on intake rates of breast milk at various ages, from which we could derive

2-7


-------
factors for the proposed age bins. Although only Neville et al. studied the infants' intake during the
first few days of life, all four studies provided data relevant to some of the age bins up to 1 year. The
four studies generally agreed that inter- and intra-individual variation is great and that there is a
predictable variation of intake among the feeding times within a day. The pattern suggests that the
early weeks after birth are a period of rapid increase in intake rates in infants. During months 3-7,
the intake rate peaks, then levels off somewhere between 700 and 800 g/day. From 8 to 12 months,
the intake rate drops off to approximately half of that value. On a g/kg body weight basis, the
decrease appears even more precipitous.

2.3.2 Studies Selected for Estimating Breast Milk Fat Content

Studies by Butte et al. (1984) and Dewey and Lonnerdal (1983) presented data on the fat
content of breast milk. However, these studies differ in their analytical approaches to measuring the
lipid content, and even in their definition of a fat. Other variables may confound the use of these
data, but they are adequate for estimating breast milk fat content.

2.4 RECOMMENDATIONS FOR PROPOSED AGE BINS

In this analysis, we expand the results of studies of homogeneous populations with small
cohorts to a large, diverse U.S. population and, therefore, cannot calculate exact time-weighted
averages or statistical boundaries for these values. We infer the variabilities and uncertainties shown
in the four studies by rounding, which avoids the appearance of precision while providing reasonable
estimations of the breast milk intake. Given these and other sources of variability, the estimates of
intake for various age groups can be derived from these data by calculating averages across the
findings in different studies or by deriving time-weighted averages across time periods. The average
is then rounded up to the nearest 10. The recommended breast milk intake values for the proposed
age bins are presented in Table 2-2. The confidence in ratings for the recommendations are shown
in Table 2-3.

2-8


-------
Table 2-2. Recommended Breast Milk Intake and Proposed Age Bins

Age Categories (Bins)

Intake (g/day)

< 1 month

650

1-2 months

680

3-5 months

780

6-11 months

740

12 months

410

Sources: Data adapted from Butte et al. (1984), Dewey et al. (1991b), Neville et al. (1988), and Pao et al.
(1980).

2.4.1 Breast Milk

Birth Through 1 Month of Age

Only Neville et al. measured milk intake from the first day postpartum. The number of
participants was low, but these are rare and valuable measurements. It is difficult to select a single
value to represent this first month, as lactation appears to commence on the second or third day, and
milk intake dramatically increases with each subsequent week during the first month. An average
of daily consumption for days 3 to 7 is 480 g/day; for days 8 through 11 it is 591 g/day; at 14 days
it is 653 g/day; at 21 days it is 651 g/day; and at 28 days it is 770 g/day. A simple weighted average
intake over the month is 643 g/day.

We recommend that the weighted average from the Neville et al. study, 643 g/day, be used
as only the basis of the estimate. This value may infer precision and conformity that is not realistic.
Also, the value is derived from a cohort that is not representative of the total U.S. population and is
averaged over time periods of great variability. Therefore, following the suggested procedure for
rounding, the recommended value for the bin is 650 g/day.

Age Bin 1-2 Months

All four selected studies measured intake during the period 1-2 months. Pao et al. estimated
intake at 1 month to be 600 g/day. Dewey et al. (1991b) measured intake at 1 month to be 673 g/day
and at 2 months to be 756 g/day. Butte et al. measured intake at 1 and 2 months to be 751 and 725
g/day, respectively. Neville et al. estimated intake at 701 g/day by time-weighted averages of the
reported values. These values, if averaged, suggest an intake of 677 g/day. Again, the value

2-9


-------
Table 2-3. Confidence in Recommendations for Breast Milk Intake and Breast Milk Fat Content









Rating (High, Medium, Low)

















Age









Considerations

<1 Month

1-2 Mos

3-5 Mos

6-11 Mos

1 Yr

2-5 Yrs

6-10 Yrs

11-15 Yrs

16-17 Yrs

Study Elements



















• Level of peer review

High

High

High

High

High

NA

NA

NA

NA

• Accessibility

High

High

High

High

High

NA

NA

NA

NA

• Reproducibility3

High/Low

High/Low

High/Low

High/Low

High/Low

NA

NA

NA

NA

• Focus on factor of interest

High

High

High

High

High

NA

NA

NA

NA

• Data pertinent to U.S.

Low

Low

Low

Low

Low

NA

NA

NA

NA

• Primary data

Med

Med

Med

Med

Med

NA

NA

NA

NA

• Currency

Med

Med

Med

Med

Med

NA

NA

NA

NA

• Adequacy of data collection periodb

Med/Low

Med/Low

Med/Low

Med/Low

Med/Low

NA

NA

NA

NA

• Validity of approach

High

High

High

High

High

NA

NA

NA

NA

• Representativeness of the
population

Low

Low

Low

Low

Low

NA

NA

NA

NA

• Characterization of variability in the
population11

Med/Low

Med/Low

Med/Low

Med/Low

Med/Low

NA

NA

NA

NA

• Lack of bias in study design

High

High

High

High

High

NA

NA

NA

NA

• Measurement errorb

Med/Low

Med/Low

Med/Low

Med/Low

Med/Low

NA

NA

NA

NA

Overall Rating

Med

Med

Med

Med

Med

NA

NA

NA

NA

a High for breast milk intake, low for breast milk fat content.
b Medium for breast milk intake, low for breast milk fat content.

2-10


-------
suggests precision and conformity beyond the facts of these studies. Using the rounding convention,
the intake for this bin is 680 g/day.

Age Bin 3-5 Months

For this age group, the mean intakes varied greatly, within studies and between studies, from
711 g/day to 833 g/day. In all four studies, the trend seems to indicate that this is a time of high
intake per day. The average of values for this time period with these studies is 779 g/day. Using the
rounding convention, the intake for this bin is 780 g/day.

Age Bin 6-11 Months

Three of the studies reported intake measurements within the period 6-11 months of age. Pao
et al. estimated intake at 6 months to be 682 g/day for completely breast-fed infants. Dewey et al.
(1991b) had a much higher estimate of 896 g/day for the same time point. Neville et al. reported
intake for infants on a monthly basis for each month in this period. The average of those months is
638 g/day, but the data suggest a high intake period during months 6-7, similar to months 3-5.
Thereafter, breast milk consumption decreases significantly with each succeeding month. This
suggests that the binning period comprises two (perhaps even three) binning populations. If all
studies are averaged and estimates at 6 months are extrapolated over the range of the binning period,
the estimated intake would be 738 g/day. Thus, the rounded bin value is 740 g/day.

Age Bin 12 Months

Neville et al. reported intake at 1 year of age to be only 403 g/day. By this time many diets
are supplemented with food and other fluids. Energy demand is high but is met by these additions
to the breast milk intake. Rounding by our convention gives us a value of 410 g/day.

2.4.2 Recommendations for Breast Milk Fat Concentration

Two of the studies, Butte et al. and Dewey and Lonnerdal, measured fat content in breast
milk within the narrowly defined populations. Analysis methodologies and the resulting definitions
of fat and lipid content differed for these studies. Several issues make it difficult to extrapolate these
values across the population for all age groups and for all measurement durations. Variables such
as diurnal variability, nutritional and health status of the mother before and during lactation, and
changes during periods of stress were addressed. Using the approximated averages in Butte et al.
(1984) (Table 2-6 in the 2001 CSEFH) and Dewey and Lonnerdal (1983) (Table 2-1 in this report),

2-11


-------
we recommend that a value of 4 percent lipid content in breast milk at all ages of nursing (all age
bins) be used for assessing exposures to toxic chemicals.

2.5	RECOMMENDATIONS FOR FURTHER ANALYSIS AND RESEARCH NEEDS

Key issues that need to be addressed include:

•	Estimation of the variability across the population for milk intake as a function of age
and/or infant weight, considering the maj or ethnic groups in the U. S. population. Studies
on black, Asian, and Hispanic mother/infant groups are needed.

•	Estimation of the effect of nutrient status of the mother before and during lactation on
the fat content of the milk. Data are needed on the types of lipids that may change
because of these variables and the mobility of such lipids in the milk during lactation.

2.6	REFERENCES

Butte, N.F.; Garza, C.; O'Brian Smith, E.; Nichols, B.L. (1984) Human milk intake and growth in
exclusively breast-fed infants. Journal of Pediatrics 104:187.

Dewey, K.G.; Lonnerdal, B. (1983) Milk and nutrient intake of breast-fed infants from 1 to 6
months: relation to growth and fatness. Journal of Pediatric Gastroenterology and Nutrition.
2:497-506.

Dewey, K.G.; Heinig, M.J.; Nommsen, L.A.; Lonnerdal, B. (1991a) Maternal versus infant factors
related to breast milk intake and residual milk volume: the DARLING study. Pediatrics
87:829-837.

Dewey, K.G.; Heinig, M.J.; Nommsen, L.A.; Lonnerdal, B. (1991b) Adequacy of energy intake
among breast-fed infants in the DARLING study: relationships to growth velocity, morbidity,
and activity levels. Journal of Pediatrics 119:538-547.

Neville, M.C., Keller, R., Seacat, J., Lutes, V., Neifert, M., Casey, C., Allen, J., Archer, P. (1988)
Studies in human lactation: milk volumes in lactating women during the onset of lactation
and full lactation. American Journal of Clinical Nutrition 48:1375-1386.

Pao, E.M., Himes, J.M., Roche, A.F. (1980) Milk intakes and feeding patterns of breast-fed infants.
Journal of the American Dietetic Association 77:540-545.

Zollner, N., Kirsch, K. (1962) Uber die quantitative Bestimmung von Lipoiden (mikromethode)
mittels der vielen naturlichem lipoiden (alien bakannten phospholipoiden) gemeinsamen
sulfophosphovanillin-reaktion. Zeitschrift fur die Gesamte Experimentelle Medizin
Einschliesslich Experimentelle Chirurgie 135:545-561.

2-12


-------
3.0 FOOD INTAKE

3.1	INTRODUCTION

Ingestion of contaminated foods is a potential pathway of exposure to toxic chemicals among
children. Children's exposure from food ingestion may differ from that of adults because of
differences in the types and amounts of food eaten. Also, for many foods, the intake per unit body
weight is greater for children than for adults. The Child-Specific Exposure Factors Handbook
provides information on food intake rates for children in the following age groups: <1 year, 1-2
years, 3-5 years, 6-11 years, and 12-19 years (U.S. EPA, 2001). The age groups that the EPA Risk
Assessment Forum are interested in are as follows: < 1 month, 1-2 months, 3-5 months, 6-11
months, 1-2 years, 3-5 years, 6-10 years, 11-15 years, and 16-17 years. This chapter addresses the
reevaluation of the data used in the CSEFH to provide food intake rates in the age categories of
interest to the Risk Assessment Forum. The reevaluation includes intake rates for various food
groups and individual foods, as well as the total dietary analysis. Intake rates for home-produced
foods were not evaluated here because the data required to conduct such analyses are not available
in USDA's 1994-96 Continuing Survey of Food Intake among Individuals (CSFII) data set (USDA,
1998). Earlier analyses of intake of home-produced food were based on the 1987-88 Nationwide
Food Consumption Survey. These data are no longer readily accessible, but could be evaluated if
adequate resources were available. The CSEFH also provides data on serving size from Pao et al.
(1982). This chapter does not reevaluate serving size because the Pao et al. report provides data for
only the age bins shown in the CSEFH and could not be reorganized into the age bins of interest to
the Risk Assessment Forum. Finally, this chapter does not address fish intake because this exposure
factor was outside the scope of this project.

3.2	EVALUATION OF EXISTING DATA

The primary source of recent information on consumption rates of foods among children is
the U.S. Department of Agriculture's (USDA) Nationwide Food Consumption Survey (NFCS) and
the USDA 1994-96 CSFII. Data from thel994-96 CSFII were used in the Child-Specific Exposure
Factors Handbook to generate children's per capita intake rates for both individual foods and the
major food groups. As stated in the CSEFH (U.S. EPA, 2001):

USDA conducts the CSFII annually to "assess food consumption behavior and nutritional
content of diets for policy implications relating to food production and marketing, food
safety, food assistance, and nutrition education" (USDA, 1995). The survey uses a
statistical sampling technique designed to ensure that all seasons, geographic regions of the

3-1


-------
U.S., and demographic and socioeconomic groups are represented. Using a stratified
sampling technique, individuals of all ages living in selected households in the 50 states and
Washington, D.C., were surveyed. Individuals provided 2 non-consecutive days of data,
based on 24-hour recall. The 2-day response rate for the 1994-96 CSFIIwas approximately
76 percent. Data from the 1994,1995, and 1996 CFSII were combined into a single data set
to increase the number of observations available for analysis. Approximately 15,000
individuals provided intake data over the 3 survey years (USDA, 1998).

Because of the relatively large sample size associated with this data set, it may be used to
generate intake rates for major crop groups and commonly eaten foods for various age groups of the
population. The CSEFH generated intake rates for the following age groups of children: <1 year,
1-2 years, 3-5 years, 6-11 years, and 12-19 years using the 1994-96 CSFII data (U.S. EPA, 2001).
The sample size for each of these age groups is shown in Table 3-1. The age groups in the CSEFH
differ from those of interest to the Risk Assessment Forum; thus, a reanalysis of the CSFII data is
needed to generate intake rates for the age groups specified in this project. Although this type of
analysis is possible, the number of observations in some of the new age groups will be lower than
those used in the analysis for the CSEFH. Table 3-2 provides the number of observations in the
1994-96 CSFII data set that maybe used for the selected age groups.

Table 3-1. Weighted and Unweighted Number of Observations used in the 1994-96 CSFII
Analysis, for the Child-Specific Exposure Factors Handbook Age Groups



Weighted

Unweighted

Age

Number of

Number of

Group

Observations

Observations

< 1 year 3,772,296 359
1-2 years 8,270,523 1,356
3-5 years 12,376,836 1,435
6-11 years 23,408,882 1,432
12-19 years	29,657,098	1,398

3-2


-------
Table 3-2. Weighted and Unweighted Number of Observations Used in the
Reanalysis of the 1994-96 CSFII for the Selected Age Bins



Weighted

Unweighted

Age

Number of

Number of

Group

Observations

Observations

< 1 month

150,104

15

1 -2 months

729,143

65

3-5 months

1,124,897

119

6-11 months

1,768,152

160

1 -2 years

8,270,523

1,356

3-5 years

12,376,836

1,435

6-10 years

19,498,495

1,189

11-15 years

19,268,648

1,005

16-17

7,760,616

363

For the purposes of this paper the 1994-96 CSFII data were used. Because Versar used the
1994-96 CSFII data and analyses for the Child-Specific Exposure Factors Handbook, the data and
programs were readily available for use in this reanalysis. It should be noted, however, that the 1998
CFSII data became available after the analysis of the 1994-96 CSFII data was conducted for the
Exposure Factors Handbooks. The 1998 data set provides additional data for children and would
be a useful supplement to the 1994-96 CSFII data set in evaluating food intake for children (it was
not used in this reanalysis because a significant effort would be required to rewrite the statistical
program to incorporate these new data). As stated in the documentation for the 1998 data set (USDA,
2000):

The goal of the sample design for the CSFII 1998 was to obtain nationally representative
samples of noninstitutionalized persons 9 years of age or younger residing in households in
the United States for each of 28 analytic domains defined by sex, age (7 age groups), and
income level (a "low-income" group and an "all-income" group). The age groups used were
under 1 year, 1 year, 2 years, 3 years, 4 years, 5 to 6 years, and 7 to 9 years.

Section 3.3 of this document provides details on the methods used to generate intake rates
for the selected age groups using the 1994-96 CSFII data. Two types of analyses were conducted:
(1) individual intake rates, and (2) total diet analysis.

3-3


-------
3.3 ANALYSES USED TO OBTAIN NEW FOOD INTAKE RECOMMENDATIONS
FOR CHILDREN

3.3.1	Individual Intake Rates

The food groups selected for this analysis include the same food groups evaluated in the
CSEFH. These include the major food groups including total fruits, total vegetables, total grains,
total meats, and total dairy; and a variety of individual fruits, vegetables, grains, meats, and dairy
products. Various USDA food categories (i.e., citrus and other fruits, and dark green, deep yellow,
and other vegetables), and protected and exposed produce were also evaluated, as in the CSEFH.
Intake rates of total vegetables, tomatoes, and white potatoes, total meats, fish, beef, pork, poultry,
dairy, eggs, and total grains were adjusted to account for the amount of these food items eaten as
meat and grain mixtures, as described in Appendix 3 A of the CSEFH. Note that fish is included here
because it is a component of total meats. If fish were excluded, the total meat value would not equal
100 percent. Food items/groups were identified in the CSFII database according to USDA-defined
food codes. Appendix 3B of the CSEFH presents the codes and definitions used to determine the
various food groups used in the analysis. Intake rates for these food items/groups represent intake
of all forms of the product (i.e., home produced and commercially produced).

Individual identifiers in the database were used throughout the analysis to categorize
populations according to demographics. These identifiers included identification number, age, body
weight, weighting factor, and number of days that data were reported. Distributions of intake were
determined for children who provided data for 2 days of the survey. Individuals who did not provide
information on body weight, or for which identifying information was unavailable, were excluded
from the analysis. Two-day average intake rates were calculated for all individuals in the database
for each ofthe fooditems/groups. These average daily intake rates were divided by each individual's
reported body weight to generate intake rates in units of g/kg/day. The data were also weighted
according to the 2-day weights provided in the 1994-96 CSFII. USDA sample weights are calculated
to account for inherent biases in the sample selection process and to adjust the sample population
to reflect the national population. Summary statistics for individual intake rates were generated on
a per capita basis. That is, both users and non-users of the food item were included in the analysis.

3.3.2	Total Diet Analysis

Using data from the 1994-96 CSFII, this reanalysis also evaluated total dietary intake. Total
dietary intake was defined as intake of the sum of all foods in the following major food groups:
dairy, eggs, meats, fish, fats, grains, vegetables, and fruits, using the same food codes and method

3-4


-------
for allocating mixtures as those described in Appendix 3B of the CSEFH. Beverages; sugar, candy,
and sweets; and nuts and nut products were not included. Distributions of total dietary intake were
generated for various age groups, as described previously for the major food groups. Means,
standard errors, and percentiles of total dietary intake were estimated in units of g/kg/day, as well
as g/day.

To evaluate variability in the contributions of the major food groups to total dietary intake,
this reanalysis ranked individuals from lowest to highest, based on total dietary intake. Three subsets
of individuals were defined, as follows: a group at the low end of the distribution of total intake (i.e.,
below the 10th percentile of total intake), a central group (i.e., the 45th to 55th percentile of total
intake), and a group at the high end of the distribution of total intake (i.e., above the 90th percentile
of total intake). Mean total dietary intake (in g/day and g/kg/day), mean intake of each of the major
food groups (in g/day and g/kg/day), and the percentage of total dietary intake that each of these food
groups represents were calculated for each of the three populations (i.e., individuals with low-end,
central, and high-end total dietary intake). A similar analysis was conducted to estimate the
contribution of the major food groups to total dietary intake for individuals at the low end, central,
and high end of the distribution of total meat intake, total dairy intake, total meat and dairy intake,
total fish intake, and fruit and vegetable intake. For example, to evaluate the variability in the diets
of individuals at the low end, central range, and high end of the distribution of total meat intake,
survey individuals were ranked according to their reported total meat intake. Three subsets of
individuals were formed as described above. Mean total dietary intake, intake of the major food
groups, and the percent of total dietary intake represented by each of the major food groups were
tabulated. This analysis was conducted for the following age groups of the population: < 1 month,
1 -2 months, 3-5 months, 6-11 months, 1-2 years, 3-5 years, 6-10 years, 11-15 years, and 16-17 years.
The data were tabulated in units of g/day and g/kg/day. Summary statistics include the number of
weighted and unweighted observations, percentage of the population using the food item/group being
analyzed, mean intake rate, standard error, and percentiles of the intake rate distribution (i.e., 0, 1,
5, 10, 25, 50, 75, 90, 95, 99, and 100th percentile or maximum observed in the survey). The food
analysis was accomplished using the SAS statistical programming system (SAS Institute, 1990).

3.4 RECOMMENDATIONS FOR PROPOSED AGE BINS

3.4.1 Results of Reanalysis

Tables 3-3 through 3-14 (at the end of this chapter) present the results of the reanalysis.
Table 3-3 provides per capita intake rates in units of g/kg/day for the total diet and major food
groups. Table 3-3a provides the same information in units of g/day. Per capita intake rates for

3-5


-------
individual foods are provided in Table 3-4. As in the CSEFH, these tables do not provide full data
distributions (i.e., only means, standard deviations, and percentage consuming), because the numbers
of observations for individual foods are lower than for the major food groups and are not believed
to be sufficient for generating distributions. Table 3-5 presents per capita intake rates for various
USDA food categories, and Table 3-6 provides per capita intake rates for exposed/protected fruits
and vegetables. Tables 3-7 and 3-7a present per capita intake rates in units of g/kg/day for the major
food groups. These tables provide the same information as in Table 3-3 and 3-3a, except that the
data are sorted in a slightly different way (i.e., these tables allow for easy comparisons between age
groups). Tables 3-8 and 3-8a present consumer-only intake rates for the major food groups in
g/kg/day and g/day, respectively. It should be noted that the terms consumer-only intake and per
capita intake are used in the same way as in the CSEFH. Consumer-only intake is defined as the
quantity of foods consumed only by children who ate these food items during the survey period. Per
capita intake rates are generated by averaging consumer-only intakes over the entire population of
children (i.e., both users and non-users). Also, the food ingestion rates are expressed as "as
consumed," since this is the fashion in which data are reported by survey respondents.

Tables 3-9 through 3-14 present the contributions of the major food groups to total dietary
intake for individuals (in the various age groups) at the low end, central, and high end of the
distribution of total dietary intake, total meat intake, total meat and dairy intake, total fish intake,
total fruit and vegetable intake, and total dairy intake in units of g/day and g/kg/day.

3.4.2 Uncertainties

As noted in the CSEFH, these exposure data have been generated from short-term data.
Although the data are suitable for estimating mean average daily intake rates representative of both
short-term and long-term consumption, the distribution of average daily intake rates generated using
short-term data (e.g., 2-day) does not necessarily reflect the long-term distribution of average daily
intake rates. Also, day-to-day variation in intake among individuals will be great for food
item/groups that are highly seasonal and for items/groups that are eaten year-round but are not
typically eaten every day (U.S. EPA, 2001). For these foods, the intake distribution generated from
short-term data will not be a good reflection of the long-term distribution. On the other hand, for
broad categories of foods (e.g., vegetables) that are eaten on a daily basis throughout the year with
minimal seasonality, the short-term distribution maybe a reasonable approximation of the true long-
term distribution, although it will show somewhat more variability. It is also important to note that
the sample sizes are small for some of the age groups used in the reanalysis (see Table 3-2). This
is particularly true for children less than 1 year of age. Table 3-15 provides the confidence ratings
for the results of this reanalysis.

3-6


-------
3.5

RECOMMENDATION FOR FURTHER ANALYSIS AND RESEARCH NEEDS

As noted in Section 3.2, the 1998 CSFII data set provides additional data for children and
would be a useful supplement to the 1994-96 CSFII data set in evaluating food intake for children.
A similar analysis to that conducted here, using the 1998 CSFII data set, would increase the number
of observations on which the child intake rates are based and improve the confidence level for intake
rates generated by such an analysis. Table 3-16 illustrates the increase in the number of children
surveyed in 1998 compared to 1994-96.

3.6 REFERENCES

Pao, E.M.; Fleming, K.H.; Guenther, P.M.; Mickle, S.J. (1982) Foods commonly eaten by
individuals: amount per day and per eating occasion. U.S. Department of Agriculture. Home
Economics Report No. 44.

SAS Institute, Inc. (1990) SAS procedures guide, version 6, 3d ed., Cary, NC: SAS Institute, Inc.

USDA (1995) Food and nutrient intakes by individuals in the United States, 1 day, 1989-91. U.S.
Department of Agriculture, Agricultural Research Service, Beltsville, MD. NFS Report No.
91-2.

USDA (1998) 1994-96 Continuing survey of food intakes by individuals (CSFII) and 1994-96 Diet
and health knowledge survey (DKHS). CD-ROM. U.S. Department of Agriculture,
Agricultural Research Service, Beltsville, MD. Available from the National Technical
Information Service, Springfield, VA.

USDA (2000) 1998 Continuing survey of food intakes by individuals (CSFII) and 1994-96 Diet and
health knowledge survey (DKHS). CD-ROM. U.S. Department of Agriculture, Agricultural
Research Service, Beltsville, MD. Available from the National Technical Information
Service, Springfield, VA.

U.S. EPA (2001) Child-specific exposure factors handbook (external review draft). Prepared by
Versar, Inc., for the Office of Research and Development, National Center for Environmental
Assessment, under EPA contract no. 68-W-99-041.

3-7


-------
Table 3-3. Per Capita Intake of the Major Food Groups (g/kg/day, as consumed)


-------
Table 3-3. Per Capita Intake of the Major Food Groups (g/kg/day, as consumed)


-------
Table 3-3a. Per Canita Ttitake of the Maior Food Oroiins fg/dav. as consumed^


-------
Table 3-3a. Per Canita Ttitake of the Maior Food Oroiins fg/dav. as consumed^


-------


Table 3-4. Per Canita Tnta

ce of Individual Foods fWktr/Hav. as

rnmumpfh










































































-------


Table 3-4. Per Canita Tnta

ce of Individual Foods fWktr/Hav. as

rnmumpfh












































-------
Table 3-5. Per Capita Intake of USDA Categories of Vegetables and Fruits (g/kg/dav, as consumed)


-------
Table 3-5. Per Capita Intake of USDA Categories of Vegetables and Fruits (g/kg/dav, as consumed)


-------
Table 3-6. Per Capita Intake of Exposed/Protected Fruit and Vegetable Categories (g/kg/day, as consumed)


-------
Table 3-6. Per Capita Intake of Exposed/Protected Fruit and Vegetable Categories (g/kg/day, as consumed)


-------
Table 3-7. Per Capita Intake of Major Food Groups (g/kg/dav, as consumed)


-------
Table 3-7. Per Capita Intake of Major Food Groups (g/kg/dav, as consumed)


-------
r.UI^ 1


-------
r.UI^ 1


-------








Table 3-8

Consumer Intake of Major Food Groups (g/kg/day, as consumed)











Food Group

PC

MEAN

SE

PI

P5

P10

P25

P50

P75

P90

P95

P99

P100

N

Nwgt













Age

<1 month

















Total Dietary Intake

100%

*

*

*

*

*

*

*

*

*

*

*

*

9

111979

Total Dairy Intake

100%

*

*

*

*

*

*

*

*

*

*

*

*

9

111979

Total Meat Intake

0%

*

*

*

*

*

*

*

*

*

*

*

*

0

0

Total Egg Intake

0%

*

*

*

*

*

*

*

*

*

*

*

*

0

0

Total Fish Intake

0%

*

*

*

*

*

*

*

*

*

*

*

*

0

0

Total Grain Intake

100%

*

*

*

*

*

*

*

*

*

*

*

*

1

6716

Total Vegetable Intake

0%

*

*

*

*

*

*

*

*

*

*

*

*

0

0

Total Fruit Intake

0%

*

*

*

*

*

*

*

*

*

*

*

*

0

0

Total Fat Intake

0%

*

*

*

*

*

*

*

*

*

*

*

*

0

0













Age

1-2 months

















Total Dietary Intake

100%

1.7e+02

l.le+01

6.9e+00

6.9e+00

2.4e+01

1.6e+02

1.9e+02

2.2e+02

2.4e+02

2.7e+02

3.1e+02

3.3e+02

46

513246

Total Dairy Intake

100%

1.9e+02

8.4e+00

4.8e+00

6.0e+01

1.4e+02

1.6e+02

1.9e+02

2.3e+02

2.5e+02

2.8e+02

3.1e+02

3.3e+02

45

470798

Total Meat Intake

0%

*

*

*

*

*

*

*

*

*

*

*

*

0

0

Total Egg Intake

0%

*

*

*

*

*

*

*

*

*

*

*

*

0

0

Total Fish Intake

0%

*

*

*

*

*

*

*

*

*

*

*

*

0

0

Total Grain Intake

100%

*

*

*

*

*

*

*

*

*

*

*

*

9

92996

Total Vegetable Intake

100%

*

*

*

*

*

*

*

*

*

*

*

*

1

5728

Total Fruit Intake

100%

*

*

*

*

*

*

*

*

*

*

*

*

5

79948

Total Fat Intake

0%

*

*

*

*

*

*

*

*

*

*

*

*

0

0













Age

3-5 months

















Total Dietary Intake

100%

1.4e+02

6.8e+00

8.1e-01

5.7e+00

2.2e+01

9.5e+01

1.4e+02

1.8e+02

2.3e+02

2.4e+02

2.9e+02

2.9e+02

109

1020261

Total Dairy Intake

100%

1.2e+02

6.6e+00

4.1e-01

2.5e+00

1.8e+01

8.1e+01

1.3e+02

1.7e+02

2.0e+02

2.3e+02

2.8e+02

2.8e+02

100

970465

Total Meat Intake

100%

*

*

*

*

*

*

*

*

*

*

*

*

12

111057

Total Egg Intake

100%

*

*

*

*

*

*

*

*

*

*

*

*

11

110853

Total Fish Intake

100%

*

*

*

*

*

*

*

*

*

*

*

*

10

92820

Total Grain Intake

100%

2.4e+00

3.6e-01

1.2e-01

2.3e-01

3.2e-01

7.4e-01

1.5e+00

2.8e+00

5.6e+00

6.2e+00

2.7e+01

2.7e+01

77

759909

Total Vegetable Intake

100%

1.0e+01

l.le+00

6.1e-01

1.4e+00

1.4e+00

3.9e+00

9.6e+00

1.7e+01

1.9e+01

2.5e+01

3.0e+01

3.1e+01

41

440422

Total Fruit Intake

100%

2.3e+01

2.3e+00

4.6e-01

2.4e+00

3.4e+00

7.6e+00

2.0e+01

3.7e+01

4.1e+01

6.3e+01

l.le+02

l.le+02

65

641787

Total Fat Intake

100%

*

*

*

*

*

*

*

*

*

*

*

*

11

111807













Age

>11 months

















Total Dietary Intake

100%

1.3e+02

4.1e+00

1.6e+01

2.4e+01

6.0e+01

1.0e+02

1.3e+02

1.7e+02

1.9e+02

2.0e+02

2.5e+02

3.1e+02

152

1720660

Total Dairy Intake

100%

8.8e+01

3.5e+00

4.9e-02

1.0e+01

3.1e+01

6.8e+01

8.7e+01

l.le+02

1.3e+02

1.7e+02

1.9e+02

2.4e+02

146

1683509

Total Meat Intake

100%

3.0e+00

2.6e-01

8.7e-02

2.7e-01

5.0e-01

9.4e-01

2.3e+00

4.1e+00

6.5e+00

8.6e+00

1.2e+01

1.2e+01

104

1329670

Total Egg Intake

100%

1.3e+00

2.4e-01

4.0e-03

1.0e-02

4.3e-02

7.0e-02

1.2e-01

1.0e+00

5.0e+00

8.3e+00

8.3e+00

l.le+01

93

1177183

Total Fish Intake

100%

4.7e-01

9.4e-02

8.0e-03

3.0e-02

3.9e-02

1.4e-01

2.9e-01

4.6e-01

8.7e-01

1.6e+00

4.7e+00

4.7e+00

65

812655

Total Grain Intake

100%

8.1e+00

6.2e-01

1.9e-01

9.6e-01

1.4e+00

2.9e+00

5.5e+00

1.2e+01

2.1e+01

2.4e+01

3.3e+01

4.0e+01

146

1684880

Total Vegetable Intake

100%

1.3e+01

8.9e-01

6.1e-01

1.7e+00

3.1e+00

7.2e+00

1.2e+01

1.6e+01

2.4e+01

2.9e+01

4.9e+01

1.0e+02

138

1617520

Total Fruit Intake

100%

2.2e+01

l.le+00

2.3e+00

5.2e+00

7.4e+00

1.5e+01

2.0e+01

2.8e+01

4.0e+01

5.1e+01

6.7e+01

7.1e+01

134

1538726

Total Fat Intake

100%

2.5e-01

2.4e-02

9.0e-03

2.0e-02

5.1e-02

1.4e-01

2.0e-01

3.2e-01

4.3e-01

5.5e-01

1.2e+00

1.7e+00

94

1178873













Ag

e 1-2 years

















Total Dietary Intake

100%

8.7e+01

9.8e-01

2.5e+01

3.7e+01

4.5e+01

6.2e+01

8.3e+01

l.le+02

1.3e+02

1.6e+02

1.9e+02

2.6e+02

1302

7943041

Total Dairy Intake

100%

3.9e+01

7.6e-01

5.1e-01

5.8e+00

1.0e+01

1.9e+01

3.3e+01

5.3e+01

7.5e+01

9.2e+01

1.3e+02

1.8e+02

1298

7918065

Total Meat Intake

100%

4.7e+00

9.1e-02

1.9e-01

8.2e-01

1.2e+00

2.2e+00

4.1e+00

6.4e+00

8.9e+00

1.0e+01

1.5e+01

2.4e+01

1274

7769335

Total Egg Intake

100%

1.3e+00

5.7e-02

7.0e-03

2.4e-02

3.5e-02

8.3e-02

2.2e-01

2.1e+00

4.1e+00

5.3e+00

8.8e+00

1.4e+01

1204

7325394

Total Fish Intake

100%

6.4e-01

4.6e-02

9.0e-03

3.4e-02

6.1e-02

1.2e-01

2.3e-01

5.0e-01

1.5e+00

2.8e+00

6.3e+00

1.4e+01

789

4790345

Total Grain Intake

100%

1.2e+01

2.0e-01

1.9e+00

3.4e+00

4.6e+00

7.0e+00

1.0e+01

1.5e+01

2.1e+01

2.5e+01

3.5e+01

4.8e+01

1297

7912242

Total Vegetable Intake

100%

1.0e+01

2.1e-01

8.0e-01

1.7e+00

3.0e+00

5.1e+00

8.2e+00

1.3e+01

1.9e+01

2.4e+01

3.3e+01

8.3e+01

1293

7885022


-------








Table 3-8

Consumer Intake of Major Food Groups (g/kg/day, as consumed)











Food Group

PC

MEAN

SE

PI

P5

P10

P25

P50

P75

P90

P95

P99

P100

N

Nwgt

Total Fruit Intake

100%

2.2e+01

5.0e-01

1.3e+00

3.5e+00

5.5e+00

1.0e+01

1.8e+01

3.0e+01

4.4e+01

5.8e+01

8.3e+01

1.3e+02

1160

7112706

Total Fat Intake

100%

4.6e-01

1.2e-02

1.9e-02

6.1e-02

8.9e-02

1.8e-01

3.4e-01

5.9e-01

9.7e-01

1.3e+00

2.3e+00

3.3e+00

1222

7481895













Ag

3-5 years

















Total Dietary Intake

100%

5.9e+01

6.3e-01

2.0e+01

2.7e+01

3.2e+01

4.1e+01

5.6e+01

7.2e+01

9.0e+01

1.0e+02

1.3e+02

1.9e+02

1337

11598922

Total Dairy Intake

100%

2.2e+01

3.9e-01

1.5e+00

4.5e+00

6.7e+00

1.2e+01

2.0e+01

3.0e+01

4.2e+01

4.9e+01

6.8e+01

9.0e+01

1333

11551343

Total Meat Intake

100%

4.5e+00

7.6e-02

4.1e-01

9.7e-01

1.5e+00

2.5e+00

4.0e+00

5.7e+00

8.0e+00

9.7e+00

1.3e+01

2.1e+01

1323

11479245

Total Egg Intake

100%

7.7e-01

3.9e-02

5.0e-03

1.6e-02

2.7e-02

5.6e-02

1.2e-01

l.le+00

2.5e+00

3.6e+00

6.1e+00

1.3e+01

1212

10495709

Total Fish Intake

100%

5.5e-01

3.7e-02

1.2e-02

2.8e-02

5.7e-02

l.le-01

2.0e-01

3.8e-01

1.5e+00

2.4e+00

5.7e+00

9.6e+00

810

7127866

Total Grain Intake

100%

l.le+01

1.9e-01

2.5e+00

3.9e+00

5.0e+00

6.9e+00

9.6e+00

1.3e+01

1.8e+01

2.1e+01

3.4e+01

1.2e+02

1336

11594549

Total Vegetable Intake

100%

7.8e+00

1.6e-01

4.9e-01

1.5e+00

2.1e+00

4.1e+00

6.6e+00

1.0e+01

1.4e+01

1.9e+01

2.9e+01

4.6e+01

1330

11552278

Total Fruit Intake

100%

1.4e+01

3.3e-01

8.9e-01

2.2e+00

3.4e+00

6.0e+00

l.le+01

1.9e+01

2.8e+01

3.5e+01

5.6e+01

l.le+02

1134

9812013

Total Fat Intake

100%

4.7e-01

1.2e-02

1.6e-02

5.4e-02

8.3e-02

1.7e-01

3.4e-01

6.2e-01

9.9e-01

1.3e+00

1.9e+00

3.1e+00

1280

11089608













Age

6-10 years

















Total Dietary Intake

100%

4.0e+01

5.2e-01

1.0e+01

1.6e+01

2.0e+01

2.8e+01

3.8e+01

4.9e+01

6.2e+01

7.2e+01

9.3e+01

1.2e+02

1105

18263131

Total Dairy Intake

100%

1.6e+01

3 .Oe-O1

4.7e-01

2.5e+00

4.3e+00

8.3e+00

1.4e+01

2.1e+01

2.9e+01

3.6e+01

4.7e+01

8.1e+01

1103

18235617

Total Meat Intake

100%

3.2e+00

6.6e-02

2.0e-01

6.8e-01

9.7e-01

1.7e+00

2.8e+00

4.4e+00

5.8e+00

7.3e+00

l.le+01

1.8e+01

1090

18042272

Total Egg Intake

100%

4.9e-01

3.0e-02

7.0e-03

1.6e-02

2.1e-02

3.9e-02

8.5e-02

4.2e-01

1.7e+00

2.5e+00

4.6e+00

9.3e+00

1008

16689310

Total Fish Intake

100%

4.6e-01

3.6e-02

1.3e-02

3.2e-02

5.1e-02

8.5e-02

1.5e-01

3.2e-01

1.2e+00

2.1e+00

4.8e+00

6.7e+00

682

11363938

Total Grain Intake

100%

8.0e+00

1.3e-01

1.3e+00

2.7e+00

3.5e+00

5.0e+00

7.3e+00

1.0e+01

1.4e+01

1.6e+01

2.0e+01

3.6e+01

1104

18252467

Total Vegetable Intake

100%

5.8e+00

1.3e-01

4.7e-01

1.3e+00

1.9e+00

2.9e+00

4.8e+00

7.5e+00

l.le+01

1.4e+01

2.1e+01

5.2e+01

1102

18222862

Total Fruit Intake

100%

7.9e+00

2.3e-01

4.4e-01

1.4e+00

2.0e+00

3.0e+00

6.0e+00

l.le+01

1.7e+01

2.1e+01

3.1e+01

4.5e+01

840

14000258

Total Fat Intake

100%

3.9e-01

l.le-02

2.0e-02

4.3e-02

6.4e-02

1.4e-01

2.7e-01

5.2e-01

8.8e-01

l.le+00

1.6e+00

3.1e+00

1069

17708968













Age

11-15 years

















Total Dietary Intake

100%

2.4e+01

3.8e-01

5.2e+00

8.1e+00

l.le+01

1.6e+01

2.2e+01

3.1e+01

3.9e+01

4.7e+01

6.0e+01

8.1e+01

975

18818601

Total Dairy Intake

100%

7.9e+00

2. le-01

1.5e-01

4.2e-01

9.3e-01

3.2e+00

6.7e+00

l.le+01

1.6e+01

2.1e+01

3.2e+01

3.8e+01

966

18682741

Total Meat Intake

100%

2.4e+00

4.9e-02

1.4e-01

4.2e-01

6.9e-01

1.3e+00

2.1e+00

3.0e+00

4.3e+00

5.4e+00

8.1e+00

l.le+01

970

18725557

Total Egg Intake

100%

3.3e-01

2.1e-02

4.0e-03

9.0e-03

1.6e-02

3.1e-02

6.3e-02

3.7e-01

l.le+00

1.4e+00

3.0e+00

7.3e+00

900

17432491

Total Fish Intake

100%

3.5e-01

2.6e-02

5.0e-03

1.6e-02

2.7e-02

6.3e-02

1.3e-01

2.6e-01

9.3e-01

1.6e+00

3.2e+00

5.9e+00

612

12076053

Total Grain Intake

100%

5.1e+00

9.5e-02

9.3e-01

1.5e+00

2.1e+00

3.0e+00

4.4e+00

6.6e+00

8.8e+00

l.le+01

1.5e+01

2.1e+01

975

18818601

Total Vegetable Intake

100%

4.4e+00

9.7e-02

3.8e-01

9.2e-01

1.4e+00

2.3e+00

3.7e+00

5.6e+00

7.9e+00

9.8e+00

1.5e+01

3.6e+01

973

18777586

Total Fruit Intake

100%

4.9e+00

1.7e-01

6.6e-02

6.0e-01

1.0e+00

1.8e+00

3.4e+00

6.5e+00

l.le+01

1.4e+01

1.8e+01

3.2e+01

681

13550269

Total Fat Intake

100%

2.8e-01

9.0e-03

1.2e-02

2.9e-02

5.2e-02

9.6e-02

1.9e-01

3.6e-01

6.6e-01

8.3e-01

1.4e+00

1.8e+00

937

18231526













Age

16-17 years

















Total Dietary Intake

100%

1.8e+01

5. le-01

4.2e+00

6.2e+00

7.6e+00

l.le+01

1.6e+01

2.2e+01

2.9e+01

3.4e+01

5.8e+01

6.4e+01

360

7718155

Total Dairy Intake

100%

5.4e+00

2.5e-01

8.8e-02

2.6e-01

4.6e-01

1.8e+00

4.3e+00

7.7e+00

1.2e+01

1.3e+01

2.0e+01

3.3e+01

357

7644914

Total Meat Intake

100%

1.9e+00

6.2e-02

1.6e-01

3.6e-01

5.2e-01

l.le+00

1.7e+00

2.5e+00

3.6e+00

4.0e+00

5.5e+00

7.0e+00

357

7640988

Total Egg Intake

100%

2.5e-01

2.4e-02

5.0e-03

l.le-02

1.7e-02

2.9e-02

5.2e-02

1.6e-01

9. le-01

1.3e+00

2.1e+00

2.5e+00

337

7212757

Total Fish Intake

100%

2.8e-01

3.8e-02

1.0e-02

1.2e-02

2.5e-02

5.4e-02

l.le-01

2.5e-01

4.9e-01

1.4e+00

3.0e+00

4.9e+00

225

4843926

Total Grain Intake

100%

3.9e+00

1.4e-01

3.5e-01

1.2e+00

1.4e+00

2.1e+00

3.4e+00

5.0e+00

6.6e+00

8.6e+00

1.4e+01

2.1e+01

360

7718155

Total Vegetable Intake

100%

3.8e+00

1.7e-01

2.1e-01

6.4e-01

9.9e-01

1.8e+00

3.2e+00

4.7e+00

7.3e+00

9.6e+00

1.5e+01

2.5e+01

357

7674724

Total Fruit Intake

100%

4.2e+00

2.5e-01

1.2e-01

5.6e-01

8.7e-01

1.5e+00

3.0e+00

5.6e+00

9.2e+00

l.le+01

1.5e+01

2.4e+01

204

4231840

Total Fat Intake

100%

2.4e-01

1.3e-02

l.le-02

2.4e-02

4.2e-02

8.5e-02

1.7e-01

3.1e-01

5.4e-01

7.3e-01

1.2e+00

1.6e+00

354

7576804

NOTES:	N = Number of Individuals Surveyed and Consuming in the 2-Day Survey Period

PC = Percent Consuming	Nwgt = Number of Observations Weighted to the US Population

SE = Standard Error	* = Data not provided for less than 20 observations

Source: Based on analysis of 1994-1996 CSFII.


-------
Table 3-8a. Consumer Intake of Major Food Groups (g/day, as consumed)


-------
Table 3-8a. Consumer Intake of Major Food Groups (g/day, as consumed)


-------

-------

-------
Table 3-9. Intake of Total Foods and Major Food Groups, and Percent of Total Food Intake for Individuals with Low-end, Mid-range, and High-end Total Food Intake






























-------
Table 3-9. Intake of Total Foods and Major Food Groups, and Percent of Total Food Intake for Individuals with Low-end, Mid-range, and High-end Total Food Intake






















-------
Table 3-10. Intake of Total Foods and Major Food Groups, and Percent of Total Food Intake for Individuals with Low-end, Mid-range, and High-end Total Intake


























-------
Table 3-10. Intake of Total Foods and Major Food Groups, and Percent of Total Food Intake for Individuals with Low-end, Mid-range, and High-end Total Intake






















-------
Table 3-11. Intake of Total Foods and Major Food Groups, and Percent of Total Food Intake for Individuals with Low-end, Mid-range, and High-end Total Meat and Dairy Intake






























-------
Table 3-11. Intake of Total Foods and Major Food Groups, and Percent of Total Food Intake for Individuals with Low-end, Mid-range, and High-end Total Meat and Dairy Intake






























-------
Table 3-12. Intake of Total Foods and Major Food Groups, and Percent of Total Food Intake for Individuals with Low-end, Mid-range, and High-end Total Fish Intake






























-------
Table 3-12. Intake of Total Foods and Major Food Groups, and Percent of Total Food Intake for Individuals with Low-end, Mid-range, and High-end Total Fish Intake


























-------
Table 3-13. Per Capita Intake of Total Foods and Major Food Groups, and Percent of Total Food Intake for Individuals with Low-end, Mid-range, and High-end Total Fruit & Vegetable Intake






























-------
Table 3-13. Per Capita Intake of Total Foods and Major Food Groups, and Percent of Total Food Intake for Individuals with Low-end, Mid-range, and High-end Total Fruit & Vegetable Intake






























-------
Table 3-14. Per Capita Intake of Total Foods and Major Food Groups, and Percent of Total Food Intake for Individuals with Low-end, Mid-range, and High-end Total Dairy Intake


































-------
Table 3-14. Per Capita Intake of Total Foods and Major Food Groups, and Percent of Total Food Intake for Individuals with Low-end, Mid-range, and High-end Total Dairy Intake


























-------
Table 3-15. Confidence in Recommendations for Food Intake


-------
Table 3-16. Number of Children Providing Intake Data in CSFII 1994-96 and CSFII1998

3 of Age

1994-96

1998

Total

<1

376

1,175

1,551

1

711

373

1,084

2

705

402

1,107

3

492

1,344

1,836

4

511

1,348

1,859

5

475

409

884

6

256

343

599

7

233

71

304

8

236

53

289

9

258

41

299

0-9	4,253	5,559	9,812

3-41


-------
4.0

DRINKING WATER AND TOTAL FLUIDS

4.1	INTRODUCTION

The legislative mandates found in the Safe Drinking Water Act (SDWA) Amendments of
1996 require EPA to gather up-to-date information on water ingestion and to identify subpopulations
that may be at elevated risk of health effects from exposure to contaminants in drinking water. To
fulfill its SDWA charge, EPA (2000) used current estimates of per capita water ingestion found in
the dietary and demographic data from the combined 1994, 1995, and 1996 Continuing Survey of
Food Intakes by Individuals (CSFII), conducted by the U.S. Department of Agriculture (USDA,
1998).

In its report to Congress, EPA noted that considerable progress has been made in the
development of improved methods for evaluating toxicity of drinking water contaminants, assessing
exposures of populations of special concern (lactating mothers, infants and children, the elderly, and
individuals whose health status has been compromised), and conducting risk assessments. One
critical issue in preparing exposure assessments for these subgroups is how to better consider age-
related changes in levels of exposure in a consistent and scientifically sound manner, especially with
regard to children. Exposure routes such as breast milk, food consumption, drinking water and total
fluid intake, ingestion of soil and other non-dietary materials, pica, and inhalation, and factors such
as body surface area, dermal soil adherence, body weight, and activity patterns, all have been
identified as important variables in these assessments. The purpose of this chapter is to describe key
published studies that provide information on drinking water consumption among children and to
provide recommendations of consumption rates that should be used in exposure assessments for the
EPA proposed age bins (<1 month, 1-2 months, 3-5 months, 6-11 months, 1-2 years, 3-5 years, 6-10
years, 11-15 years, and 16-17 years).

4.2	EVALUATION OF EXISTING DATA

Currently, EPA uses the quantity o f 1 L/day as a default drinking water intake rate for infants
(10 kg body mass or less) and children (U.S. EPA, 1980, 1991). This rate includes drinking water
consumed in the form of juices and other beverages containing tap water. The National Academy
of Sciences estimated that daily consumption of water may vary with levels of physical activity and
fluctuations in temperature and humidity (NAS, 1977). It is reasonable to assume that some
individuals in physically demanding occupations or living in warmer regions may have high levels
of water intake.

4-1


-------
Two studies, the USDA 1994-96 Continuing Survey of Food Intakes by Individuals (1998)
and Estimated Per Capita Water Ingestion in the United States (U.S. EPA, 2000), have generated
the most recent data on drinking water intake rates. These studies were summarized in the CSEFH.
These two studies reported intake rates for both direct and indirect ingestion of water, and in general,
support EPA's use of 1 L/day as an upper percentile rate for children under 10 years of age. Direct
intake is defined as direct consumption of water as a beverage, while indirect intake includes water
added during food preparation but not water intrinsic to purchased foods. Data for consumption of
various sources (i.e., the community water supply, bottled water, and other sources) are also
presented. EPA (2000) assumed that bottled water, and other purchased foods and beverages, are
widely distributed and therefore less likely to contain source-specific water. As a result, the use of
total water intake to estimate exposure may overestimate the potential exposure to toxic substances
present only in local water supplies; therefore, intake of tap water (community water), rather than
total water intake, was emphasized in the two studies.

The USDA study was carried out over 2 nonconsecutive days. Estimates of per capita
ingestion of plain drinking water (direct ingestion) and water ingested indirectly were derived from
responses to questionnaires provided by the USDA. W ater used in the final preparation of foods and
beverages at home or by food service establishments such as school cafeterias or restaurants was
defined as indirect water. Quantities of ingested water were averaged by participant to generate a
2-day average. These daily averages comprise the empirical distributions from which mean and
upper percentile per capita ingestion rates were developed from 2-day averages for more than 15,000
individuals in 50 States and the District of Columbia, which were then extrapolated to the
population of the entire United States.

Using the CSFII data, EPA estimated the per capita drinking water ingestion rate for
subpopulations segregated by (1) gender and age, and (2) by pregnant, lactating, and childbearing-
age women. These estimates are found in the EPA report Estimated Per Capita Water Ingestion in
the United States (U.S. EPA, 2000). EPA noted that the CSFII does not support estimates of water
intake levels for subpopulations with traditional lifestyles (Native Americans and recent immigrants),
those who live in hot climates, those with health conditions that affect water ingestion, and those
who consume large amounts of water because of physical activity.

In general, EPA estimated that the 90th percentile of the empirical distribution of 2-day
average per capita ingestion of community water was 2.016 L/person/day, approximately equal to
the 2 L/person/day estimate used as a standard ingestion value by many Federal agencies. EPA also
estimated that the mean water ingestion for women of childbearing age (15 to 44 years) was similar
to the general population and that lactating women had the highest community water ingestion rate

4-2


-------
of any subpopulation (2.872 L/person/day 90th percentile; 3.434 L/person/day at the 95th percentile),
For infants less than 1 year of age, the estimated mean community water ingestion rate was 878
mL/person/day (90th percentile) and 1,040 mL/person/day (95th percentile). For children 1 to 10
years of age, the mean community water ingestion was estimated to be 400 mL/person/day (90th
percentile) and 905 mL/person/day (95th percentile), consistent with the standard 1 -liter ingestion rate
used in risk assessments for a 10 kg child.

4.3 STUDIES SELECTED FOR ANALYSIS TO OBTAIN NEW RECOMMENDATIONS

The two studies discussed in the previous section were selected for the new analysis. In its
study Estimated Per Capita Water Ingestion in the United States (U.S. EPA, 2000), EPA reported
that community water (i.e., tap water from the public water supply) accounts for 75 percent of the
mean ingested water in the United States. Total water consumption consists of community water
supply, bottled water, and other sources, plus missing sources. "Other sources" include household
wells, cisterns, or a spring (either household or community). The data also distinguish between
direct and indirect water consumption. Table 4-1 provides the 1994-96 CSFII estimates, combined
for all ages, for the mean total direct and indirect water consumption, by water source, per person.
The estimates also include consumption rates for the 90th and 95th percentiles.

Table 4-1. Estimated Direct and Indirect Total Water Ingestion
by Source for U.S. Population (mL/person/day)a

Water Source

Sample
Size

Mean

90th Percentile

95th Percentile

Community Water

15,303

927

2,016

2,544

Bottled Water

15,303

161

591

1,036

Other Sources

15,303

128

343

1,007

Missing Sources

15,303

16

NAb

NA

All Sources

15,303

1,232

2,341

2,908

a Estimates are based on 2-day averages for nonconsecutive days.
b NA ~ Not available.

Source: USDA(1998).

Table 4-2 presents EPA's estimated total direct and indirect water ingestion rates and
percentiles for children, from all sources, by broad age groups (i.e., <1 year, 1-10 years, 11-19 years).

4-3


-------
Table 4-2. Estimate of Total Direct and Indirect Water Ingestion,
All Sources, by Broad Age Category for U.S. Children
(mL/person/day)a

Direct Water Ingestion

Age (Years)

Sample Size

Mean

90th Percentile

95th Percentile

< 1

359

484

949

1,182

1 - 10

3,980

528

1,001

1,242

11 - 19

1,641

907

1,780

2,185

Indirect Water Ingestion

< 1

359

67

156

170

1 - 10

3,980

25

49

64

11 - 19

1,641

16

30

39

a Estimates are based on 2-day averages for nonconsecutive days.

Sources: U.S. EPA (2000), USDA (1998).

EPA (2000a) used the drinking water ingestion data to derive estimates of water consumption
rates by age group, gender, water source, and potentially more highly exposed subgroups of the
population. To better present these data for use in risk assessments, ingestion rates were expressed
in both volume per person per day (mL/person/day) and volume per body weight per day
(mL/kg/day). As shown in Table 4-3, younger children consume more water in terms of volume per
body weight and therefore are a potentially more highly exposed subgroup of the general population.

The data in the CSFII study have both strengths and limitations. The strengths of the data set
lie in the design of the survey. First, it was intended to collect a statistically representative sample
of the U.S. population. Second, the survey was sufficiently specific in detailing types of food and
drink. The large sample size (>15,000 people, including 6,000 children) enhances the precision and
accuracy of the estimates. Furthermore, the survey was conducted over nonconsecutive days, which
improves the variance over consecutive days of consumption. In addition, the survey was
administered in such a manner that an interviewer went through the data collection process for the
initial day to show participants how to properly respond to the questionnaire. The second day of the
survey was reported by the participant alone. Finally, the survey uses parameters that enable
differentiation of water sources, ages, gender, weight, and potentially vulnerable populations and
represents the most up-to-date data on water consumption.

4-4


-------
Table 4-3. Estimate of Total Direct and Indirect Water Ingestion,
All Sources, by Fine Age Category for U.S. Children"

Age (Years)

Sample Size

Mean

90th Percentile

95th Percentile

mL/person-day

<0.5

199

280

861

945b

0.5-0.9

160

412

884

l,101b

1-3

1834

313

691

942

4-6

1,203

420

917

1,165

7-10

943

453

978

1,219

11-14

816

594

1,365

1,722

15-19

825

760

1,610

2,062

mL/kg/day

<0.5

191

47

139

170b

0.5-0.9

153

45

103

122b

1-3

1,752

23

51

67

4-6

1,113

21

44

64

7-10

879

15

32

39

11-14

790

12

26

34

15-19

816

12

25

32

a Estimates are based on 2-day averages for nonconsecutive days.

b Sample size was insufficient for minimum reporting requirements according to Third Report on

Nutritional Monitoring in the U.S. (1994-1996).

Sources: U.S. EPA (2000), USDA (1998).

The limitations of the CSFII include the short duration of the study, which diminishes the
precision of the individual water ingestion rate, and some of the data reporting methods. The large
sample population might have contributed to greater accuracy, but the precision could still be low.
The mean for an individual can easily be skewed. Also, the data reporting did not always support
variance estimation for some reported population subsets. As such, the differences in mean could
not always be statistically tested except for the larger population subsets. Therefore, the reported
differences were derived empirically instead of statistically. Also, no effort was made to verify
estimated water consumption rates. Available data on urinary volume rates (see Table 4-4), or body
composition data that include water values, could have been used. Most importantly, the CSFII data
set does not appear to contain data to gauge the water intake for the potentially more highly exposed
members of the subpopulation of children (those in the early age bins as proposed by EPA: <1, 1-2,
3-5, and 6-11 months).

4-5


-------
Table 4-4. Urinary Volume Rates
(mL/kg/day)

Age

Sample Size

Mean

Std. Dev.

Newborns - 1st day

9

8.5

3.5

Newborns 7th day (Breast Fed)

16

76

17

0-6 months

8

34

6

6-12 months

19

29

12

1 -2 years

14

25

7

2-3 years

6

33

9

3-4 years

8

34

7

4-5 years

10

29

10

5-7 years

8

25

7

7-11 years

12

25

7

11-14 years

8

19

3

Young males

11

20

3.2

Source: Lentner, C. (1981).

Basically, the younger the child, the more sensitive that child is to the adverse effects of any
contaminants that may be present in the drinking water (Fomon, 1967). Apart from the toxic
properties of the contaminants, there are substantial differences between the renal function of
children and adults that affect water intake and fluid balance in their bodies. During this period of
rapid growth, substantial changes in body composition occur and can appear contradictory to the
stable composition normally observed in adults (Fomon, 1967). For example, the mineral, protein,
water, and lipid contents of the body increase with age during early life, each at markedly different
rates (Figure 4-1).

The single molecule that constitutes the highest fraction of body mass is water. In healthy
adults, total body water (TBW) constitutes 60 percent of body weight for non-obese subjects. These
fractional contents, however, are not constant across the life span, nor are they invariant with
diseases. At full-term birth, a healthy infant's total body water decreases rapidly over the next 3 to
5 years until the hydration fraction reaches that observed for adults. The change in hydration reflects
a change in the ratio of water between the intracellular and extracellular compartments. In some
clinical conditions and with certain drugs, the body can retain or lose significant amounts of water.
In the healthy state, total body water tends to be well regulated, although a loss of only 15 percent,
such as in dehydration, can be significantly life threatening. Data have shown that even with

4-6


-------
significant changes in the extracellular:intracellular ratio, the normally fixed hydration constant
remains relatively firm. The kidney, the primary organ involved in maintaining water balance, is
physiologically vastly different in the fetus, newborn, and child up to at least 2 years of age
(Rhoades, 1992). The fetal kidney is not subj ect to the influence of antidiuretic hormone, which also
influences thirst in the adult1 and is only able to produce hypotonic urine. The fetal glomeruli are not
completely developed until the 35th week of gestation; however, in premature infants maturation
takes place immediately after delivery.

Figure 4-1. Changes in human body composition during fetal development and early life. These
data often serve as a reference standard for assessing growth in the preterm infant.

Source: Fomon (1967).

Once the body weight of the fetus has reached 0.2 to 0.25 kg, the formation of new glomeruli
stops. Given the number of glomeruli present at birth, the glomerular filtration rate is initially
inhibited due to the presence of the fetal cuboid epithelium. During the first 2 years of life, this
epithelium is slowly replaced by the final thin epithelial layer from the juxtamedullary to the cortical
region. At term, the proximal tubules of the fetus are still primitive. The length of the Loops of
Henle is important for the production of concentrated urine. At birth, the Loops of Henle are very
short compared with the adult. It is not until postnatal months 5 and 6 that the glomerular filtration

1 Thirst is a sensation aroused by a need for water. Although it is desirable that we restrict the word thirst to the
sensation aroused by a lack of water, in general usage it incorporates both an idea of appetite for water as well as a drive
toward relief of a need.

4-7


-------
rate in relation to body surface area approaches adult values; glomerular filtration rate and kidney
mass increase uniformly. After the first 6 months the other renal functions mature more slowly.
With normal development, children's tubular functions do not reach full maturity until they are at
least 2 years of age.

4.4	RECOMMENDATIONS FOR PROPOSED AGE BINS

A new analysis was conducted on the CSFII data to obtain recommended values for the
proposed age bins. Table 4-5 provides the recommended drinking water ingestion estimates for the
U.S. populations within the selected age bins using the data adapted from the CSFII data as presented
in the EPA water ingestion report (U.S. EPA, 2000). Note that the CSFII data set does not provide
enough data for the proposed age groups of children up to 1 year of age. Confidence ratings for the
recommended data to support recommendations presented in Table 4-5 are shown in Table 4-6.

4.5	RECOMMENDATIONS FOR FURTHER ANALYSIS AND RESEARCH NEEDS

Assessing exposure as volume of intake per unit body mass clearly indicates the greater
potential for exposure of the young child. Fetuses, newborns, and toddlers up to 2 years of age are
vulnerable as a result of their different renal function and fluid intake needs.

More research is needed to collect data on intake rates for drinking water within the
proposed age bins. It is recommended that EPA determine if responses to the questionnaires used
in the CSFII would allow researchers to derive drinking water ingestion data based on the proposed
age bins through reanalysis of the data. Also, if CSFII data are available on the very young, it should
be determined if the sample size is adequate to extrapolate this data to provide estimates of drinking
water ingestion rates for the general children's population.

According to McPherson et al. (2000), assessing the diets of children presents unique
methodological challenges. Validity and reliability studies of recall, records, food frequency
questionnaires (FFQs), diet histories, water intake, and other observations among children are
difficult. McPherson et al. evaluated the dietary assessment methods among school-aged children
for 47 studies published in peer-reviewed English journals between January 1970 and April 1999.
Each study for children 5-18 years of age had at least 30 samples. Most of the 24-hour recall
validation studies assessed only a portion of the day, not a 24-hour period, and had higher
agreements for meal versus complete day intake (McPherson et al., 2000). Food records
underestimated energy intake when compared with intake of doubly labeled water. Few studies
evaluated children's ability to complete records alone or to record an entire day. FFQs overestimated

4-8


-------
Table 4-5. Recommended Values for Direct, Indirect, and Both Direct and Indirect Water
Ingestion Excluding Commercial and Bottled Water

Direct Ingestion (mL/person/day)

Age Group

Sample Size

Population

Mean

95th Percentile

< 1 month

NAa



NA

NA

NA

1-2 months

NA



NA

NA

NA

3-5 months

NA



NA

NA

NA

6-11 months

160



1,768,152

96

b

1-2 years

1,834



12,262,345

184

677

3-5 years

1,203



12,531,561

274

880

6-10 years

943



15,351,948

317

1,030

11-15 years

816



15,578,741

414

1,531

16-17 years

825



17,988,744

531

2,618

Indirect Ingestion (mL/person/day)

< 1 month

NA



NA

NA

NA

1-2 months

NA



NA

NA

NA

3-5 months

NA



NA

NA

NA

6-11 months

160



1,768,152

316

b

1-2 years

1,834



12,262,345

129

432

3-5 years

1,203



12,531,561

145

458

6-10 years

943



15,351,948

136

482

11-15 years

816



15,578,741

180

629

16-17 years

825



17,988,744

531

2,618

Direct and Indirect Ingestion (mL/person/day)

< 1 month

NA



NA

NA

NA

1-2 months

NA



NA

NA

NA

3-5 months

NA



NA

NA

NA

6-11 months

160



1,768,152

412

b

1-2 years

1,834



12,262,345

313

942

3-5 years

1,203



12,531,561

420

1,165

6-10 years

943



15,351,948

453

1,219

11-15 years

816



15,578,741

594

1,722

16-17 years

825



17,988,744

760

2,062

a NA = Not available

b The sample size does not meet minimum reporting requirements, as described in the Third Report on Nutrition
Monitoring in the United States, 1994-96.

Source: Adapted from U.S. EPA (2000), Part I, Table A2.

4-9


-------
Table 4-6. Confidences in Recommendations for Drinking Water Ingestion

Rating (High, Medium, Low)
Age

Considerations

<1 Mo

1-2 Mos

3-5 Mos

6-11 Mos

1 -2 Yrs

3-5 Yrs

6-10 Yrs

11-15 Yrs

16-17 Yrs

Study Elements



















• Level of peer review

NA

NA

NA

NA

Low

Low

Low

Medium

Medium

• Accessibility

NA

NA

NA

NA

Low

Low

Low

Medium

Medium

• Reproducibility

NA

NA

NA

NA

Low

Low

Medium

Medium

High

• Focus on factor of interest

NA

NA

NA

NA

Low

Low

Low

Medium

High

• Data pertinent to U.S.

NA

NA

NA

NA

Low

High

High

High

High

• Primary data

NA

NA

NA

NA

Low

Low

Low

Medium

Medium

• Currency

NA

NA

NA

NA

Low

High

High

High

High

• Adequacy of data collection period

NA

NA

NA

NA

Low

Low

Low

Low

Low

• Validity of approach

NA

NA

NA

NA

Low

Low

Low

Medium

Medium

• Representativeness of the population

NA

NA

NA

NA

Low

Low

Medium

Medium

Medium

• Characterization of variability in the
population

NA

NA

NA

NA

Low

High

High

High

High

• Lack of bias in study design

NA

NA

NA

NA

Low

High

High

Medium

Medium

• Measurement error

NA

NA

NA

NA

Low

Medium

Medium

Medium

Medium

Overall Rating

NA

NA

NA

NA

Low

Low

Medium

Medium

Medium

NA = No data available or data are not suitable for use in the age bin specified.

4-10


-------
energy intake; however, validation standards may have over- or underestimated intake or used
different referent periods. Results of reliability studies for FFQs and diet history showed higher
energy intake at first, compared with subsequent administrations. Limited data were available on age,
ethnicity, and gender effects. Correlations between the validation standard and dietary method were
generally higher for recalls and records than FFQs. It was difficult to generalize the validity and
reliability results of dietary assessment methods because of discrepancies in study design, referent
periods, and validation standards. Given the experience of McPherson et al. (2000), new research
should consider not only the matrix used to rate the confidence in recommended values presented
in Table 1-1 of the Child-Specific Exposure Factors Handbook, but also data validity and data
reliability. A new questionnaire should be developed that considers independent data validation, such
as use of physiological parameters to gauge water balance and fluid intake values.

It is a frequent practice in anesthesiology and critical care medicine to use estimates of human
body surface area (BSA) to reflect the body's metabolic functions, such as ventilation rate, fluid
requirements, and extracorporeal circulation. Fetuses, infants, and adults have distinct shape-weight
relationships. Examination of published human BSA data, although complex, yields a simple linear
relationship between BSA and weight in infants and children weighing between 3 and 30 kg.
Application of linear regression analysis to published data results in a formula relating BSA in
square centimeters and weight in grams.

BSA = 1321 +0.3433 * Wt

Drinking water consumption and total fluid intake should be normalized by BSA as well as by body
weight using linear regression analysis (Current, 1998).

4.6 REFERENCES

Current, J.D. (1998) A linear equation for estimating the body surface area in infants and children.
Internet Journal of Anesthesiology 1998, vol. 2N2: http://www.ispub.com
/journals/IJA/Vol2N2/bsa.htm. Published April 1, 1998; last updated April 1, 1998.

Fomon, S.J. (1967) Body composition of the male reference infant during the first year of life.
Pediatrics 40:863-870.

Lentner, C., ed. (1981) Geigy scientific tables. Vol. 1. Units of measurement; body fluids;
composition of the body; nutrition. 8th ed. rev. West Caldwell, NJ: Ciba-Geigy Corporation
(Novartis), Medical Education Division, pp. 53-100.

4-11


-------
McPherson, R.S.; Hoelscher, D.M.; Alexander, M., et al. (2000) Dietary assessment methods among
school-aged children: validity and reliability. Preventive Medicine 31 :S 11 -33. Published
electronically August 11, 2000; doi:10.1006/pmed.2000.0631. Available online at

http://www. idealibraiy. com.

National Academy of Sciences. (1977) Drinking water and health, vol. 1. National Academy of
Sciences, National Research Council. Washington, DC : National Academy Press. Available
at http://www.nap.edn/catalog/l 780.html.

Rhoades, R.; Pflanger, R. (1992) Regulation of fluid and electrolyte balance. In: Human
Physiology, 2d ed. New York: Sanders College Publishing, pp. 857-886.

USDA (1998) 1994-96 Continuing survey of food intakes by individuals (CSFII) and 1994-96 Diet
and health knowledge survey (DKHS). CD-ROM. U.S. Department of Agriculture,
Agricultural Research Service, Beltsville, MD. Available from the National Technical
Information Service, Springfield, VA.

U.S. EPA (1980) Water quality criteria documents. Federal Register 45(231):79318-79379.

U.S. EPA (1991) Final Rule. Federal Register 56(20):3256-3597.

U.S. EPA. (2000) Estimated per capita water ingestion in the United States. Office of Science and
Technology, Office of Water, Washington, DC. EPA/822/R-00/008. Available at

http://www.epa.gov/waterscience/drinking/percapita/text.pdf.

4-12


-------
5.0

SOIL INGESTION AND PICA

5.1 INTRODUCTION

Children's exposure to toxicants via ingestion of contaminated soil and/or dust is of potential
concern. Ideally, the toxic substance(s) ingested by a child can be quantified over time. Using the
exposure profile, exposure parameters (such as the average daily quantity of a toxin ingested over
a 2-year period, or the maximum quantity ingested in a 24-hour period) relevant for risk assessment
can be selected and computed. These parameters, linked with knowledge of absorption and the
ultimate toxic effect, can inform the assessment of risks due to ingestion of toxic substances.
Although absorption of toxicants and toxic effects have obvious importance, the following
discussion is limited to quantifying ingestion of toxic substances.

The purpose of this chapter is to determine if soil ingestion and pica recommendations can
be made from existing data for the following proposed age groups: <1 month, 1-2 months, 3-5
months, 6-11 months, 1-2 years, 3-5 years, 6-10 years, 11-15 years, and 16-17 years.

Two facts make quantifying ingestion of toxic substances in soil and/or dust difficult. First,
it has not been possible to directly quantify the amount of a toxic substance ingested. Instead,
several indirect strategies for constructing such estimates have been proposed, including a behavioral
strategy (Lepow et al., 1974; National Research Council, 1980; Day et al., 1975; Kimbrough et al.
1984); a mass-balance strategy (Binder et al., 1986; Clausing et al., 1987; Calabrese et al., 1989a,
b; Davis et al., 1990; van Wijnen et al., 1990; Thompson and Burmaster, 1991; Calabrese et al.,
1997a); an implicit estimation approach based on average blood lead population comparisons
between areas with different soil lead concentrations (de Silva, 1994); and a combined approach
using behavioral data to develop a model for soil ingestion with age, and then applying this model
to mass-balance soil ingestion estimates (Sedman, 1989; Sedman and Mahmood, 1994).

Each of the methods has its strengths and limitations. A key consideration in comparing the
various methods is the ability to experimentally validate the approach. Only one method, the
mass-balance method, has been experimentally validated and only among adults (Calabrese et
al.,1990, 1991a, 1997a). Since the validation offers an objective point of reference, we limit this
discussion to mass-balance studies of soil ingestion. In a mass-balance soil ingestion study, the
amount of soil ingested is estimated using a trace element that is contained in soil. By quantifying
the trace element excreted, after subtracting the amount of the element ingested from other sources
and assuming minimal bioavailability of the element, the amount of soil ingested is back-calculated.

5-1


-------
A second fact makes estimation of soil ingestion confusing, and hence difficult. The
confusion occurs because the time period relevant for soil ingestion exposure assessment is usually
different from the duration of soil ingestion studies; soil ingestion studies are limited in size; the day-
to-day variability as well as the uncertainty in daily soil ingestion estimates appear to be large; and
for a given subject, the distribution of soil ingestion over days is often highly skewed. As a result
of these factors, simple tabulations and summaries of study data do not directly provide appropriate
soil ingestion distribution estimates. These factors are addressed in more recent evaluations of soil
ingestion and are particularly relevant when evaluating age-specific soil ingestion.

Pica, defined as deliberate soil ingestion, is discussed at the end of each section in this
chapter. Quantifying pica in terms of the amount of soil ingested on a pica day is particularly
difficult because the behavior is episodic. Since soil ingestion was quantified on a limited number
of days, and deliberate ingestion was not self-reported, quantifying the amount of soil ingested on
a pica day is difficult. We address these difficulties in the context of the relevant literature.

5.2 EVALUATION OF EXISTING DATA

5.2.1 Soil Ingestion

In this review, we focus on four primary studies and numerous manuscripts developed to
estimate soil ingestion based on the data in those studies. The studies include the study in Helena,
Montana (Binder et al., 1986); the Amherst, Massachusetts, study (Calabrese et al., 1989a); the
Washington State study (Davis et al., 1990); and the Anaconda, Montana, study (Calabrese et al.,
1997a). Common features of these studies are the mass-balance methodology, their conduct in the
United States, and the use of the trace elements aluminum (Al), silicon (Si), and titanium (Ti).

We briefly review other mass-balance studies that have appeared in the peer reviewed
literature, along with what we consider to be significant limitations of the other studies. Other
reviews have been given by Calabrese et al. (1993a) and Calabrese and Stanek (1994,1998). Since
the focus of this report is on the relationship between age and soil ingestion, we also review reports
that directly discuss the relationship between soil ingestion and age.

Three other mass-balance studies have been reported in the literature in addition to the four
primary studies. One study was conducted on a limited number of subjects in an attempt to
document soil pica behavior (Calabrese etal., 1997b). Since the subjects in this study were selected
to be non-representative of the general population, we do not discuss this study further. The other
two studies were conducted in the summer with Dutch children (Clausing et al., 1987, and

5-2


-------
van Wijnen et al., 1990). In each study, Al, Ti, and acid-insoluble residue (AIR) was used to
quantify soil ingestion, with the estimate based on the minimum of the three estimates considered
to be the best estimate. The Clausing study included 18 children ages 2-4, and collected fecal
samples over a 5-day period. A concern in the study was the completeness of sample collection
(only 27 samples were collected), leading the investigators to inflate fecal dry weights to a constant
10 g/day. These factors, plus the estimation of trace element intake from food from an independent
hospital control group of subjects, are principal study limitations. Biases are possible both as aresult
of use of the limiting tracer method, and as a result of the hospital food controls.

The van Wijnen study included 292 children between the ages of 1 and 5, and used similar
methods as Clausing's study. Children were recruited from day-care centers (n=199), camps (n=78),
and hospitals (n=15). No attempt was made to collect complete fecal samples, with one sample
considered adequate for a child. To compensate for the missing fecal samples, all dry fecal weights
were assumed to be 15 g. As in the Clausing study, the fecal sampling protocol is a limitation.
Collection of a small fecal sample for one child, and a large fecal sample for a second child, given
that both children ingest the same amount of soil, will result in very different estimates. The choice
of 15 g dry weight (versus 10 g dry weight) will alter soil ingestion estimates by 50 percent. Other
concerns for the Clausing study are shared by the van Wijnen study. These limitations, along with
the possibility of different cultural practices between Dutch and U.S. child rearing that may affect
behavior and soil ingestion, resulted in exclusion of these studies from further consideration.

The relationship between age and soil ingestion is directly discussed in two publications.
Sedman (1989) discusses the relationship between blood lead levels and age, and uses this, plus the
Helena, Montana, data (Binder et al., 1986), to construct age specific soil ingestion estimates. Citing
results given by Annest and Mahaffey (1984) from the Second National Health and Nutrition
Examination Survey (1976-1980), average blood lead levels decline from 16.3 mg/dL for children
between the ages of 6 months and 3 years to 11.4 mg/dL for children ages 12-14 years. Noting the
decline in blood lead levels, Sedman expressed the blood lead for older age groups as a percentage
of the blood lead levels for children in the 2-year-old group, and then fit a simple exponential model
with age (x) to the percent of blood lead (y), giving the expression v = 1.0594 exp{-0.0305x}.
Similar models were fit to other blood lead data, and to mouthing prevalence data, resulting in
different age coefficients. Sedman assumed that some combination of the age coefficients would
be applicable to represent soil ingestion by age and chose an average of the blood lead and mouthing
coefficients, resulting in an age coefficient of-0.112. Assuming a soil ingestion rate of 590 mg/d
for children aged 1-3 years (obtained by taking the mean of five estimates, plus one standard
deviation) age-specific soil ingestion values were calculated. The five estimates used to construct
the soil ingestion rate came from two estimates using lead as a tracer in a study of 18 children by

5-3


-------
Ter Haar and Aronow (1974) (8 of whom were hospitalized for high levels of lead and pica like
behavior), while three of the estimates were based on the Helena, Montana, study (Binder et al.,
1986) using the tracers Al, Si, and Ti. Using the exponential age model with a slope coefficient of
-0.112, Sedman tabulated age specific soil ingestion values. In a subsequent publication, Sedman
and Mahmood (1994) used soil ingestion estimates from Davis et al. (1990) and Calabrese et al.
(1989a) studies, but retained the same basic exponential model and age slope to compute age specific
soil ingestion.

We consider the assumptions underlying Sedman's approach to be arbitrary. Although the
blood lead continues to decline with age based on the Third National Health and Nutrition
Examination Survey (NHANES III), 1991-94 (CDC, 1997), a cross-sectional decline in blood lead
may not imply a decline in soil ingestion. Many changes occur during this age range, including
weight and height gains and hormonal changes. Mouthing behavior is reported to decline with age,
but is not strongly related to soil ingestion (Davis et al., 1990). The limited rationale for the choice
of the age slope results in little confidence in the age-specific soil ingestion results.

A second publication that addresses age directly uses the Amherst, Massachusetts, data
(Staneket al., 1991). Assuming a linear relationship between age and soil ingestion, estimates were
given for linear regression slopes of soil ingestion for four trace elements (Al, Si, Ti, and zirconium
[Zr]) with age (in months) for 59 of the study children separately. The estimated slope was positive
with age (but substantially different) for each trace element, and statistically significant for Si.
Nevertheless, the proportion of the variance explained by age (the R-square) was small (14.5 percent
for Si and less than 6 percent for other trace elements). Further discussion of the relationship
between age and soil ingestion in these data was given by Calabrese and Stanek (1991, 1994) and
Calabrese et al. (1993a), who cautioned that the positive slope with age for Si and Al maybe due to
increased ingestion of toothpaste by older children, a factor not controlled in the Amherst study. In
light of these observations, Calabrese et al. (1993a) concludes that the Amherst study data are
insufficient to quantify an age effect. For this reason, we consider the results of Stanek et al. (1991)
to offer limited insight into the age relationship.

Additional discussion of age and soil ingestion was given in Calabrese et al. (1993a). The
authors postulated that children age 6-12 years ingest 25 percent of the soil ingested by a child age
1-6 years, while those greater than age 12ingest 10 percent ofthat ingested by a child age 1-6 years.
The authors noted that these estimates were not based on data. We do not consider this assertion to
have a sufficient foundation for their use as a basis for a recommendation.

5-4


-------
5.2.2 Soil Pica

For purposes of the discussion in this section, we defined pica as the deliberate ingestion of
soil. Although there is much literature on ingestion of non-food items (which is also referred to as
pica), there is limited information on soil pica. Most of this information is anecdotal or for special
populations (see Wong, 1986, and Calabrese and Stanek, 1993b). Some effort has been made to
quantify the frequency of soil pica in children using retrospective parental questionnaires (Stanek
et al., 1998b), with mass-balance study follow-up on children reported to display soil pica (Calabrese
et al., 1997b). Such efforts fall short of defining the amount of soil ingested in a pica event, but they
provide insights that may help guide the design of studies that can quantify soil pica.

Nevertheless, very high soil ingestion, presumed to be soil pica, was observed in one subject
in the Amherst study (Calabrese et al., 1991b). We discuss this subject in detail, in an effort to
highlight the difficulty in framing a recommendation concerning soil pica that has value for exposure
assessment.

5.3 STUDIES SELECTED FOR ANALYSIS TO OBTAIN NEW RECOMMENDATIONS
5.3.1 Soil Ingestion

Recommendations for soil ingestion are based on four mass-balance studies conducted in the
United States. The studies include the Helena, Montana, study (Binder et al. 1986); the Amherst,
Massachusetts, study (Calabrese et al., 1989a); the Washington State study (Davis et al., 1990); and
the Anaconda, Montana, study (Calabrese et al., 1997a). These studies were conducted on children
1-7 years of age. As a result, recommendations for soil ingestion are made only for the age groups
1-2 years, 3-5 years, and 6-10 years.

The four studies have been reviewed in many places; we describe them briefly here. The
studies share the advantage that they include multiple trace elements, attempt to collect complete
food (except for the Helena study) (Binder et al., 1986) and fecal samples over a defined study time
period, and follow similar protocols. All the studies use the trace elements Al, Si, and Ti. In all four
studies, estimates based on Ti are markedly different from other trace elements (for reasons not
related to soil ingestion) (Stanek and Calabrese, 2000, Table IV), and hence will not be presented.
It is important to note that these differences are considered the result of source error, that is,
ingestion of trace elements is a result of nonfood/nonsoil ingestion. Although there are important
differences between the studies, the common features support combining results to form a
recommendation. We summarize the studies in chronological order.

5-5


-------
The first mass-balance soil ingestion study used aluminum, silicon and titanium as tracers
to estimate soil ingestion for 59 1- to 3-year-old children in diapers, from East Helena, Montana, for
3 days in the summer of 1984 (Binder et al., 1986). Food was not collected, and soil ingestion
estimates were constructed by dividing the average daily trace element totals in fecal samples
(assuming 15 g dry fecal weight/day) by trace element concentrations in soil from pooled one-inch
deep samples in the children's yards. Adjustments were not made for food, toothpaste, and medicine
ingestion, nor was urine collected. Data were reanalyzed by Thompson and Burmaster (1991) using
actual dry stool weights (assuming complete fecal sample collection) yielding average soil ingestion
estimates about one half as large as the soil ingestion estimates reported by Binder et al. (1986). We
judge Thompson and Burmaster's estimates to be more appropriate in light of subsequent studies
results (Calabrese et al., 1989a, 1997a) that obtained similar dry fecal weights for a similar age group
of children. Estimates of average soil ingestion over three days given by Thompson and Burmaster
were 97 mg/d and 85 mg/d based on Al and Si, respectively.

Soil ingestion estimates provided by Thompson and Burmaster did not adjust for trace
element intake from food. Lacking adjustment for food intake of trace elements, estimates of
average soil ingestion are positively biased, and estimates of the distribution of soil ingestion are
artificially spread out. For this reason, we consider estimates from the study only for the mean and
median soil ingestion. Average daily trace element ingestion from food in other studies with similar
age children range from 1.87 to 6.3 mg/d for Al, and from 15.4 to 18 mg/d for Si (Calabrese et al.,
1989b (Table 10); Calabrese etal., 1997a (Table 3); Sedman, 1989). Using Binder's geometric mean
trace element concentrations in soil, the average trace element ingestion from food is equivalent to
between 28-95 mg soil/d for Al, and between 51-60 mg/d for Si. Applying these adjustments for
food, estimates of the average (over 3 days) soil ingestion for children ages 1-3 from Al and Si in
the Helena, Montana, study range from 2 to 69 mg/d.

The second mass-balance soil ingestion study used eight trace elements (including Al and
Si) to estimate soil ingestion for 64 children ages 1-3 from Amherst, Massachusetts, for 4 days in
each of 2 consecutive weeks in late September/early October 1986 (Calabrese et al., 1989a, b).
Subjects were recruited from day-care centers, and were eligible regardless of whether or not they
wore diapers. Duplicate food samples and medicines were collected corresponding to three of the
four fecal sample days (lagged by 12 hours), and used to estimate trace element intake from food
over 4 days. Measures were made daily. Trace element amounts from food were subtracted from
fecal totals, and divided by soil concentrations from composite 3-inch-deep home soil samples to
estimate soil ingestion. Adjustments were not made for toothpaste ingestion, nor was urine
collected. One child in the study ingested particularly large amounts of soil (-20 g) on 2 days in 1
week (Calabrese et al., 1991b, 1993c). Excluding this child, the study resulted in soil ingestion

5-6


-------
estimates corresponding to an average over 6-8 days. Soil ingestion estimates can be calculated on
48 children age 1-2 years and 15 children age 3 years.

The third mass-balance study used three trace elements (including A1 and Si) to estimate soil
ingestion for 101 children age 2-7 years from Washington State on 4 consecutive days in July and
August 1987 (Davis et al., 1990). All subjects were out of diapers. Subjects were recruited via
random digit dialing; 73 percent of households with eligible children participated. Duplicate food
samples, medicines, fecal samples, and some urine were collected on 4 consecutive days. Collected
quantities were pooled over days for each subject, and used to estimate trace element intake. These
amounts were subtracted from fecal and urine totals and divided by soil concentrations from
composite 1-inch-deep home soil samples to estimate average daily soil ingestion. The study
resulted in soil ingestion estimates corresponding to average (over 4 days) soil ingestion for 11
children age 2 years, 54 children age 3-5 years, and 36 children age 6-7 years. The impact of
toothpaste ingestion on soil ingestion estimates was discussed and estimated to reduce soil ingestion
estimates by 1.8 mg/d for Al, and 45.4 mg/d for Si.

The fourth mass-balance study used eight trace elements (including Al and Si) to estimate
soil ingestion for 64 children aged 1-3 from Anaconda, Montana, on seven consecutive days in
September/October 1992 (Calabrese et al., 1997a). Subjects were selected via a stratified simple
random sample of subjects in the area, balancing for geographic area, gender, and age. Duplicate
food samples and medicines were collected (lagging fecal collections by 1 day) for 7 consecutive
days, and used to estimate trace element intake from food. Low silicon/aluminum toothpaste was
used. Measures were made daily. Trace element amounts from food were subtracted from fecal
amounts and divided by soil concentrations from composite 3-inch-deep home soil samples to
estimate soil ingestion. Adjustments were not made for toothpaste ingestion, nor was urine
collected. The study resulted in soil ingestion estimates corresponding to the average (over 5-7 days)
ingestion for soil calculated on 44 children age 1-2 years and 20 children age 3 years.

Interpretations of estimates of soil ingestion from these studies have several limitations. Two
of the studies (in Montana) were conducted on children in areas with contaminated soil. All studies
were conducted in the north in the summer or early fall. In one study (in Amherst), children were
predominantly from two-parent, highly educated households. These factors serve to limit
generalizability.

Other limitations were noted. Adjustments for medicine were not made in the Helena,
Montana, study. Adjustments for toothpaste were only made in the Washington State and Anaconda
studies. In light of the magnitude of the adjustment cited by Davis et al. (1990), estimates of soil

5-7


-------
ingestion based on Si in the Amherst and Helena studies are positively biased. Urine was partially
collected only in the Washington State study, and hence soil ingestion in other studies is
underestimated unless absorption of trace elements is 0 percent. Finally, the short time periods used
in the study designs, along with the presumed skewed nature of the daily soil ingestion distribution,
implies that simple estimates are likely to be positively biased when used to estimate longer-term
average soil ingestion (such as over a 2-year period) (Stanek et al., 1998a; Stanek and Calabrese,
2000).

5.3.2 Pica

We discuss soil pica with reference to the child identified as displaying pica in the Amherst
soil ingestion study (Calabrese et al., 1993c). This child, a 3.5-year-old female, had fecal samples
collected for four days in each of two consecutive weeks. The amount of Al and Si ingested on each
day from food was less than 100 mg of equivalent soil for each day of food sampling. Using the
home soil concentrations of 44 and 354 mg/g for Al and Si, respectively, and Table 2 in Calabrese
et al. (1993 c), the amount of soil ingested on each day for the subject (not accounting for food) is
presented in Table 5-1.

Table 5-1. Soil Equivalent Amount in Fecal Samples for Pica Subject, by Week (mg/day)
(not subtracting amounts of trace elements from food)

Week

Tracer

Day 1

Day 2

Day 3

Day 4

1

Al

168

28

51

155

1

Si

249

220

49

178

2

Al

6

18,789

4

35,662

2

Si

16

19,961

8

23,976

Source: Data adapted from Calabrese et al., 1993b.

Note that minimal soil ingestion appears to have occurred during the first study week, and
in fact no soil ingestion appears to have occurred on days 1 and 3 of the second week. Nevertheless,
on the second day of week 2, the child appears to have ingested over 18 g of soil, while on day 4, the
child appears to have ingested over 23 g of soil. For this child, using 18 g and 23 g as soil ingestion
levels for these days, and assuming soil was ingested only on these 2 days during the 1,095-day
period that the child was between the age of 3 and 5 years, the child's average daily soil ingestion
would be 37 mg/day, not an unusual quantity. However, if the high soil ingestion behavior was
repeated over this time period, average soil ingestion could be much higher. Distinguishing between

5-8


-------
the dramatically different conclusions requires long-term data on soil ingestion. Such data are
lacking. Efforts to link parental reported soil pica in children may be reproducible (see Stanek et al.,
1998b). However, follow-up soil ingestion studies on such children (Calabrese et al., 1997b) have
not successfully characterized the apparently observed soil pica behavior. In the absence of
knowledge of the longer-term frequency and magnitude of pica behavior in a child, we do not
consider a recommendation possible for soil pica based on available data.

5.4 RECOMMENDATIONS FOR PROPOSED AGE BINS

5.4.1 Soil Ingestion

Recommendations for soil ingestion are summarized in Table 5-2. The recommendations
are based principally on three of the four primary mass-balance soil ingestion studies previously
discussed (omitting the Helena, Montana, study). The estimates were constructed in the following
manner. First, for each study, trace element (Al and Si), and age group, an estimate of mean,
median, and 90th percentile was made in each study based on children that fell within that age range.
The trace elements Al and Si were used since they were included in all studies. Estimates based on
Ti were not included because of the likelihood of nonfood, nonsoil sources of ingestion. The range
of these estimates is reported. Note that this range is not a confidence interval. It simply indicates
the extent of variability of the individual study estimates. Next, three strategies were used to
combine the study's trace-element-specific estimates and form a single estimate. The strategies
corresponded to taking the median of the six estimates and the weighted (by sample size) mean of
the estimates, with negative estimates replaced by zero. In most cases, the combined estimates
differed among themselves by less than 10 mg/day. By way of comparison, we also tabulated the
long-term soil ingestion estimates using results from the Anaconda, Montana, study (Stanek et al.,
2001). Our inclusion of the long-term estimates is based on basic methodological considerations.
Simple percentile estimates using short-term study data result in estimates on both ends of the
distribution being too extreme in comparison to what would be expected if a longer study period
were employed (Stanek etal., 1998a). The combined estimates suffer from this limitation. Weused
the results of Stanek et al. (2001) to provide a reference for the extent of this bias. For the age
groups 1-2 and 3-5 years, the recommended values attempt to more closely correspond to what we
would expect from estimates based on a longer study design. No long-term estimates were available
for the age group 6-10 years, and hence, recommended values are based solely on the short-term
study data. Results from the Helena, Montana, study were not available as age specific, and so they
have not been explicitly used. Nevertheless, the range of estimates reported is consistent with the
Helena estimates (if a correction is made for toothpaste ingestion).

5-9


-------
Table 5-2. Recommended Values for Soil Ingestion

Age Group

Mean value (mg/day)

Median (mg/day)

90th Percentile

< 1 month

NAa

NA

NA

1 -2 months

NA

NA

NA

3-5 months

NA

NA

NA

6-11 months

NA

NA

NA

1 -2 years

30 (0-47)b

24 (0-86)

100 (42-257)

3-5 years

30 (0-76)

20 (0-41)

150 (54-258)

6-10 years

71 (65-77)

37 (35-40)

187(164-211)

11-15 Years

NA

NA

NA

16-17 Years

NA

NA

NA

a NA ~ Not available

b The range accompanying an estimate describes estimates from different studies, not confidence intervals.

Primary data were available for each of the other studies, with the exception of toothpaste
data from the Washington State study. Using the primary data, simple soil ingestion estimates for
the mean, median, and 90th percentile were tabulated for each of three age groups. An adjustment
(subtraction of 1.8 mg/day for A1 and 45.4 mg/day for Si) was made for toothpaste ingestion based
on the average difference reported by Davis et al. (1990) (see Section 5.3) for the mean and median
estimates for the age groups 1 -2 and 3 -5 years in the Washington State study. A similar adjustment
(subtraction of 2.2 mg/day for Al and 42.7 mg/day for Si, with the difference due to the different
average concentrations of the trace elements in soil) was made for the mean and median estimates
for the age groups 1-2 years and 3-5 years in the Amherst study. No adjustment for toothpaste
ingestion was made for the age group 6-10 years in the Washington State study, since ingestion of
toothpaste at this age is reported to be low (Barnhart et al., 1974). Estimates for the age group 6-10
years are based solely on 36 subjects ages 6-7 years in the Washington State study. The
recommended estimates for the age group 6-10 years are based simply on the midpoint of the Al and
Si estimates for these 36 subjects.

We briefly provide details of the combined estimates. For reference, among 64 children ages
1-4 from Anaconda, Montana, the long-term estimate of the mean was 31 mg/day; the median, 25
mg/day; and the 90th percentile, 75 mg/day (Staneketal., 2001). A total of 103 children contributed
to the estimate for 1-2 year olds. The three combined estimates of the mean ranged from 10 to 15
mg/day, and the three combined estimates of the median ranged from 4 to 11 mg/day. There were
two combined estimates of the 90th percentile, which ranged from 161 to 171 mg/day. A total of 89
children contributed to the estimate for 3-5 year olds. The three combined estimates of the mean
ranged from 31 to 39 mg/day and for the median from 14 to 16 mg/day. The two estimates of the

5-10


-------
90th percentile ranged from 168 to 198 mg/day. A total of 36 children were in the age range of 6-10
years from the Washington State study. Estimates for these children correspond to estimates for A1
and Si for the study time period for these children.

Recommendations for the mean and 90th percentile soil ingestion that are provided in Table
5-2 differ from recommendations provided in Table 5-19 of the draft Child Specific Exposure
Factors Handbook. The CSEFH estimates the mean as 100 mg/day(as compared with30-71 mg/day
in Table 5-2) and an upper percentile as 400 mg/day (as opposed to 100-187 mg/day in Table 5-2).
The estimates differ by up to fourfold, even though data used in forming the estimates are the same
for four of the six studies, and both reports base estimates principally on the trace elements of A1 and
Si. Inclusion of the two additional studies by the CSEFH is not the principal source of the
difference, since estimates from these two studies overlap the estimates from the other four studies.
Instead, the difference is attributable to differences in how soil ingestion is determined. There are
differences in the way the soil ingestion estimate is determined for three of the four studies. For the
Helena, Montana, study (Binder et al., 1986), the differences result from using actual reported fecal
sample weights, subtracting an estimate of trace element intake from food, and subtracting an
estimate of trace element intake from toothpaste. For the Amherst, Massachusetts, study (Calabrese
et al., 1989), the differences result from excluding the pica subject and subtracting an estimate of
trace element intake from toothpaste. For the Washington State study (Davis et al., 1990), the
difference is due to the subtraction of trace element intake from toothpaste. It should be noted that
the adjustment for toothpaste principally affected estimates based on Si for children in the categories
1-2 and 3-5 years. As aresult ofthese differences, the estimate ofmean (and median) soil ingestion
in Table 5-2 is lower than the estimate given in Table 5-19 in the CSEFH. Differences are also
evident between the recommended estimates for the 90th percentile, and the estimates given in Table
5-19 in the CSEFH. No adjustments to the data used were made when forming the estimates
presented in Table 5-2. Differences are likely due to the use of a different percentile in Table 5-19
in the CSEFH, and the inclusion of the Dutch studies.

Table 5-3 (at the end of this section) summarizes a measure of confidence in the estimates.
It should be noted that confidence in the 90th percentile estimates is much lower than confidence in
estimates of the mean or median.

5.4.2 Pica

As previously mentioned, in the absence of knowledge of the longer-term frequency and
magnitude of pica behavior in a child, we do not consider a recommendation possible for soil pica
based on available data.

5-11


-------
5.5

RECOMMENDATIONS FOR FURTHER ANALYSIS AND RESEARCH NEEDS

As is clear from Table 5-2, there are many areas where little is known about soil ingestion.
Even for the few cells in which recommendations have been provided, there is no evidence to
support generalizing the estimates to other seasons of the year or to other parts of the country. It is
unlikely that there are sufficient data available to reliably distinguish soil ingestion in different age
ranges. While such a distinction may in fact exist, we consider it yet to be uncovered.

It is possible using existing data to improve on the estimates of soil ingestion for age
groupings. Such estimation can be accomplished by taking advantage of the multiple estimates of
soil ingestion (using the multiple trace elements) for a subject. This strategy is particularly effective
for removing effects of source errors (such as toothpaste, or other nonfood/nonsoil ingestion) that
occur for individual elements. Using such a strategy (similar to that in Stanek et al, 2001), a single
estimate of soil ingestion can be obtained and the distribution characterized. Such a strategy would
help focus and possibly narrow the range of estimates provided and would have higher reliability.

The main limitation in completing the cells in Table 5-2 is the lack of adequate data. Soil
ingestion studies are difficult to conduct. Data collection and chemical processing are expensive.
Nevertheless, new soil ingestion data are needed. The last soil ingestion study data were collected
in 1992. Although much has been learned about the conduct of such studies and analysis of the data,
the most critical need at this point is new data. The agenda for new data is large and will not be
accomplished with a single large study. Instead, our understanding of soil ingestion needs to be
pursued in the context of a research program.

Data are needed that span a broader age range, perhaps initially expanding the age range to
extend from 3 months to 12 years. Data on children in future soil ingestion studies need to span the
range of demographic variables such as geography, race, and economic status so that results can be
more confidently applied to the general U.S. population. Estimates of soil ingestion need to reflect
longer time periods. Seasonal effects and longitudinal studies (both over seasons and over years)
are important to identify tracking that may lead to a broader or narrower soil ingestion distribution.
Finally, soil ingestion studies need to be integrated with behavioral studies and made efficient.
Much has been learned as a result of the conduct of soil ingestion studies in the past, and this needs
to be taken advantage of in the future.

5-12


-------
5.6 REFERENCES

Annest, J.L.; Mahaffey, K. (1984) Blood lead levels for persons ages 6 months to 74 years, United
States, 1976-1980. U.S. Department of Health and Human Services. Data from the National
Health and Nutritional Survey, Series 11, No 233. Hyattsville, MD.

Barnhart, W.E.; Hiller, L.K.; Leonard, G.J.; Michaels, S.E. (1974) Dentifrise usage and ingestion
among four age groups. Journal of Dental Research 53:1317-1322.

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. Archives of Environmental Health
41:341-345.

Calabrese, E.J.; Barnes, R.; Stanek, E.J. Ill; Pastides, H.; Gilbert, C.E.; Veneman, P.; Wang, X.;
Lasztity, A.; Kostecki, P.T. (1989a) How much soil do young children ingest?: an
epidemiologic study. Regulatory Toxicology and Pharmacology 10:123-137.

Calabrese, E.J.; Pastides, H.; Barnes, R.; Edwards, C.; Kostecki, P.T.; Stanek, E.J. Ill; Veneman, P.;
Gilbert, C.E. (1989b) How much soil do young children ingest?: an epidemiologic study. In:
Petroleum contaminated soils. Calabrese, E.J., Kostecki, P.T., eds. pp. 363-356.

Calabrese, E.J.; Stanek, E.J. Ill; Gilbert, C.E.; Barnes, R.M. (1990) Preliminary adult soil ingestion
estimates: results of a pilot study. Regulatory Toxicology and Pharmacology 12:88-95.

Calabrese, E.J.; Stanek, E.J. Ill (1991) A guide to interpreting soil ingestion studies: II. Qualitative
and quantitative evidence of soil ingestion. Regulatory Toxicology and Pharmacology
13:278-292.

Calabrese, E.J.; Stanek, E.J. Ill; Gilbert, C. (1991) Evidence of soil-pica behaviour and
quantification of soil ingestion. Human and Experimental Toxicology 10:245-249.

Calabrese, E.J.; Stanek. E.J. (1993) Soil pica: not arare event. Journal of Environmental Science and
Health, Part A 28(2):373-384.

Calabrese, E. J.; Stanek, E.J. Ill; Gilbert, C.E. (1993 a) A preliminary decision framework for deriving
soil ingestion rate. In: Principles and practices for petroleum contaminated soils. Calabrese,
E.J., Kostecki, P.T., eds. pp. 613-624.

Calabrese, E.J.; Stanek, E.J. Ill; Gilbert, C.E. (1993b) Lead exposure in a soil pica child. Journal of
Environmental Science and Health, Part A 28(2):353-362.

Calabrese, E.J.; Stanek, E.J. Ill (1994) Soil ingestion issues and recommendations. Journal of
Environmental Science and Health, Part A 29(3):517-530.

5-13


-------
Calabrese, E.J.; Stanek, E.J. III; Pekow, P.; Barnes, R. (1997a) Soil ingestion rates in children
residing on a Superfund site. Ecotoxicology and Environmental Safety 36:258-268.

Calabrese, E.J.; Stanek, E.J. Ill; Barnes, R. (1997b) Soil ingestion rates in children identified by
parental observation as likely high soil ingesters. Journal of Soil Contamination 6(3): 271 -
279.

Calabrese, E.J.; Stanek, E.J. HI. (1998) Soil ingestion estimation in children and adults: a dominant
influence in site specific risk assessment. Environmental Law Reporter 28(11): 10660-10672.

CDC (Centers for Disease Control and Prevention). (1997) Update: blood lead levels—United
States, 1991-1994. Morbidity and Mortality Weekly Report 46(07): 141-146. Available at

http://www.cdc.gov/mmwr/preview/mmwrhtml/00048339.htm.

Clausing, P.; Brunekreef, B.; van Wijnen, J.H. (1987) A method for estimating soil ingestion by
children. International Archives of Occupational and Environmental Health 59:73-82.

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. Archives of Environmental
Health 45:112-122.

Day, J.P.; Hart, M.; Robinson, M.S. (1975) Lead in urban street dust. Nature 253:343-45.

De Silva, P.E. (1994) How much soil do children ingest? a new approach. Applied Occupational and
Environmental Hygiene 9:40-43.

Kimbrough, R.D.; Falk, H.; Stehr, P.; Fries, G. (1984) Health implications of 2,3,7,8-
tetracholorodibenzodioxin (tcdd) contamination of residential soil. Journal of Toxicology and
Environmental Health 14:47-93.

Lepow, M.L.; Bruckman, L.; Rubino, R.A.; Markowitz, S.; Gillette, M.; Kapish, J. (1974) Role of
airborne lead in increased body burden of lead in Hartford children. Environmental Health
Perspectives 7:99-102.

National Research Council. (1980) Lead in the human environment. Washington, DC: National
Research Council.

Sedman, R.M. (1989) The development of applied action levels for soil contact: a scenario for the
exposure of humans to soil in a residential setting. Environmental Health Perspectives.
79:291-313.

Sedman, R.M.; Mahmood, R.J. (1994) Soil ingestion by children and adults reconsidered using the
results of recent tracer studies. Journal of the Air and Waste Management Association
44:141-144.

5-14


-------
Stanek, E.J. III; Calabrese, E.J.; Zheng, L. (1991) Soil ingestion estimates in children: influence of
sex and age. Trace Substances in Environmental Health 24:33-43.

Stanek, E.J. Ill; Calabrese, E.J.; Xu, Limin. (1998a) A caution for Monte Carlo risk assessment of
long term exposures based on short term exposure study data. Human and Ecological Risk
Assessment 4:409-422.

Stanek, E.J. Ill; Calabrese, E.J.; Mundt, K.; Pekow, P.; Yeatts, K.B. (1998b) Prevalence of soil
mouthing/ingestion among healthy children age 1 to 6. Journal of Soil Contamination
7(2):227-242.

Stanek, E.J. Ill; Calabrese, E.J. (2000) Daily soil ingestion estimates for children at a superfund site.
Risk Analysis 20:627-635.

Stanek, E.J. Ill; Calabrese, E.J.; Zom, M. (2001) Soil ingestion distributions for Monte Carlo risk
assessment in children. Human and Ecologic Risk Assessment 7:357-368.

Ter Haar, G.; Aronow, G.R. (1974) New information on lead in diet and dust as related to the
childhood lead problem. Environmental Health Perspectives 7:83-89.

Thompson, K.M.; Burmaster, D.E. (1991) Parameteric distributions for soil ingestion by children.
Risk Analysis 2:339-342.

van Wijnen, J.H.; Clausing, P.; Brunekreef, B. (1990) Estimated soil ingestion by children.
Environmental Research 51:147-162.

Wong, M.S. (1986) The role of environmental and host behavioural factors in determining exposure
to infection with Ascaris lumbricoldes and Trichuris trichlura. Ph.D. dissertation, Faculty
of Natural Sciences, University of the West Indies.

5-15


-------
Table 5-3. Confidence in Recommendations for Soil Ingestion and Pica

Rating (High, Medium, Low)
Considerations	7



<1 Month

1-2 Mos

3-5 Mos

6-11 Mos

1-2 Yrs

3-5 Yrs

6-10 Yrs

11-15 Yrs

16-17 Yrs

Study Elements



















• Level of Peer Review

NA

NA

NA

NA

High

High

High

NA

NA

• Accessibility

NA

NA

NA

NA

Medium

Medium

Medium

NA

NA

• Reproducibility

NA

NA

NA

NA

High

High

High

NA

NA

• Focus on factor of interest

NA

NA

NA

NA

High

High

High

NA

NA

• Data pertinent to U.S.

NA

NA

NA

NA

High

High

High

NA

NA

• Primary data

NA

NA

NA

NA

High

High

High

NA

NA

• Currency

NA

NA

NA

NA

Medium

Medium

Medium

NA

NA

• Adequacy of data collection period

NA

NA

NA

NA

Low

Low

Low

NA

NA

• Validity of approach

NA

NA

NA

NA

Medium

Medium

Medium

NA

NA

• Representativeness of the population

NA

NA

NA

NA

Medium

Medium

Medium

NA

NA

• Characterization of variability in the population

NA

NA

NA

NA

Low

Low

Low

NA

NA

• Lack of bias in study design

NA

NA

NA

NA

Medium

Medium

Low

NA

NA

• Measurement error

NA

NA

NA

NA

Low

Low

Low

NA

NA

Overall Rating

NA

NA

NA

NA

Medium

Medium

Low

NA

NA

5-16


-------
6.0

NON-DIETARY INGESTION EXPOSURE

6.1	INTRODUCTION

Non-dietary ingestion pathways of exposure are simultaneously one of the most important
pathways for exposure in children and one of the least studied. For young children, mouthing
activities offer one of the most common ways for the child to explore his or her environment.
However, contamination of any object used in mouthing activities may lead to elevated exposure to
a variety of chemical compounds, including metals, pesticides, and other potentially toxic
compounds.

The purpose of this document is to identify sources of data on the non-dietary pathway of
exposure for children and to assess the utility of that data in EPA's proposed age bins for the
purposes of exposure assessment. The Child-Specific Exposure Factors Handbook, Chapter 6,
identified a number of studies related to non-dietary exposure (U.S. EPA, 2001). In this paper, we
evaluate those data that pertain to non-dietary ingestion exposure, identify other data of interest on
this topic, discuss strengths and weaknesses of the data, and suggest new research activities to
identify important factors related to children's non-dietary exposure.

Gurunathan et al., (1998) sum up the literature on non-dietary exposure factors: "There is
a paucity of data available for making an accurate assessment of the relative importance of oral (non-
dietary), dermal, and inhalation exposures to household pesticides." In the work of Gurunathan et
al., as well as that of others, the difficulty of gathering high-quality, representative data for this
important pathway is evident. Only a few studies are available that provide these type of data
effectively. It is safe to say that no systematic, representative study of any large population exists
in the current literature.

In the next section, we review the studies used to develop the CSEFH for their applicability
to age categories suggested by EPA. All of the studies are lacking in some fundamental way. We
then examine the literature since 1999 to assess the applicability of any studies not considered by the
authors of the CSEFH.

6.2	EVALUATION OF EXISTING DATA

As part of the development of the draft CSEFH, EPA researchers selected several studies to
use as the basis for proposed factors in non-dietary exposures. The specific works include Davis,

6-1


-------
1995; Zartarian et al, 1997; Groot et al, 1998; Stanek et al, 1998; and Reed et al, 1999. Each is
discussed below with emphasis on the question of age-group categorization.

The Davis (1995) study, as described in the draft CSEFH, observed mouthing activities of
a total of 92 children ranging in age from birth to 48 months. The study relied on observations made
by the child's parent and invoked the so-called interval method of observing the child for a certain
interval every hour. Presumably, age data were available on the children so that activities could be
placed in the proposed age bins. However, with only 92 children in the study, the number in each
age bin is likely to be small.

Zartarian et al. (1997) presented results of a four-child pilot investigation designed to assess
the use of a videotaping methodology to help assess activities that could result in non-dietary
exposure. The children monitored were two girls ages 2 years 5 months and 4 years 2 months, and
two boys ages 2 years 10 months and 3 years 9 months. The data were collected in September 1993
and comprise a total of 33 hours of activity. Activities were coded using a coding program called
VideoTraq, which allows coding of specific events. Coders watched the videotaped activities of
each child and coded activities such as "right hand touching hard toy," as well as the location and
activity level of each child. The Dermal Exposure Reduction Model (DERM) software totaled the
number of seconds in the monitoring period that the child was in contact with the object. For each
of the four children, Zartarian et al. reported the total amount of time each hand was in contact with
each of 19 different items, as well as summary statistics regarding the central tendency of contact
time for each hand of each child.

The study was well designed and well executed given its initial intent, which was to show
the efficacy of a videotaping methodology. Given that only four children were studied, two from
each gender, and that they were the children of farm workers in Salinas, California, the
representativeness of the data set is certainly in question. However, the technique showed merit and
precipitated the Reed et al. (1999) work discussed later in this section. The data were presented in
both tabular and graphical format and thus were readily available.

The Groot et al. (1998) study is an observational study focusing on mouthing activities of 42
Dutch children, with data collected in the summer of 1998. The children ranged in age from 3
months to 36 months. The researchers placed the children in four age categories: 3 to 6 months, 6
to 12 months, 12 to 18 months, and 18 to 36 months, with age categories selected by expected
behavioral (not physiological) characteristics. The number of children in each category was small:
15, 14, 12, and 11, respectively. No differences were found between boys and girls with regard to
activities, while large differences were found between age groups. Further, within an age group,

6-2


-------
variability, as measured by standard deviation in duration of mouthing activities, often exceeded the
mean value.

The Groot et al. investigation was quite well designed, with a number of good quality
assurance procedures. Most interesting among these is the attempt to validate observations using
shadow observations and studies of inter- and intraobserver variability. In each case it was found
that observers gave like estimates of total mouthing time during the observing intervals. The
children were placed in age categories fairly closely aligned with the proposed age bins, so if the data
could be obtained from the authors (likelihood unknown), they could be applied to the 3- to
36-month categories. The study suffers in at least two respects. First, the sample size was small,
especially in light of the large variability in results observed. Second, the study was done in the
Netherlands and on a population of children of well-educated parents. Thus the applicability to U. S.
populations is unknown.

Stanek et al. (1998) is a study of the activities of 533 children, focusing on activities that are
thought to affect non-dietary exposure. The study was conducted using face-to-face interviews with
the parent or guardian of the child, with questions regarding frequency of 28 mouthing behaviors for
the children. The children ranged in age from 1 to 6 years. Responses were given as "daily, at least
weekly, at least monthly, and never."

Although the study had the largest number of participants, it had many limitations that most
likely would preclude the use of its data for the proposed age bins. In many ways, the Stanek et al.
data are not comparable to the other studies. Rather than reporting the frequency of events per day
or per hour, the results were presented in events per week, and the frequencies reported are much
lower than those reported in other studies. For example, outdoor ingestion rates were estimated to
be 4.73 times per week compared with an estimate of approximately 9 times per hour for indoor
mouthing activities reported by both Reed et al. (1999) and Zartarian et al. (1997). This discrepancy
is most likely due to the nature of the data collection, which was through recall of activities occurring
in the past. Also, there was no sensitization of the adults to the required observations before the data
were collected. Unlike the videotaping or observational studies, these data were collected "cold."
Other studies have suggested that untrained observers need time to develop their understanding of
what is meant by a mouthing activity.

The data in the Stanek et al. study were collected in detail. Age categories were blocked by
year of age, but the actual ages may be available from the authors, allowing finer resolution to be
gleaned from the data. In general, the younger children (1-year-olds) participated in most activities
of interest far more frequently than older children (e.g., 6-year-olds). Although the number of

6-3


-------
children evaluated in the Stanek et al. study is greater than in all the other studies combined, the
difference in study design and reporting preclude its use because its data are not comparable.

Reed et al. (1999) reported the results of a study for mouthing activities collected using a
videotaping methodology. The sample of children monitored included 20 at a day-care center and
10 in private residences. Children at the day-care center ranged in age from 3 to 6 years, with a mean
age of 56.8 ± 9.7 months. The children in the private residences ranged in age from 2 to 5 years,
with a mean of 43.2 ± 12.6 months. The coding of videotaped activities focused on different types
of mouthing activities including directly mouthing surfaces and objects and placing objects in the
mouth. A significant effort was made toward maintaining quality assurance of the coding activities;
training programs were implemented to ensure inter- and intracoder reliability. Thus, the data can
be considered of high quality. The paper presented hourly frequency data for hand and mouthing
activities for each child observed. Data presented include the range, mean, and median frequencies
for each of four activities, grouped by those in day-care settings and those in residential settings. The
authors also presented two figures, in sufficient detail to assess inter- and intra-individual variability
in these activities, which was substantial.

With regard to the study's applicability to the proposed age bins, no age group data can be
fit to the bins. Although the authors gave the age ranges for the groups as a whole, no data were
supplied on the individual ages of the children, so age-specific contact rates cannot be ascertained.
Such data could presumably be obtained from the authors and would prove useful for the needs of
EPA; however, we present two cautions. First, the overall age span of the individuals studied was
limited. Second, the small numbers of individuals and the lack of representativeness of the sample
(20 children drawn from a single university-affiliated day-care center and 10 individuals from two
cities in New Jersey) make generalization of the data suspect.

It should be noted that authors of the Reed et al. (1999) study reported a much higher value
for mouthing frequency, approximately 9.5 contacts per hour versus the value of 1.5 contacts per
hour reported in the CSEFH. The value reported by Reed et al. is in agreement with that found by
Zartarian et al. (1997) in their pilot investigation.

6.3 NEW STUDIES

A literature search performed using both Medline and Science Citation Index (using the
identifiers "children" and "non-dietary exposure") identified 13 references, dated 1999 and later, to
be of interest. None of these affords a better analysis of the age-specific components of non-dietary
exposure. However, the data and models identified may be of use in further characterizations of

6-4


-------
these parameters. Several of these (Kissel and Fenske, 2000; Oliver et al., 2000; Buck et al., 2001;
Lunchick, 2001; and Youngren et al, 2001) are modeling efforts geared toward understanding
exposures experienced by not only children, but adults as well. Several papers discuss specific
monitoring studies. These include Fenske et al. (2000), Lu et al. (2000), Wilson et al. (2000), Wong
et al. (2000), and Moschandreas et al. (2001). Still others (Faustman et al., 2000, and Hubal et al.,
2000) discuss other aspects of non-dietary exposure effects, some of which are specific to children.
The following sections briefly discuss these modeling efforts and monitoring studies.

6.3.1 Modeling Efforts

Oliver et al. (2000) discussed the need for using probabilistic techniques in all assessments
of exposure under the Food Quality Protection Act. Although not specifically geared toward
understanding children's exposure, their work suggests the need for more relevant data collection
efforts.

The work of Kissel and Fenske at the University of Washington has pushed the field of
dermal and non-dietary exposure for children forward in the past 10 years. In an effort to model the
effects of dermal transfer rates, these two authors have combined talents to estimate dermal exposure
doses (Kissel and Fenske, 2000). Although the major application of the model is to an
occupationally exposed cohort, such efforts should be applied to children's exposure with an eye
toward both physiological and behavioral differences in children that would improve understanding
of the effects of the contact exposure and inadvertent ingestion.

Buck et al. (2001) developed a probabilistic, multimedia, multipathway exposure model and
assessment for chlorpyrifos as part of the National Human Exposure Assessment Survey (NHEXAS).
The model was developed using data on the general population, collected in Arizona, and on children
ages 3-12 in Minneapolis-St. Paul. Such a model, especially using the Minneapolis-St. Paul data,
could prove useful in identifying the importance of non-dietary exposure in a representative sample.
Multiple exposure and control scenarios could be explored using such a modeling approach.

Lunchick (2001) discussed the role of probabilistic modeling in the risk assessment and
regulatory process. Although not a model as such, the author presented an interesting discussion of
policy implication and the potential for modeling to aid in the process of determining likely exposure
distributions. This work focused on occupational exposures; however, the "philosophy" of the paper
is applicable to children's exposures through non-dietary routes as well.

6-5


-------
Youngren et al. (2001) produced a companion piece to the Lunchick paper outlining the need
to address uncertainty in exposures determined through modeling efforts. One focus of the paper
is the manner in which available data are used to develop estimates of exposure factors. This is
relevant to the process being undertaken in the present work on child-specific exposure factors.

6.3.2 Monitoring Studies

Fenske et al. (2000) presented the results of a study of organophosphorus pesticide exposure
in 109 children using a biological marker of exposure. Most of the study subjects were children of
agricultural workers in Washington State. No specific data on non-dietary exposure are given;
however, this data set is rich and, coupled with the biological measurement of exposure, should be
explored for non-dietary exposure markers. Lu et al. (2000) (from the same research group)
discussed dermal concentrations using results from the same study as that of Fenske et al. (2000).
Such data, coupled with modeling efforts and, perhaps, the earlier videotaping studies, may prove
useful in modeling the uptake of pesticides and may be correlated with urinary biomarker output.

Wilson et al. (2000) reported on an investigation of nine children ages 2-5 years. Data
collected include floor dust samples, outdoor play area soil, hand samples, and solid and liquid food.
Biomonitoring samples (i.e., urine samples) were also collected. Such data could prove useful in
identifying the role of inadvertent ingestion relative to exposure experienced through other
pathways/routes. Although the sample size is very small, few such data are available anywhere.

Wong et al. (2000) reported on the results of a telephone survey assessing the amount of time
children have contact with soil, grass, and other outdoor surfaces. Such data are useful in
establishing larger databases of information regarding non-dietary exposure-related activities but
may be more focused on time-activity data than on non-dietary exposure. However, telephone
surveys rely on recall and thus are subject to the errors discussed previously in this chapter. These
data are part of the University of Washington Research Program.

Moschandreas et al. (2001) reported results of exposure to two organophosphorus pesticides,
chlorpyrifos and diazinon, experienced by a population in Arizona, and attempted to correlate the
biomarker results with exposure. This study was part of the NHEXAS investigations, a
multicomponent investigation undertaken by several research groups. The exposures were modeled
in Buck et al. (2001). The NHEXAS data sets, with data collected in Arizona, the Upper Great
Lakes states, Minneapolis-St. Paul, and Baltimore may prove quite valuable in establishing non-
dietary exposure factors for children when the data become generally available within the next year.

6-6


-------
6.3.3 Other Papers of Interest

Faustman et al. (2000) discussed the importance of child-specific risk assessment in a special
children's health supplement to the journal Environmental Health Perspectives. Hubal et al. (2000)
presented an overview of factors important in the analysis of children's environmental exposure-
related risk.

6.4	STUDIES SELECTED FOR ANALYSIS TO OBTAIN NEW RECOMMENDATIONS

At this time, no studies are recommended as definitive in aiding the selection of age-
appropriate factors for the age bins proposed by EPA. There have been no systematic,
probability-based studies undertaken that would afford a reasonable assessment of such, nor have
any studies been designed and implemented that would determine whether age-bin-specific factors
differ from one another. Sample sizes in the studies outlined above are too small. Study designs are
not consistent and often span only part of the range (e.g., very little work has been done on children
ages 10-20).

6.5	RECOMMENDATIONS FOR PROPOSED AGE BINS

No recommendations can be made for factors for the proposed age bins at this time. The
existing data are derived from a series of small studies conducted by pioneers in a nascent field. The
data are of high quality, as evidenced by their placement in the top exposure-related, peer-reviewed
journals. However, studies of four (or even 92) children, or of children in other countries, as well
as more sophisticated modeling efforts being undertaken, do not give confidence in establishing
exposure factors for children of various ages, despite the apparent agreement between some of the
studies. Such efforts are grounded in the data currently available, and those data are simply too
sparse. Choosing from among these studies or from among the small number of new investigations
is simply not warranted.

Exposure assessment, especially in a new area such as non-dietary ingestion, is data driven.
It is difficult to address the uncertainties introduced by using the existing data to assess exposures
for children. The uncertainty associated with using data from, say, a single 2-year-old child with
regard to mouthing activities and projecting that to a population of 2 to 4-year-olds across the
country cannot be judged. In addition, numerical estimates derived from the studies evaluated are
not addressed. It is difficult to discuss whether numerical estimates for a population are good
without at least a modicum of data to assess their validity. Until new data are collected in a
systematic, representative manner, factors selected will have to be used only with the strongest

6-7


-------
cautions. A confidence evaluation of how well the existing non-dietary exposure studies used in the
current CSEFH address the proposed age groups is presented in Table 6-1.

6.6	RECOMMENDATION FOR FURTHER ANALYSIS AND RESEARCH NEEDS

Throughout this discussion, numerous recommendations have been made regarding analyses
of existing data sets or modifications of study design. The most important research need at present
in the non-dietary exposure factors area is a systematic method of data collection. Current meta-
analysis is limited by inconsistencies in data and small sample sizes. One clear recommendation to
be made from this assessment of existing data is that a new, comprehensive data collection effort to
determine non-dietary ingestion exposure factors should be designed and undertaken.

6.7	REFERENCES

Buck, R.J.; Ozkaynak, H.; Xue, J.; Zartarian, V.G.; Hammerstrom, K. (2001) Modeled estimates of
chlorpyrifos exposure and dose for the Minnesota and Arizona NHEXAS populations.
Journal of Exposure Analysis and Environmental Epidemiology 11(3):253-268.

Davis, S. (1995) Soil ingestion in children with pica (Final report). EPA Cooperative Agreement
CR816334-01.

Faustman, E.M.; Silbernagel, S.M.; Fenske, R.A,; Burbacher, T.M.; Ponce, R.A. (2000) Mechanisms
underlying children's susceptibility to environmental toxicants. Environmental Health
Perspectives 108(Suppl 1): 13-21.

Fenske, R.A.; Kissel, J.C.; Lu, C.; Kalman, D.A.; Simcox, N.J.; Allen, E.H.; Keifer, M.C. (2000)
Biologically based pesticide dose estimates for children in an agricultural community.
Environmental Health Perspectives 108(6):515-520.

Groot,M.; Lekerkerk, M.; Steenbekkers, L. (1998) Mouthing behavior in children: an observational
study. Wageningen, Netherlands: Agricultural University Wageningen.

Gulson, B.L.; Davis, J.J.; Mizon, K.J.; Korsch, M.J.; Law, A.J.; Howarth, D. (1994) Lead
bioavailability in the environment of children: blood lead levels in children can be elevated
in a mining community. Archives of Environmental Health 49(5):326-331.

Gurunathan, S.; Robson, M.; Freeman, N.; Buckley, B.; Roy, A.; Meyer, R.; Bukowski, J.; Lioy, P.
(1998) Accumulation of chlorpyrifos on residential surfaces and toys accessible to children.
Environmental Health Perspectives 106(1):9-16.

6-8


-------
Hubal, E.; Sheldon, L.; Burke, J.; McCurdy, T.; Berry, M.; Rigas, M.; Zartarian, V.; Freeman, N.
(2000) Children's exposure assessment: a review of factors influencing children's exposure,
and the data available to characterize and assess that exposure. Environmental Health
Perspectives 108(6).

Kissel, J.; Fenske, R. (2000) Improved estimation of dermal pesticide dose to agricultural workers
upon reentry. Applied Occupational and Environmental Hygiene 15(3):284-290.

Lu, C.; Fenske, R.A.; Simcox, N.J.; Kalman, D. (2000) Pesticide exposure of children in an
agricultural community; evidence of household proximity to farmland and take home
exposure pathways. Environmental Research 84(3):290-302.

Lunchick, C. (2001) Probabilistic exposure assessment of operator and residential non-dietary
exposure. Annals of Occupational Hygiene 45(Suppl l):S29-42.

Moschandreas, D.J.; Kim, Y.; Karuchit, S.; Ari, H.; Lebowitz, M.D.; O'Rourke, M.K.; Gordon, S.;
Robertson, G. (2001) In-residence, multiple route exposures to chlorpyrifos and diazinon
estimated by indirect method models. Atmospheric Environment 35(12):2201-2213.

Oliver, G.R.; Bolles, H.G.; Shurdut, B.A. (2000) Chlorpyrifos: probabilistic assessment of exposure
and risk. Neurotoxicology 21 (1 -2):203-208.

Reed, K.; Jimenez, M.; Freeman, N.; Lioy, P. (1999) Quantification of children's hand and mouthing
activities through a videotaping methodology. Journal of Exposure Analysis and
Environmental Epidemiology 9(5):513-520.

Stanek, E.J. Ill; Calabrese, E.J.; Mundt, K.; Pekow, P.; Yeatts, K.B. (1998) Prevalence of soil
mouthing/ingestion among healthy children age 1 to 6. Journal of Soil Contamination
7(2):227-242.

U.S. EPA (2000) Summary report on the technical workshop on issues associated with considering
developmental changes in behavior and anatomy when assessing exposures to children. Risk
Assessment Forum, Washington, DC; EPA/63 0/R-00/005.

U.S. EPA (2001) Child-specific exposure factors handbook (external review draft). Prepared by
Versar, Inc., for the Office of Research and Development, National Center for Environmental
Assessment, under EPA contract no. 68-W-99-041.

Wilson, N.K.; Chuang, J.C.; Lyu, C. (2000) PAH exposures of nine preschool children. Polycyclic
Aromatic Compounds 21(l-4):247-259.

Wong, E.Y.; Shirai, J.H.; Garlock, T.J.; Kissel, J.C. (2000) Adult proxy responses to a survey of
children's dermal soil contact activities. Journal of Exposure Analysis and Environmental
Epidemiology 10(6, Pt.l):509-517.

6-9


-------
Youngren, S.H.; Youngren, M.A.; Barraj, L. (2001) Challenges of probabilistic assessment of
operator and residential non-dietary exposure. Annals of Occupational Hygiene 45:S49-S54.

Zartarian, V.; Ferguson, A.; Leckie, J. (1997) Quantified dermal activity from a four-child pilot field
study. Journal of Exposure Analysis and Environmental Epidemiology 7(4):543-552.

6-10


-------
Table 6-1. Confidence Evaluation of the Existing Non-Dietary Exposure Studies"

11-15 Years 16-17 Years

Considerations

< 1 Month

1-2 Months

3-6 Months

6-11 months

1-2 Years

3-5 Years

6-10 Years





Study Elements





Davis
Groot

Davis

Groot

Davis
Groot
Stanek

Reed Zartarian
Davis to age 4
Groot
Stanek

Reed
Stanek





• Peer Review

NA

NA

Low/Medium

Low/Medium

Low/Medium

High

High

NA

NA

• Accessibility

NA

NA

Medium

Medium

Medium

High

High

NA

NA

• Reproducibility

NA

NA

Medium

Medium

Medium

Medium

Medium

NA

NA

• Focus on factor of
interest

NA

NA

High

High

High

High

High

NA

NA

• Data pertinent to U.S.

NA

NA

Davis High
Groot Low

Davis High
Groot Low

Davis High
Groot Low

High

High

NA

NA

• Primary data

NA

NA

High

High

High

High

High

NA

NA

• Currency

NA

NA

High

High

High

High

High

NA

NA

• Validity of approach

NA

NA

Medium

Medium

Medium

Medium
Differing
methods with
strengths and
weaknesses

Medium
Differing
methods with
strengths and
weaknesses

NA

NA

• Representativeness

NA

NA

Low

Low

Low

Low

Low

NA

NA

• Characterization of
variability

NA

NA

Low

Low

Low

Low

Low

NA

NA

• Lack of bias in study

NA

NA

Unknown

Unknown

Unknown

Unknown

Unknown

NA

NA

• Measurement error

NA

NA

Unknown

Unknown

Unknown

Unknown

Unknown

NA

NA

Other Elements



















• Number of studies

0

0

2

Unknown
Numbers

2

Unknown
Numbers

2+ 1
Unknown
Numbers

4+ 1
Unknown Numbers

2 age 6 only

0

0

• Agreement between
researchers

Not
Applicable

Not Applicable

High

High

Medium
Different
Methods

Medium
Different Methods

Medium
Different Methods

Not
Applicable

Not
Applicable

Overall Rating

Low

Low

Low
Very Low
Sample Sizes
with Unknown
Effects

Low
Very Low
Sample Sizes
with Unknown
Effects

Low
Very Low
Sample Sizes +
Potential Recall
Problems

Low
Very Low Sample
Sizes + Potential
Recall Problems

Low
Very Low Sample

Sizes with
Unknown Effects

Low

Low

a The table represents a criticism of the data as a whole, not of the work done by individual researchers. As can be seen from inspection of the table, many categories have no data at all, while
the remaining have unknown (but small) sample sizes. Confidence that these data are generally representative of exposure factors for each individual age category must be uniformly low.
This confidence will remain low until such time as a large, population-based investigation is undertaken, analyzed, and evaluated.

NA = Study data not available.

6-11


-------
7.0

EXPOSURE FACTORS FOR INHALATION

7.1	INTRODUCTION

The Child-Specific Exposure Factors Handbook (U.S. EPA, 2001) reviewed several studies
in order to estimate age-dependent inhalation rates for children. However, to develop more refined
estimates of inhalation rates for propo sed EPA age groupings, from birth through age 17, the existing
studies must be reevaluated and, where possible, updated to reflect new data or analyses.

Three basic techniques are used for estimating inhalation rates. The first technique is to
assign breathing rates to various activities and then to calculate daily values as the sum of the
products of the various breathing rates and their duration. This is referred to as an activity-based
approach. In contrast, ametabolically based approach determines breathing rate as a function of the
oxygen demand needed to provide the metabolic energy for sustaining a given lifestyle. The third
approach is a hybrid of the first two. A physiological measure of oxygen consumption, such as heart
rate, is used along with activity data and a personal calibration curve of heart rate to inhalation to
estimate how much air is inhaled during an individual's daily activities.

The sections below review studies that have used these approaches, or modifications of them,
to develop inhalation estimates for different age groups. Our focus is on the data and methods that
can best be used to develop recommendations for inhalation exposure factors for the following
proposed age groups: < 1 month, 1-2 months, 3-5 months, 6-11 months, 1-2 years, 3-5 years, 6-10
years, 11-15 years, and 16-17 years.

7.2	EVALUATION OF EXISTING DATA

An individual's breathing rate changes during the course of the day as a function of the kinds
of activities that are carried out. For inhalation exposure assessment, research must characterize
inhalation rates that are indicative of specific activities as well as inhalation rates associated with
different locations (e.g., indoors vs. outdoors) or times of day. In those types of assessments, short-
term breathing rates are used; however, other exposure assessments may require daily, or chronic,
breathing rates for the purpose of determining the inhalation of airborne contaminants. The
emphasis in this analysis is on the chronic or daily inhalation rates applicable to the proposed
childhood age groups.

7-1


-------
7.2.1 Activity-Based Estimation of Inhalation Rates

The study conducted by Allan and Richardson (1998) estimated breathing rates for different
age groups using an activity-based approach in which five separate activity levels were defined that
reflect varying levels of physical effort. These levels were (a) resting, (b) very light activity, (c) light
activity, (d) light to moderate activity, and (e) moderate to heavy activity. Probability distributions
were developed to represent the minute volumes for each activity level for the following age groups:
infants, birth to 6 months; toddlers, 7 months to 4 years; children, 5 to 11 years; teenagers, 12 to 19
years; adults, 20 to 59 years; and seniors, 60 years and older. Probability distributions were also
constructed for the various activities based on time-activity studies. The probability density
functions (PDFs) for the age-dependent inhalation rates were developed using a Monte Carlo
sampling technique to propagate the uncertainty in the products of the PDFs for activity-specific
levels and associated time periods.

Table 7-1 presents the results of simulations for the age groups through 19 years. The
primary methodological issues associated with this approach are the misspecification of the activity
levels (i.e., the time- orminute-volume PDF for a given activity level is biased in some way) or their
misclassification (i.e., a given activity, such as walking, is assigned to the wrong activity level).

Table 7-1. Summary of the Inhalation Rate Estimates for Selected Age Groups

Breathing Rate, m3/day (CV)a

Age

Male

Female

Male and Female

Birth to 6 months

NA

NA

2.1 (0.27)

7 months to 4 years

9.7 (0.28)

8.8 (0.27)

9.2 (0.28)

5 to 11 years

15.1 (0.22)

14.0 (0.21)

14.5 (0.22)

12 to 19 years

17.7 (0.23)

14.0 (0.2)

15.8 (0.25)

a Coefficient of variation
NA = Data not available
Source: Allan and Richardson (1998).

Unfortunately, this study does not easily lend itself to a reanalysis that would support
recommendations for the new age bins because the authors developed and organized their supporting
data around the age bins defined above. Another problem with their approach is the lack of
inhalation rate data to cover the broad range of activities that are associated with individuals in

7-2


-------
different age groups. So, even ifreanalysis was attempted, the fundamental challenge would remain;
that is, properly linking activities recorded in time-activity studies with the applicable breathing rate
distributions for each age group. An interesting observation that Allan and Richardson make is that
the inhalation rates are particularly sensitive to the minute volumes associated with the lowest
activity level (i.e., resting). They note that time-activity studies show that most people spend a
significant amount of time conducting activities requiring low expenditures of metabolic energy, and
hence the breathing rates selected to represent them will have a significant influence on estimates
of daily breathing rates.

7.2.2 Metabolically Based Inhalation Rates

Layton (1993) presented three different approaches for determining inhalation rates that each
rely on estimates of energy expenditure or food energy intake. The basic equation used to estimate
inhalation in the study was:

VE = E x H x VQ x F

where:

VE = ventilation rate, m3/day;

E = energy expenditure or intake, kcal/day;

H = volume of oxygen at standard temperature and pressure, dry air (STPD)
consumed in the production of 1 kcal of energy expended, L/kcal, (equal to
0.21 L 02/kcal);

VQ = ventilatory equivalent, ratio of the minute volume at body temperature •
ambient pressure • with air saturated with water vapor (BTPS) to the oxygen
uptake rate, unitless; and

F = conversion factor, 0.001 m3/L.

The two methods for estimating inhalation rates based on energy expenditure—one using
activity patterns and the other a multiplier of basal metabolic rate—cannot directly support the
proposed age bins for chronic inhalation rates. Specifically, the activity-based approach did not
cover children, and the basal metabolic rate (BMR) multiplier approach has a very limited data set
upon which to determine applicable multipliers for different age groups and genders. Nevertheless,
the BMR multiplier approach for determining short-term inhalation rates associated with graded
levels of physical activity is still applicable for children.

7-3


-------
The inhalation estimates for children based on food energy intakes, however, can be updated
to reflect more recent dietary surveys of children in the United States as well as more realistic
estimates of the ventilatory equivalent for children. The ventilatory equivalent, which is the ratio
of minute volume to oxygen consumption, is used to relate oxygen uptake to breathing rate. In
Layton (1993) this parameter was mainly estimated using adult data, and consequently updated
values would provide more realistic estimates of children's breathing rates.

7.2.3 Inhalation Rates Determined from Heart Rate Measurements and Activity Data

Spier et al. (1992) and Linn et al. (1992) implemented a protocol for determining inhalation
rates that relied on the use of personal heart rate monitors to record diurnal changes in heart rate for
subjects and individual calibration equations that related heart rate with inhalation rate. Relevant
cohorts in these studies included elementary age and high school students. The advantage of this
methodology is that a real-time physiological measure of activity is monitored that can be related to
specific activities, locations, and times of day. Unfortunately, the test cohorts involved in these
studies were small, and the age groupings do not directly relate to the proposed age groups.

7.3 STUDIES SELECTED FOR THE ANALYSIS TO OBTAIN NEW

RECOMMENDATIONS

The metabolically based methods described by Layton (1993) for determining age-dependent
inhalation rates can serve as the foundation for developing new recommendations for age-specific
inhalation rates. Accordingly, the sections that follow review more recent information that addresses
data gaps pertaining to the calculation ofbreathing rates for children of different ages. The principal
emphasis is the development of recommendations for determining daily breathing rates using food-
energy intakes to represent daily energy expenditures.

7.3.1 Food Energy Intakes for Children

The food energy intake data presented in Layton (1993) were based on the 1977-78 National
Food Consumption Survey (NFCS) conducted by the U.S. Department of Agriculture (USD A, 1984).
Since then the USDAhas conducted additional dietary studies of the U.S. population. For example,
the Continuing Survey of Food Intakes by Individuals (CSFII) included a number of surveys targeted
at different population groups. Of particular interest here are the 1994-96 and 1998 CSFII studies
(USDA, 1999). The 1998 survey focused on food and nutrient intakes for5,559 children from birth
to 9 years of age. This particular study was prompted by the Food Quality Protection Act of 1996,
which required additional data on the dietary habits of children in order to improve EPA's estimates

7-4


-------
of childhood exposures to pesticides in diets. A total of 9,812 children were sampled, including the
4,253 children of the same age in the earlier 1994-96 CSFII. An important attribute of the CSFII
studies is that they are probability samples of the U.S. population, so they are specifically designed
to be representative of children's dietary practices across the country.

Table 7-2 compares the food energy intakes for the combined 1994-96 and 1998 CSFII
studies and the 1977-78 NFCS for overlapping age groups from birth to 5 years of age. The food
energy intakes are approximately 10 percent higher in the newer dietary surveys, which means that
the associated breathing rates would also be proportionately higher. Males in the 6 to 11 age group
sampled in the CSFII studies consumed 2,050 kcal/day of food energy, whereas females consumed
1,825 kcal/day, or nearly 90 percent of the male value. The distinction between male and female
food energy intakes as a function of age should be evaluated further to determine at what age the two
genders should be considered separately for the purposes of determining inhalation rates. The food
surveys, for example, have previously used ages between 5 and 8 years as cutoffs when presenting
gender-specific results.

Table 7-2. Comparison of Food Energy Intakes for Children Under 5 Years of Age
Sampled in the 1994-96 and 1998 CSFII and the 1977-78 NFCS



CSFII Surveys

NFCS Survey



Age Bins (years)

kcal/day (number)

Ratio, CSFII/NFCS

<1

856(1,126)

793 (421)

1.08

1-2

1,330 (2,118)

1,209 (1,035)

1.10

3-5

1,658 (4,574)

1,466 (1,719)

1.13

The other applicable age bin presented in a summary table of CSFII results consisted of ages
12 through 19 years, which is not consistent with the proposed age bins of 11 to 15 years and 16 to
17 years. To determine food energy intakes for those particular cohorts will require statistical
analyses of the actual CSFII databases. Moreover, such analyses should also address the other age
groups as well in order to determine the variances of the results. Until additional analyses can be
completed on the CSFII data, Table 7-3 presents estimates of food energy intakes for the two cohorts
using data in the 1977-78 NFCS.

7-5


-------
Table 7-3. Summary of Energy Expenditures for Children and Adolescents (kcal/day)

Age Bins (years)

Male

Female

Male and Female

3-5a
6-10b
11 -15C
16-17d

NA
2,026
2,700
3,100

NA
1,796
2,200
2,100

1,658

a From the 1994-96, 1998 CSFII studies.

b Average values of age bins 6-9 and 6-11 years given in the summary table of the CSFII study.
c Estimated as the product of the value of the age bin 12-14 years in the 1977-78 NFCS and the factor 1.2 to
account for underreporting of food intakes in the NFCS (see Layton, 1993). Results are rounded to two
significant figures.

d Estimated as the product of the value of the age bins 15-18 years in the 1977-78 NFCS and the factor 1.2
to account for underreporting of food intakes in the NFCS (see Layton, 1993). Results are rounded to two
significant figures.

7.3.2 Ventilatory Equivalents for Children

The ventilatory equivalent (VQ) is basically a measure of breathing efficiency, with lower
values representing higher efficiencies—that is, the lungs require less ventilation per unit of oxygen
consumed in generating the metabolic energy necessary for sustaining a given level of physical
activity. As the respiratory system of a child matures, the ventilatory equivalent gradually decreases
until it reaches a minimum value sometime in adolescence. Layton (1993) used a VQ value of 27 to
represent all age groups, based on a literature review of VQ data for children and a statistical analysis
of values in the open literature for adults. However, additional data have since been identified that
can be used to define VQ values that are more applicable to children.

Adams (1993), for example, conducted a series of respiratory measurements on children
(6-12.9 years of age), adolescents (13-18.9 years of age), adults (19-59.9 years of age), and seniors
(older than 60 years of age). In addition to these groups, a pilot group consisting of 12 young
children ages 3 to 5 years participated in a limited set of respiratory measurements. The average VQ
value of the 6- to 12.9-year-old group was 32.3 for a series of 6 respiratory experiments in which 31
to 40 children (including both males and females) were walking on a treadmill at different velocities.
No significant differences between genders were observed for ventilation and oxygen uptake rates.
In contrast, the VQ values for the pilot group averaged 38.9 for three different walking velocities
(i.e., 1.5, 1.88, and 2.24 mph; n= 12, n = 11, andn= 11). Adams (1993) cited the work of Astrand
(1952), who also found that young children in this age cohort have higher VQ values than older

7-6


-------
children or early adolescents. Indeed, the average VQ for male adolescents in the Adams study,
walking at two different velocities (i.e., 2.5 and 3.3 mph; n = 20 and n = 11), was 24.8, compared
with a mean value of 28.0 for female adolescents at two walking velocities (i.e., 2.5 and 3.0 mph;
n = 20 and n = 14).

The higher VQ values for the preadolescents are consistent with the results of other studies.
Wilmore and Sigerseth (1967) measured ventilation rates and oxygen uptake rates for 62 females
ages 7 to 13 years who were using a bicycle ergometer. The mean VQ value for their measurements
was 31.4. More recently, Rowland and Cunningham (1997) conducted a longitudinal 5-year study
of changes in respiratory responses in a group of 20 children consisting of 11 girls and 9 boys, with
an average age of 9.2 years at the beginning of the study (range of 7.9 to 10.3 years). For the boys,
the average VQ in the first year of the study (for submaximal oxygen uptake) was 31.4 (standard
deviation = 5.10), declining to 26.34 (2.66) in the fifth year. The average VQ for girls in the first year
was 32.3 (2.8), compared with 29.0 (2.34) in the final year. Differences between genders were not
statistically significant. In another study, Armstrong et al. (1997) reported the results of respiratory
measurements made on 101 boys and 76 girls with an average age of 11.1 years. At an exercise rate
of 70 to 75 percent of peak aerobic capacity, the average VQ value for boys was 23.3 (standard
deviation = 2.9, n = 63), while for girls the average VQ was 24.1 (2.4). At 80-85 percent of aerobic
capacity, the mean VQ values increased slightly to 25.0 (Standard Deviation = 2.7, n = 58) for boys
and 24.9 (Standard Deviation = 3.3, n = 42) for girls.

Table 7-4 summarizes the "best" estimates for the different age bins based on available
literature values. Unfortunately, little information is available on the respiratory characteristics of
children younger than 3 years of age. More research clearly needs to be carried out on this particular
age group. The VQ values selected reflect the increasing ventilation efficiency (i.e., demonstrated
by decreasing VQ values) as children mature. The higher VQ values for the younger cohorts
somewhat compensate for the lower energy intakes/expenditures for those groups compared with the
older children. Additional longitudinal studies would provide valuable information on how VQ
values decrease with age. Of special interest is the age at which ventilation efficiency reaches the
adult level. The coefficients of variation (CV) for the VQ values given in Rowland and Cunningham
(1997) and Armstrong et al. (1997) are about 10 percent, so the VQ estimates given in Table 7-4 are
expected to have similar CVs.

7-7


-------
Table 7-4. Summary of Ventilatory Equivalents for Children and Adolescents

Age Bins (years)	Male

3-5	NAa

6-10	NA
11 — 15C 25
16 — 17c 25

Female	Male and Female

NA	39b

NA	32

28	27

28	27

a NA ~ data not available.

b Provisional value based on the pilot study by Adams (1993).
c Gender differences for this age bin may not be statistically significant.

7.4 RECOMMENDATIONS FOR PROPOSED AGE BINS

Recommended daily inhalation rates for children and adolescents are presented in Table 7-5.
The values are calculated as the product of the energy intake/expenditure values given in Table 7-3,
the VQ values recommended in Table 7-4, the parameter H (i.e., 0.21 L 02/kcal), and the conversion
factor 0.001 m3/L. The resulting estimates indicate that the inhalation rates for the younger children
represented by the 6- to 10-year-old age group are comparable to the older age groups. This occurs
because the increase in VQ values as age decreases more than compensates for the reduction in
energy intake/expenditure. For the age groups that are 6 years and above, the variability in the
inhalation rates is about a factor of 1.3, based on the variability in energy expenditure [(CV -= 0.2 to
0.3) and VQ data (expenditure CV » 0.1)]. For comparison, inhalation estimates from the activity-
based approach of Allan and Richardson (given in Table 7-1) have CV values ranging from 0.2 to
0.28, which represent similar levels of variability.

A principal source of uncertainty involves the inhalation estimates for children under 5 years
of age. The inhalation rate for the 3- to 5-year-old bin clearly needs to be verified as it was based
on an average VQ value determined from a limited set of measurements made by Adams (1993)
using a pilot study group. No estimates are provided for the under 3 age groups because of a lack of
respiratory data in the literature. Finally, Table 7-6 provides an overview of the metrics used to
gauge the reliability of the recommended values.

7-8


-------
Table 7-5. Daily Inhalation Rates Estimated for Children and Adolescents" (m3/day)

Age Bins (years)

Male

Female

Male and Female

3 - 5a
6 - 10b
11 - 15c
16 - 17d

14

14
14
16

12

13
12

a Values calculated as the product of the ventilatory equivalents in Table 7.3, the energy expenditure rates
in Table 7-4, the parameter H (0.21 L/kcal), and the factor 0.001 m3/L. Results are presented to two
significant figures.

b Average values of age bin 6-9 and 6-11 years given in the summary table of the CSFII study.
c Estimated as the product of the value for the age bin 12-14 years in the 1977-78 NFCS and the factor 1.2
to account for underreporting of food intakes in the NFCS (see Layton, 1993). Results are rounded to two
significant figures.

d Estimated as the product of the value for the age bin 15-18 years in the 1977-78 NFCS and the factor 1.2
to account for underreporting of food intakes in the NFCS (see Layton, 1993). Results are rounded to two
significant figures.

7.5 RECOMMENDATIONS FOR FURTHER ANALYSIS AND RESEARCH NEEDS

There are basically two lines of research needed to improve inhalation estimates of children,
especially those under 5 years of age for which little data exist to make reliable estimates. First,
additional studies are needed to analyze data on food-energy intakes as well as energy expenditures.
For example, statistical analyses of the CSFII dietary surveys can provide valuable information on
energy intakes for age groups considered in the Child-Specific Exposure Factors Handbook as well
as the more refined age bins addressed in this assessment. The dietary data can be used to develop
estimates of chronic inhalation rates, but to derive diurnal inhalation estimates, time-activity data
for children would have to be obtained and relevant breathing rates assigned, as done in the factorial
approach by Allan and Richardson (1998). However, improved estimates could also be developed
using an energy expenditure-based approach originally described by Layton (1993), but improved
upon by McCurdy (2000) of the U.S. EPA National Exposure Research Laboratory. Specifically,
inhalation is calculated in the same manner as Equation 1, except that E is estimated using a factorial
approach in which an individual's activities are assigned energy expenditure values based on a
multiplier of basal metabolic rate (termed a MET). The total daily value is equal to the sum of
calculated energy expenditures. This approach was enhanced by McCurdy to reflect the energy
expenditures of hundreds of different activities associated with activity patterns of children and
adults characterized by various investigators. The core component of the improved methodology is

7-9


-------
the Consolidated Human Activity Database (CHAD), which includes 16,900 person-days of diurnal
activities. But perhaps more importantly, CHAD includes MET values for a given activity that can
be explicitly incorporated in a Monte Carlo sampling procedure. Values oftheBMR, H, and VQ can
also be treated stochastically. As a quality assurance check, daily energy expenditures computed
using CHAD can also be compared with food-energy intakes obtained from the USD A food-
consumption surveys for children in the different age bins.

One critical area of research for improving estimates of inhalation rates for children under
5 years of age would be to develop expanded data sets of VQ measurements on children from the
applied physiology literature together with new, direct measurements. Reanalysis of raw data from
selected studies should be attempted first. For example, age-dependent distributions ofVQ could be
obtained from a reanalysis of the raw measurements developed in the study by Adams (1993).
Results of such analyses could be used to design experimental studies targeting children in selected
age bins. Another related issue concerns the nature of the age-dependent decline in VQ as well as
related gender differences (e.g., at what age should males and females be treated separately with
respect to VQ?). In addition, Linn et al. (1991) noted that asthmatic children in their study had higher
inhalation rates. Given those children's enhanced susceptibility to ozone, this also should be an area
of special research.

In summary, until additional data on VQ values are obtained for children ages 5 years and
younger, estimates of inhalation rates for this particular age cohort will have potentially significant
uncertainties. The existing recommendations given in the Child-Specific Exposure Factors
Handbook could underestimate actual inhalation rates by 50 percent or more, depending on the
results of additional measurements of VQ values as well as the development of improved estimates
of age-specific energy intakes and expenditures.

7-10


-------
Table 7-6. Confidence in Recommendations for Inhalation Rates

Rating (High, Medium, Low)

Age

Considerations

<1 Month

1-2 Mos

3-5 Mos

6-11 Mos

1 -2 Yrs

3-5 Yrs

6-10 Yrs

11-15 Yrs

16-17 Yrs

• Level of peer review

N/A

N/A

N/A

N/A

N/A

Low

Medium

Medium

Medium

• Accessibility

N/A

N/A

N/A

N/A

N/A

Low

Medium

Medium

Medium

• Reproducibility

N/A

N/A

N/A

N/A

N/A

Medium

High

High

High

• Focus on factor of interest

N/A

N/A

N/A

N/A

N/A

Medium

Medium

Medium

Medium

• Data pertinent to U.S.

N/A

N/A

N/A

N/A

N/A

Medium

Medium
to high

Medium
to high

Medium
to high

• Primary data

N/A

N/A

N/A

N/A

N/A

Medium

Medium

Medium

Medium

• Currency

N/A

N/A

N/A

N/A

N/A

Medium

High to
medium

High to
medium

High to
medium

• Adequacy of data collection
period

N/A

N/A

N/A

N/A

N/A

Low

Medium

Medium

Medium

• Validity of approach

N/A

N/A

N/A

N/A

N/A

High

High

High

High

• Representativeness of the
population

N/A

N/A

N/A

N/A

N/A

Low

Medium

Medium

Medium

• Characterization of variability
in the population

N/A

N/A

N/A

N/A

N/A

Low

Medium

Medium

Medium

• Lack of bias in study design

N/A

N/A

N/A

N/A

N/A

Low

Medium

Medium

Medium

• Measurement error

N/A

N/A

N/A

N/A

N/A

Low

Medium

Medium

Medium

Overall Rating

N/A

N/A

N/A

N/A

N/A

Low

Medium
to high

Medium
to high

Medium
to high

NA = Not applicable; data were not found for these age bins.

7-11


-------
7.6 REFERENCES

Adams, W.C. (1993) Measurement of breathing rate and volume in routinely performed daily
activities, final report. California Air Resources Board (CARB) Contract No. A033-205.
June 1993.

Allan, M.; Richardson, G.M. (1998) Probability density functions describing 24-hour inhalation rates
for use in human health risk assessments. Human and Ecological Risk Assessment 4:379-
408.

Armstrong, N.; Kirby, J.B.; McManus, A.M.; Welsman, J.R. (1997) Prepubescents' ventilatory
responses to exercise with references to sex and body size. Chest 112:1554-1560.

Astrand, P-O. (1952) Experimental studies of physical work capacity in relation to sex and age,
Copenhagen: Ejnar Munksgaard.

Layton, D.W. (1993) Metabolically consistent breathing rates for use in dose assessments. Health
Physics 64:23-36.

Linn, W.S.; Shamoo, D.A.; Hackney, J.D. (1992) Documentation of activity patterns in "high-risk"
groups exposed to ozone in the Los Angeles area. In: Proceedings of the second
EPA/AWMA conference on tropospheric ozone, Atlanta, Nov. 1991, Pittsburgh, PA: Air
and Waste Management Association; pp. 701-712.

McCurdy, T. (2000) Conceptual basis for multi-route intake dose modeling using an energy
expenditure approach. Journal of Exposure Analysis and Environmental Epidemiology
10:86-97.

Rowland, T.W.; Cunningham, L.N. (1997) Development of ventilatory responses in exercise in
normal white children. Chest 111:327-332.

Spier, C.E.; Little, D.E.; Trim, S.C.; Johnson, T.R.; Linn, W.S.; Hackney, J.D. (1992) Activity
patterns in elementary and high school students exposed to oxidant pollution. Journal of
Exposure Analysis and Environmental Epidemiology 2:277-293.

USDA (1984) Nutrient intakes: individuals in the United States, 1977-1978; Nationwide Food
Consumption Survey 1987-88, USDA, Human Nutrition Information Service, Washington,
DC; Report no. 1-2.

USDA (1999) Food and nutrient intakes by children, 1994-96,1998, Table Set 17. U.S. Department
of Agriculture, Agricultural Research Service, Beltsville, MD. Available at

http://www.barcMsda.gov/bhnrc/foodsurveyAiome.htm.

U.S. EPA (2001) Child-specific exposure factors handbook. Report prepared for U.S. EPA, Office
of Research and Development, National Center for Environmental Assessment, by Versar,
Inc.

Wilmore, J.H.; Sigerseth, P.O. (1967) Physical work capacity of young girls, 7-13 years of age.
Journal of Applied Physiology 22:923-928.

7-12


-------
8.0

EXPOSURE FACTORS FOR THE DERMAL ROUTE

8.1 INTRODUCTION

Dermal exposure is estimated, in part, by the amount of body surface area available for
contact with contaminated media. The amount of body surface area exposed during an event is
influenced by age-specific behavioral factors. For children, such factors include playing and
crawling on contaminated surfaces, and the amount of clothing worn during play activities. Surface
area of the skin is determined via direct measurement or regression models that consider the
dependence of surface area on such other body dimensions as height and weight. The principles
upon which the regression models are based are that body density and shape are roughly the same,
and the relationship of the surface area to any dimension may be represented by the curve of central
tendency of their plotted values or by the algebraic expression for the curve (U.S. EPA, 1997). The
Child-Specific Exposure Factors Handbook (U.S. EPA, 2001) described various measurement
techniques and reviewed pertinent surface area studies as a basis for recommending body surface
areas for children that are representative of the sub-population under consideration (i.e., age and sex-
dependent).

For scenarios that involve contact with contaminated soil, dermal exposure is a function of
how much soil adheres to the skin during a specific event. A number of studies were described in
Dermal Exposure Assessment: Principles and Applications (U.S. EPA, 1992). The Exposure
Factors Handbook (U.S. EPA, 1997) and the Child-Specific Exposure Factors Handbook (U .S. EPA,
2001) evaluated more recent studies for use as exposure factors.

The U.S. EPA is currently interested in developing exposure factors for the following
proposed age bins: <1 month, 1 -2 months, 3-5 months, 6-11 months, 1-2 years, 3-5 years, 6-10 years,
11-15 years, and 16-17 years. The purpose of this section is to evaluate the studies considered by
U.S. EPA in the Child-Specific Exposure Factors Handbook to determine if, and how, the existing
data can be presented within the proposed age bins. Additionally, more recent studies that might
supplement the existing data will be discussed, and recommendations will be made, where possible,
regarding the analytical approach to redistribute the data into the proposed age bins.

8-1


-------
8.2

EVALUATION OF EXISTING DATA

8.2.1 Surface Area Studies

Direct surface area measurement data presented in Gehan and George (1970) were analyzed
in EPA (1985). In their study, Gehan and George used all of the records for postnatal subjects that
had been reported in a comprehensive study by Boyd (1935), and for which direct measurements of
surface area, height, and weight were reported. The Boyd data used in the Gehan and George study
consisted of a total of 401, including observations of a relatively high number of Japanese and
Chinese subjects, and some individuals with unusual body types (U.S. EPA, 1985). Of the 401
subjects, 229 were less than 5 years of age, 42 were between 5 and 20 years of age, and 130 were
greater than 20 years of age (Gehan and George, 1970). The data were analyzed using a least squares
method of multiple regression to develop revised constants for abi-exponential surface area model
proposed in an earlier study by DuBois and DuBois (1916). In that study, DuBois and DuBois
developed the model based on a sample of only nine individuals for whom surface area was already
measured (U.S. EPA, 1997). Additionally, the subjects were predominantly male, two were
deformed, and the only child was described as sickly. The constants developed by Gehan and
George (1970) explain more than 99 percent of the variations in surface area among the 401
measured individuals.

Although the revised model proposed by Gehan and George (1970) was determined by EPA
to be the best choice for estimating total body surface area, that study gave insufficient information
to estimate the standard error about the regression (U.S. EPA, 1985). Subsequently, EPA used the
least squares to reanalyze the direct measurement data for the 401 individuals reported in Gehan and
George to determine the standard errors of the individual constants as well as the standard error
about the regression. The resulting standard error about the regression was 0.00374. EPA was
satisfied that the revised model explained more than 99 percent of the total variation in surface area
among the observations and is identical to two significant figures with the model developed by
Gehan and George (U.S. EPA, 1985).

For the purpose of estimating percentiles (and their standard errors) for total surface area of
male and female children, EPA used height and weight data gathered in the second National Health
and Nutrition Examination Survey (NHANESII). Those data were analyzed using QNTLS, an SAS
macro developed by Rochon and Kalsbeek (1983) that performs variance estimation of multistage
sample survey data using the jackknife repeated replicate approach (JRRA) (U.S. EPA, 1985).
JRRA uses resampling to calculate estimates of linear statistics (e.g., means, distributions, totals)
and more complex nonlinear measures such as covariances, correlations, and the slope and intercept

8-2


-------
from a simple (weighted) linear regression in addition to their standard errors (Rochon and Kalsbeek,
1983). The technique used in JRRA, resampling with replacement from the observed data, is
necessary because resampling observations of the observed data mimics the process of sampling
observations of the population.

The total surface area percentile estimates for male and female children were calculated using
the Gehan and George bi-exponential model and the height and weight results of the QNTLS
analysis. However, because the NHANESII height data were not available for children younger than
2 years of age, total surface area was not estimated for that age group.

The lack of NHANES II height data precludes estimation of recommended total surface area
values for children less than 1 month old, 1 to 2 months, 3 to 5 months, 6 to 11 months, or 1 to 2
years of age. However, values for children 3 to 5, 6 to 10, 11 to 15, and 16 to 17 years of age can
be estimated by compiling subsets of the NHANES II data that correspond to the recommended age
bins. Subsequently, the subsets can be analyzed using QNTLS or any other statistical software
package that is able to estimate the variance of multistage sample survey data using JRRA. In view
of the large data gap for children under 2 years of age, additional data are needed to support the age
bin recommendations for that age group.

Another precaution regarding use of the total body surface area values derived from the
NHANES II data is the currency of those data. During the two decades since NHANES II data were
collected, there has been an upward shift in the prevalence of overweight among children and
adolescents. The U.S. Department of Health and Human Services, Centers for Disease Control and
Prevention reported that from the 1960s to 1980, overweight in children and adolescents was
relatively stable. However, from NHANES II (1976-80) to NHANES III (1988-94), the prevalence
of overweight nearly doubled among those age groups. During that time period, the prevalence of
overweight among children ages 6 to 11 years increased from 7 percent to 11 percent, and
adolescents ages 12 to 19 years increased from 5 percent to 11 percent (U.S. CDC, 2001). A review
of NHANES 1999 data suggests that overweight in children and adolescents may be increasing to
even higher levels than in 1994 (Figure 8-1). Subsequently, age-specific total surface area values
based on NHANES II data may not be representative of the current population.

8-3


-------
Figure 1. Prevalence of overweight among
children and adolescents ages 6-19 years

15

10
3

0

NOTES Excludes pregnant women starti^ witfi 1971-74 P-Egnancy status not available to"
1i-63-G5arid 1 QBB-^O Data for I9B3-G5 arc for Chidrcn B-11 y#3rs J age. data for 1D66-70art
for ad olcsccnt; 12-17 years of a:c .nun 12-19 vcar;

SOURCE COOKCHS NHES and NHANES. '

Figure 8-1. Prevalence of Overweight Among Children and
Adolescents Ages 6-19 Years

Phillips et al. (1993) calculated surface area to body weight ratios for three age groups of the
population. The study evaluated the 401 individual observations that were reported by Boyd (1935)
and used by Gehan and George (1970) to develop their bi-exponential surface area model. Phillips
et al. used Pearson's product moment correlation analysis to compare surface area data with
corresponding body weights. Subsequently, surface area/body weight (SA/BW) ratios were
calculated for each of the individuals in the data set, and summary statistics were calculated for the
entire data set. Data were sorted by sex and age to evaluate the effect they had on SA/BW
distributions. The resulting SA/BW were plotted against the corresponding ages of the individuals
in the data set, and visual inspection was used to divide the SA/BW curve into three segments at the
ages where obvious changes had occurred. The age segments corresponded approximately to the
ages of infants (0 to 2 years), children (2.1 to 17.9 years), and adults (>18 years). Comparisons of
the mean SA/BW ratio differences for males and females were not found to be statistically different
at the 0.05 level of confidence. However, a strong negative correlation was observed between age
and SA/BW for all ages combined, and for infants and children when these age groups were tested
separately.

Age in years

¦ 6-11 B12-1

Q

11
¦

11



13



7

&

4

Mm 1











1963-70 1971-74 1976-00 1980-94 1999

8-4


-------
The results of the Phillips et al. (1993) study can be presented within the age bins proposed
by EPA. However, the data upon which the SA/BW ratios were derived are over six decades old and
may not be representative of the characteristics of the current population, based on the discussion
above regarding the increased prevalence of obesity among children and adolescents. Additionally,
the conclusions of the Phillips et al. study could not be modified using NHANES data, because
height measurements are not available for children less than 2 years of age in those surveys.

8.2.2 Soil Adherence Studies

In the past, the U.S. EPA recommended one value to represent dermal soil adherence to all
body parts, regardless of the soil type or conditions, or type of activity that leads to soil contact. The
work of Kissel et al. (1996a, 1996b, 1998) and Holmes et al. (1999) has attempted to determine the
magnitude of dermal soil adherence. Kissel et al. (1996a) conducted sieved and unsieved soil
adherence experiments for five soil types. Kissel et al. (1996b, 1998) and Holmes et al. (1999)
estimated soil adherence associated with various indoor and outdoor activities. The results of those
studies showed that soil adherence generally could be directly correlated with moisture content,
inversely correlated with particle size, and independent of clay content or organic carbon.
Additionally, the rate of soil adherence is higher for hands than for other parts of the body.

These studies have made great strides in our understanding of the magnitude of soil
adherence under different soil conditions and human activities, but all of the aforementioned studies
were based on a small number of participants, and, for at least one study (Kissel et al., 1998), a
relatively short activity duration. Additionally, the activity settings and children selected for the
study may not be representative of the U.S. population. With regard to age bins, soil adherence is
more activity- than age-specific, therefore, the values recommended in the Child-Specific Exposure
Factors Handbook (U.S. EPA, 2001) can be used for any age group, depending on the activity
considered.

8.3 STUDIES SELECTED FOR THE ANALYSIS TO OBTAIN NEW

RECOMMENDATIONS

EPA did a thorough job of reviewing the peer-reviewed literature to derive reasonable values
for total body surface area and dermal soil adherence values for the Exposure Factors IIandbook
(U.S. EPA, 1997) and the Child-Specific Exposure Factors Handbook (U.S. EPA, 2001). For this
present effort, a thorough search of peer-reviewed literature back to 1997 was conducted; however,
no new studies were identified that have been performed relative to those factors since the
development of the EPA documents (1997,2001). Although one expert expressed optimism during

8-5


-------
the technical workshop that the NHANES IE body weight data might supply new insight into surface
area in children less than 2 years of age, the NHANES IE format is apparently the same as that of
NHANES II. The reason that total surface area could not be estimated from the NHANES II data
set is because height information, which is equally important to calculate surface area, was not
included. That same information apparently is missing from the NHANES III data. Consequently,
the new data should not provide any new insight into total surface area within the context of the age
bins recommended by EPA.

8.4 RECOMMENDATIONS FOR PROPOSED AGE BINS

Presently, no recommendations can be made with respect to the proposed age-bin-specific
surface area values. As indicated previously, the lack of NHANES II height data precludes
estimation of recommended total surface area values for children less than 1 month of age, 1 to 2
months, 3 to 5 months, 6 to 11 months, or 1 to 2 years of age. With regard to children in the age
groups of 3 to 5 years, 6 to 10 years, 11 to 15 years, and 16 to 17 years, estimation of surface area
values would require compilation of the NHANES II data into subsets that correspond to the
recommended age bins. Subsequently, the subsets would need to be analyzed using a routine to
estimate the variance of multistage sample survey data using JRRA. Unfortunately, this analysis
cannot be performed within the scope of this document for various reasons. First, the analysis
requires that the NHANES data be in the same format as the original evaluation, conducted for
Development of Statistical Distributions or Ranges of Standard Factors Used in Exposure
Assessments (U.S. EPA, 1985). Second, the JRRA for variance estimation, described by Rochon and
Kalsbeek (198 3), has some weaknesses. Although the procedure is conceptually straightforward and
moderately easy to use, error checking is limited. The authors cautioned that the JRRA macros
assume that information entered through the parameter macros is accurate, and consequently, users
may be left to the mercy of the PROC MATRIX error message (Rochon and Kalsbeek, 1983).
Because of this, considerable time may be required to verify the information entered into the
programs. Until a new analysis of an existing dataset or new data are collected, factors selected to
address the proposed age bins will have to be used with caution. A confidence evaluation of the
existing studies used in the current CSEFH is presented in Table 8-1.

As previously stated, soil adherence is more activity- than age-specific. Therefore, the values
recommended in the Child-Specific Exposure Factors Ilandbook can be used for any age group
depending on the activity considered.

8-6


-------
8.5

RECOMMENDATIONS FOR FURTHER ANALYSIS AND RESEARCH NEEDS

Presently, the most important research need is to compile appropriate height and weight data
for children under 2 years of age. Because of their behavior patterns (e.g., playing and crawling on
contaminated surfaces with fewer clothes for protection) and physical factors (i.e., higher surface
area relative to body weight), children in this age group may have potentially higher exposure to
environmental toxicants than other age groups.

With regard to dermal soil adherence, the Child-Specific Exposure Factors Handbook (U.S.
EPA, 2001) has recommended that, because the controlled experiments and field studies conducted
to date were based on specific situations and a limited number of measurements, more-detailed
studies are necessary to determine variation among individuals, the effects of time of activity,
protective clothing, and seasonal factors on dermal soil adherence.

8.6 REFERENCES

Boyd, E. (1935) The growth of the surface area of the human body. Minneapolis, MN: University
of Minnesota Press.

CDC (Centers for Disease Control and Prevention). (2001) Fast stats A to Z. Overweight prevalence.
National Center for Health Statistics. Web page: http://www.cdc.gov/nchs/fastats/overwt. htm.

DuBois, D.; DuBois, E.F. (1916) A formula to estimate the approximate surface area if height and
weight be known. Archives of Internal Medicine 17:863-871.

Gehan, E.; George, G.L. (1970) Estimation of human body surface area from height and weight.
Cancer Chemotherapy Reports 54(4):225-235.

Holmes, K.K. Jr.; Shirai, J.H.; Richter, K.Y.; Kissel, J.C. (1999) Field measurements of dermal
loadings in occupational and recreational activities. Environmental Research, Section A
80:148-157.

Kissel, J.; Richter, K; Duff, R.; Fenske, R. (1996a) Factors affecting soil adherence to skin in hand-
press trials. Bulletin of Environmental Contamination and Toxicology 56:722-728.

Kissel, J.; Richter, K; Fenske, R.. (1996b) Field measurements of dermal soil loading attributable
to various activities: implications for exposure assessment. Risk Analysis 16(1): 116-125.

Kissel, J.C.; Shirai, J.H.; Richter, K.Y.; Fenske, R.A. (1998) Investigation of dermal contact with
soil in controlled trials. Journal of Soil Contamination 7(6):737-752.

8-7


-------
Phillips, L.J.; Fares, R.J.; Schweer, L.G. (1993) Distributions of total skin surface area to body
weight ratios for use in dermal exposure assessments. Journal of Exposure Analysis and
Environmental Epidemiology 3(3):331-338.

Rochon, J.; Kalsbeek, W.D. (1983) Variance estimation from multi-stage sample survey data: the
jackknife repeated replicate approach. Presented at SAS Users Group Conference; January
1983; New Orleans, LA: SAS Institute, Inc.

U.S. EPA (1985) Development of statistical distributions or ranges of standard factors used in
exposure assessments. EPA/600/8-85/010. Available from the National Technical
Information Service, Springfield, VA; PB85-242667.

U.S. EPA (1992) Dermal exposure assessment: principles and applications. Office of Research and
Development, Washington, DC; EPA/600/8-91/01 IB.

U.S. EPA (1997) Exposure factors handbook, vol. I: general factors. Office of Research and
Development, Washington, DC; EPA/600/P-95/002Fa-c.

U.S. EPA (2001) Child-specific exposure factors handbook. Report prepared for U.S. EPA, Office
of Research and Development, National Center for Environmental Assessment, by Versar,
Inc.

Wong, E.Y.; Shirai, J.H.; Garlock, T.J.; Kissel, J.C. (2000) Adult proxy responses to a survey of
children's dermal soil contact activities. Journal of Exposure Analysis and Environmental
Epidemiology 10(6, Pt l):509-517.

8-8


-------
Table 8-1. Evaluation of Existing Surface Area Studies

Rating (High, Medium, Low)

Considerations		

<1 Month 1-2 Mos 3-5 Mos 6-11 Mos 1-2 Yrs 3-5 Yrs 6-10 Yrs 11-15 Yrs 16-17 Yrs

Study Elements

Level of Peer Review

NA

NA

NA

NA

NA

High

High

High

High

Accessibility

NA

NA

NA

NA

NA

Medium

Medium

Medium

Medium

Reproducibility

NA

NA

NA

NA

NA

High

High

High

High

Focus on factor of interest

NA

NA

NA

NA

NA

High

High

High

High

Data pertinent to U.S.

NA

NA

NA

NA

NA

High

High

High

High

Primary data

NA

NA

NA

NA

NA

High

High

High

High

Currency

NA

NA

NA

NA

NA

Low

Low

Low

Low

Adequacy of data collection period

NA

NA

NA

NA

NA

Low

Low

Low

Low

Validity of approach

NA

NA

NA

NA

NA

High

High

High

High

Representativeness of the population

NA

NA

NA

NA

NA

Medium

Medium

Medium

Medium

Characterization of variability in the
population

NA

NA

NA

NA

NA

High

High

High

High

Lack of bias in study design

NA

NA

NA

NA

NA

Medium

Medium

Medium

Medium

Measurement error

NA

NA

NA

NA

NA

High

High

High

High

rail Rating

NA

NA

NA

NA

NA

Medium

Medium

Medium

Medium

8-9


-------
9.0

CHILD-SPECIFIC ACTIVITY PATTERNS

9.1 INTRODUCTION

Each exposure route (dermal, inhalation and ingestion) is influenced by physiological,
behavioral, physical activities, and demographic characteristic (Hubal et al., 2000). For each
exposure route, an algorithm (see Section 1.1) is used to express exposure as a function of (1) the
chemical concentration in the exposure medium, (2) contact rate, (3) rate of transfer of the chemical
from the exposure medium to the portal of entry, and (4) the exposure duration (Hubal et al., 2000).
Exposure duration, addressed as time spent in various microenvironments and in various
microactivities in the home environment, is discussed in this section; other exposure factors
expressed in the algorithms are addressed in other sections of this issue paper and the CSEFH.

Chemical exposures of children are influenced by the types of activities in which they are
engaged as well as the locations of the activities and the level of participation in those activities.
Consequently, exposure to chemicals in the environment can vary among children of similar
developmental stages because of the variability in their behavior. Additionally, seasonal and
geographic differences among children of similar developmental stages influence the variability of
exposure. In any microenvironment (the location that a child occupies at a particular time), exposure
to contaminants is influenced, among other things, both by the activity in which the child is engaged
(e.g., such macroactivities as watching TV, eating, playing games, and crawling on the floor) and,
the detailed actions that occur within the macroactivity (e.g., such microactivities as hand-to-surface
and hand-to-mouth behavior) (Hubal et al., 2000). Macroactivity data are discussed in Section 6,
Non-Dietary Ingestion. This section focuses on macroactivity data only. The Child-Specific
Exposure Factors Handbook (U.S. EPA, 2001) and Hubal et al. (2000) described various
measurement techniques and reviewed pertinent activity studies as a basis for recommending
activity factors for children that are representative of the subpopulations (i.e., age- and sex-
dependent) under consideration.

The U.S. EPA is interested in developing exposure factors for the following age bins: <1
month, 1-2 months, 3-5 months, 6-11 months, 1-2 years, 3-5 years, 6-10 years, 11-15 years, and 16-
17 years. The purpose of this section is to evaluate the studies considered by U.S. EPA in the Child-
Specific Exposure Factors Handbook (U.S. EPA, 2001) to determine if, and how, the exi sting data
can be presented within the proposed age bins. Additionally, more recent studies that might
supplement the existing data will be discussed, and recommendations will be made, where possible,
regarding the analytical approach to redistribute the data into appropriate age bins.

9-1


-------
supplement the existing data will be discussed, and recommendations will be made, where possible,
regarding the analytical approach to redistribute the data into appropriate age bins.

9.2 EVALUATION OF EXISTING DATA

The Child-Specific Exposure Factors Handbook (U. S. EPA, 2001) evaluated several activity
studies in an attempt to derive recommended default values for use in exposure assessments for
children. The most significant studies are described below, including a discussion of their inherent
strengths and weaknesses.

Timmer et al. (1985) - How Children Use Time

Timmer et al. (1985) evaluated children's time use data obtained from a 1981 -82 panel study
of children between the ages of 3 and 17 years. The authors concluded that more time was spent by
girls than boys performing household work and personal care activities, and less time playing sports.
Additionally, children spent most of their free time watching television. Older children spent more
time doing household and market work, studying, and watching television, and less time eating,
sleeping, and playing. The study was limited with respect to the age of the data as well as the time
span upon which the children's time use was based. Specifically, since the data were collected two
decades ago, they may no longer be representative of children today. Additionally, data in the panel
study were collected only during the time of year when children attend school, and therefore do not
provide overall annual estimates of children's time use (U.S. EPA, 2001).

Robinson and Thomas (1991) - Time Spent in Activities, Locations, and

Microenvironments: A California-National Comparison Project

Robinson and Thomas (1991) evaluated and compared children's time use data obtained from
the 1987-88 California Air Resources Board (CARB) study for California residents and the 1985
national study, America's Use of Time. Both studies comprised children 12 years and older. The
authors concluded that, in both studies, males spent more time in work locations, automobiles and
other vehicles, garages, and physical outdoor activities. Females, on the other hand, spent more time
cooking, engaging in other kitchen activities, performing other chores, and shopping (U.S. EPA,
2001). This study was limited with respect to the area in which the data were collected, the age of
the data, and the length of the study. Specifically, since data were collected only for California
residents, the results may not be representative of other areas of the United States. Also, as with the
Timmer et al. (1985) study, because of the time lapse since the data were collected, the results may
no longer be representative of children today.

9-2


-------
Tsang and Klepeis (1996) - National Human Activity Pattern Survey (NHAPS)

Tsang and Klepeis (1996) produced the largest and most current human activity pattern
survey available. The survey gathered data from 9,386 study participants of all ages regarding the
duration and frequency of selected activities and the time spent in selected microenvironments. The
NHAPS data set is representative of the U.S. population and has been adjusted to be balanced
geographically, seasonally, and temporally (U.S. EPA, 2001). It is also race-specific and is
representative of all ages and gender. However, the data are limited with respect to children ages
1 to 17 as well as time spent in certain activities. Specifically, for children ages 1 to 17, data for
most activities are sparse. Additionally, for selected activities, means cannot be calculated for time
spent at the high end of the distribution for those activities. For example, for swimming events,
Tsang and Klepeis (1996) reported the 90th percentile value of 180 minutes per swimming event
(based on one event/month) and 99th percentile as 181 minutes. The 181 -minute value was used to
indicate that more than 180 minutes were spent in the activity, thereby masking actual time spent.

Hubal et al. (2000) - Children's Exposure Assessment: A Review of Factors Influencing

Children's Exposure and the Data Available to Characterize and Assess That Exposure

Hubal et al. (2000) reviewed the U.S. EPA National Exposure Research Laboratory's
Consolidated Human Activity Database (CHAD), a storehouse of data from several studies on
human activities. Although CHAD contains 4,300 person-days of information for children younger
than 18 years and 3,009 person-days of microactivity data for 2,640 children younger than 12 years,
the database is limited with respect to activity codes. Specifically, the authors noted that although
CHAD contains approximately 140 activity codes and 110 location codes, the data generally are not
available for all activity locations for any single respondent (U.S. EPA, 2001). Furthermore, many
of the activity codes are broadly defined, thereby ignoring many child-oriented behaviors. This
author queried CHAD for male and female children ranging in age from 1 to 17 years and
experienced the same problem with the activity codes as Hubal et al. (2000). Moreover, for some
activity codes, the activity maybe spelled differently on different records or is described colloquially,
thereby making compilation of those data subgroups difficult or impossible to analyze.
Consequently, the descriptions of the activities in the database do not really give a clear indication
as to the levels of activity among the children. Nonetheless, Hubal et al. (2000) compiled data from
three studies incorporated into CHAD that contained activity data for children under 12 years of age.
Those studies included the 1990 California "children and youth" recall survey (Wiley et al., 1991),
the 1983 Cincinnati diary study sponsored by the Electric Power Research Institute (Johnson, 1989),
and the "air" and "water" versions of the 1992-1994 National Human Activity Pattern Survey

9-3


-------
(NHAPS) recall survey (Klepeis et al., 1995). The results of the analysis conducted by Hubal et al.
(2000) are presented in Tables 9-1 through 9-3.

Table 9-1. Number of Person-Days/Individuals" for Children in CHAD Database

Age Group

All Studiesb

California

Cincinnati11

NHAPS-Air

NHAPS-Water

0 yr

223/199

104

36/12

39

44

0-6 mo



50

15/5





6-12 mo



54

21/7





1 yr

259/238

97

31/11

64

67

12-18 mo



57







18-24 mo



40







2 yr

317/264

112

81/28

57

67

3 yr

278/242

113

54/18

51

60

4 yr

259/232

91

41/14

64

63

5 yr

254/227

98

40/14

52

64

6 yr

237/199

81

57/19

59

40

7 yr

243/213

85

45/15

57

56

8 yr

259/226

103

49/17

51

55

9 yr

229/195

90

51/17

42

46

10 yr

224/199

105

38/13

39

42

11 yr

227/206

121

32/11

44

30

Total

3009/2640

1200

556/187

619

634

a CHAD = Consolidated Human Activity Database, available on U.S. EPA Intranet.

b The number of person-days of data are the same as the number of individuals for all studies except for the Cincinnati
study. Since up to 3 days of activity pattern data were obtained from each participant in this study, the number of
person-days of data is approximately three times the number of individuals.

Source: Hubal et al. (2000, Table 1).

9-4


-------
Table 9-2. Number of Hours Per Day Children Spend in Various Microenvironments by Age
(Average ± std. dev. percent of children reporting >0 hours in microenvironment)

MICROENVIRONMENT

Age

(years)

Indoors

Outdoors

Indoors

Outdoors



at Home

at Home

at School

at Park

In Vehicle

0

19.6±4.3 (99%)

1,4±1.5 (20%)

3.5±3.7 (2%)

1.6±1.5 (9%)

1.2±1.0 (65%)

1

19.5±4.1 (99)

1,6±1.3 (35)

3.4±3.8 (5)

1.9±2.7 (10)

1.1±0.9 (66)

2

17.8±4.3 (100)

2.0±1.7 (46)

6.2±3.3 (9)

2.0±1.7 (17)

1,2±1.5 (76)

3

18.0±4.2 (100)

2.1±1.8 (48)

5.7±2.8 (14)

1.5±0.9 (17)

1.4±1.9 (73)

4

17.3±4.3 (100)

2.4±1.8 (42)

4.9±3.2 (16)

2.3±1.9 (20)

1,1±0.8 (78)

5

16.3±4.0 (99)

2.5±2.1 (52)

5.4±2.5 (39)

1,6±1.5 (28)

1,3±1.8 (80)

6

16.0±4.2 (98)

2.6±2.2 (48)

5.8±2.2 (34)

2.1±2.4 (32)

1,1±0.8 (79)

7

15.5±3.9 (99)

2.6±2.0 (48)

6.3±1.3 (40)

1,5±1.0 (28)

1.1±1.1 (77)

8

15.6±4.1 (99)

2.1±2.5 (44)

6.2±1.1 (41)

2.2±2.4 (37)

1.3±2.1 (82)

9

15.2±4.3 (99)

2.3±2.8 (49)

6.0±1.5 (39)

1,7±1.5 (34)

1.2±1.2 (76)

10

16.0±4.4 (96)

1,7±1.9 (40)

5.9±1.5 (39)

2.2±2.3 (40)

l.lil.l (82)

11

14.9±4.6 (98)

1.9±2.3 (45)

5.9±1.5 (41)

2.0±1.7 (44)

1.6±1.9 (74)

Source: Hubal et al. (2000, Table 2).

9-5


-------
Table 9-3. Number of Hours Per Day Children Spend Doing Various
Macroactivities While Indoors at Home by Age
(percent of children reporting >0 hours for microenvironment/macroactivity)

MACRO ACTIVITY IN HOME MICROENVIRONMENT

Age

(yr)

Eat

Sleep
or Nap

Shower
or Bathe

Play
Games

Watch TV
or Listen
to Radio

Read,
Write,
Homework

Think,
Relax,
Passive

0

1.9 (96%)

12.6 (99%)

0.4 (44%)

4.3 (29%)

1.1 (9%)

0.4 (4%)

3.3 (62%)

1

1.5 (97)

12.1 (99)

0.5 (56)

3.9 (68)

1.8 (41)

0.6 (19)

2.3 (20)

2

1.3 (92)

11.5 (100)

0.5 (53)

2.5 (59)

2.1 (69)

0.6 (27)

1.4 (18)

3

1.2 (95)

11.3 (99)

0.4 (53)

2.6 (59)

2.6 (81)

0.8 (27)

1.0 (19)

4

1.1 (93)

10.9 (100)

0.5 (52)

2.6 (54)

2.5 (82)

0.7 (31)

1.1 (17)

5

1.1 (95)

10.5 (98)

0.5 (54)

2.0 (49)

2.3 (85)

0.8 (31)

1.2 (19)

6

1.1 (94)

10.4 (98)

0.4 (49)

1.9 (35)

2.3 (82)

0.9 (38)

1.1 (14)

7

1.0 (93)

9.9 (99)

0.4 (56)

2.1 (38)

2.5 (84)

0.9 (40)

0.6 (10)

8

0.9 (91)

10.0 (96)

0.4 (51)

2.0 (35)

2.7 (83)

1.0 (45)

0.7 (7)

9

0.9 (90)

9.7 (96)

0.5 (43)

1.7 (28)

3.1 (83)

1.0 (44)

0.9 (17)

10

1.0 (86)

9.6 (94)

0.4 (43)

1.7 (38)

3.5 (79)

1.5 (47)

0.6 (10)

11

0.9 (89)

9.3 (94)

0.4 (45)

1.9 (27)

3.1 (85)

1.1 (47)

0.6 (10)

Source: Hubal et al. (2000, Table 3).

9.3	STUDIES SELECTED FOR THE ANALYSIS TO OBTAIN NEW
RECOMMENDATIONS

This author used the Medline and Toxline databases to conduct a thorough search of peer-
reviewed literature relative to children's activity factors. The search was limited to those articles that
were published after 1997 to the present date in order to identify any studies not considered in the
Child-Specific Exposure Factors Handbook (U.S. EPA, 2001). Unfortunately, no new pertinent
studies have been performed relative to those factors since the development of those two documents.
Therefore, Hubal et al. (2000) was selected for the analysis to obtain new recommendations.

9.4	RECOMMENDATIONS FOR EACH AGE BIN

Presently, no recommendations can be made with respect to activity factor values for children
less than 1 month old, 1 to 2 months, 3 to 5 months, 6 to 11 months, 11 to 15 years, or 16 to 17

9-6


-------
years. As stated by Hubal et al. (2000), the current database on children's macroactivities is sparse
and data are insufficient to adequately assess exposures to environmental contaminants. However,
the results of the Hubal et al. (2000) evaluation of CHAD data for children less than 12 years of age
(see Tables 9-1 through 9-3) are sufficient to estimate values for time in microenvironments and
participation in certain macroactivities for children in age bins of 1-2 years, 3-5 years, and 6-10
years. Therefore, weighted average values for the microenvironments and macroactivities within
those age bins were estimated based on the averages of the year groups within each of the age bins
and the respective person-days reported by Hubal et al. (2000) for those year groups. The
recommended values for microenvironments and macroactivities for children in age bins of 1-2
years, 3-5 years, and 6-10 years are presented in Tables 9-4 and 9-5, respectively. The level of
confidence for the recommended activity factor values is presented in Table 9-6.

Table 9-4. Estimated Number of Hours Per Day Children Spend in Various Microenvironments by Age Bin

Age Bin





MICROENVIRONMENT





Indoors
at Home

Outdoors
at Home

Indoors
at School

Outdoors
at Park

In Vehicle

<1 mo

NA

NA

NA

NA

NA

1 -2 mo

NA

NA

NA

NA

NA

3-5 mo

NA

NA

NA

NA

NA

6-11 mo

NA

NA

NA

NA

NA

1-2 yr

18.6

1.9

5.5

2.0

1.2

3-5 yr

17.2

2.3

5.4

1.8

1.3

6-10 yr

15.7

2.3

6.0

2.0

1.2

11-15 yr

NA

NA

NA

NA

NA

16-17 yr

NA

NA

NA

NA

NA

NA = Data are insufficient to estimate time spent in microenvironment.
Source: Adapted from Hubal et al. (2000).

9-7


-------
Table 9-5. Estimated Number of Hours Per Day Children Spend Doing
Various Macroactivities While Indoors at Home by Age Bin

MACRO ACTIVITY IN HOME MICROENVIRONMENT

Age Bin

Watch TV	Read,	Think,

Sleep Shower Play or Listen	Write,	Relax,

Eat or Nap or Bathe Games to Radio	Homework	Passive

<1 mo. NA NA NA NA NA	NA	NA

1-2 mo. NA NA NA NA NA	NA	NA

3-5 mo. NA NA NA NA NA	NA	NA

6-11 mo. NA NA NA NA NA	NA	NA

I-2	yr 1.4 11.8 0.5 3.2 2.0	0.6	1.8
3-5 yr 1.1 10.9 0.5 2.5 2.5	0.8	1.1
6-10 yr 1.0 9.9 0.4 1.9 2.8	1.1	0.8

II-15	yr NA NA NA NA NA	NA	NA
16-17 yr NA	NA	NA	NA	NA	NA	NA

NA = Data are insufficient to estimate time spent doing macroactivity.
Source: Adapted from Huban et al. (2000).

9-8


-------
Table 9-6. Confidence for Recommendations for Activity Factors

Rating (High, Medium, Low)

Considerations

Age

<1 Month 1-2 Mos 3-5 Mos 6-11 Mos 1-2 Yrs

3-5 Yrs

6-10 Yrs

11-15 Yrs 16-17 Yrs

Study Elements

• Level of peer review

NA

NA

NA

NA

High

High

High

NA

NA

• Accessibility

NA

NA

NA

NA

Medium

Medium

Medium

NA

NA

• Reproducibility

NA

NA

NA

NA

Medium

Medium

Medium

NA

NA

• Focus on factor of interest

NA

NA

NA

NA

Medium

Medium

Medium

NA

NA

• Data pertinent to U.S.

NA

NA

NA

NA

High

High

High

NA

NA

• Primary data

NA

NA

NA

NA

High

High

High

NA

NA

• Currency

NA

NA

NA

NA

Medium

Medium

Medium

NA

NA

• Adequacy of data collection period

NA

NA

NA

NA

Medium

Medium

Medium

NA

NA

• Validity of approach

NA

NA

NA

NA

Medium

Medium

Medium

NA

NA

• Representativeness of the
population

NA

NA

NA

NA

Medium

Medium

Medium

NA

NA

• Characterization of variability in
the population

NA

NA

NA

NA

Medium

Medium

Medium

NA

NA

• Lack of bias in study design

NA

NA

NA

NA

Medium

Medium

Medium

NA

NA

• Measurement error

NA

NA

NA

NA

Medium

Medium

Medium

NA

NA

Overall Rating

NA

NA

NA

NA

Medium

Medium

Medium

NA

NA

9-9


-------
9.5

RECOMMENDATIONS FOR FURTHER ANALYSIS AND RESEARCH NEEDS

The present state of knowledge regarding children's exposures and activities are inadequate
to assess exposures to environmental contaminants sufficiently. This author supports the
conclusions and recommendations presented by Hubal et al. (2000). Research needs to be conducted
in three specific areas in order to improve the database that is currently available to assess children's
exposures. Methods for monitoring children's activities and exposures need to be improved.
Additionally, physical activity data for children, but especially young children less than 4 years of
age and in age bins 11-15 years and 16-17 years, need to be collected in order to assess exposure by
all routes. In order to accomplish this, population-based data are required to improve the
characterization of children's activities and exposures as a function of age, gender, environmental
setting (residence, school, day care), socioeconomic status, race/ethnicity, location (urban, suburban,
rural), region, and season.

9.6 REFERENCES

Hubal, E.A.; Sheldon, L.S.; Burke, J.M.; McCurdy, T.R.; Berry, M.R.; Rigas, M.L.; Zartarian, V.G.;
Freeman, N.G. (2000) Children's exposure assessment: a review of factors influencing
children's exposure and the data available to characterize and assess that exposure.
Environmental Health Perspectives 108(6):475-486.

Robinson, J.P.; Thomas, J. (1991) Time spent in activities, locations, and microenvironments: a
California-National Comparison Project report. Las Vegas, NV: U.S. EPA, Environmental
Monitoring Systems Laboratory.

Tsang, A.M.; Klepeis, N.E. (1996) Results tables from a detailed analysis of the National Human
Activity Pattern Survey (NHAPS) response. Draft report prepared by Lockheed Martin for
U.S. EPA, Contract No. 68-W6-001, Delivery Order 13.

U.S. EPA (1992) Dermal exposure assessment: principles and applications. Office of Research and
Development, Washington, DC; EPA/600/8-91/01 IB.

U.S. EPA (1997) Exposure factors handbook. Office of Research and Development, Washington,
DC; EPA/600/P-95/002 Fa-c.

U.S. EPA. (2001) Child-specific exposure factors handbook. Prepared by Versar, Inc., for the
Office of Research and Development, National Center for Environmental Assessment, under
EPA contract no. 68-W-99-041.

Wiley, J.A.; Robinson, J.P.; Cheng, Y.; Piazza, T.; Stork, L.; Plasden, K. (1991) Study of children's
activity patterns. California Environmental Protection Agency, Air Resources Board
Research Division, Sacramento, CA.

9-10


-------
10.0 BODY WEIGHT

10.1	INTRODUCTION

Body weight is an important parameter in several aspects of the metrics of toxicology, the
elements of exposure, and the expression of risk. Exposure and risk assessments are frequently
expressed as a function of dose normalized to the average body weight of the exposed population.
Body weight is one of the parameters in the calculation of the body mass index, by which overall
fitness is categorized and body fat content estimated. It can also serve as one parameter in estimating
body surface area, which is a key factor in some exposure and risk scenarios.

Creating an average growth reference of the relationship between weight and age requires
a database that is representative of the population under consideration, contains accurate
measurements from the sample subj ects, and uses a statistical process that appropriately fits smooth
percentile curves to the data (Ulijaszek et al., 1998). These three principles are good guides for the
evaluation of data and consideration of appropriate age bins.

Since the 1970s, one database has been the cornerstone for data on growth parameters for the
population of the United States. That is the National Health and Nutrition Examination Survey and
its predecessor surveys. The most recent survey is the Third National Health and Nutrition
Examination Survey (NHANES III), conducted between 1988 and 1994. In 1977 the National
Center for Health Statistics (NCHS), within the Centers for Disease Control (CDC), created
multipurpose growth charts that have been used by many research, clinical, and international health
organizations, including the World Health Organization (WHO). The underlying data, statistical
treatment of those data, and various other aspects of creating the growth charts have been subjected
to scrutiny by the users of the charts. In December 2000, NCHS released the revised growth charts
for the United States using the NHANES III data (CDC, 2000). Documentation on the population
surveyed, the survey techniques, and the statistics applied to the data are provided with the growth
charts.

10.2	EVALUATION OF EXISTING DATA

The NHANES IE database, distributed by NCHS, contains measured physical parameters on
a representative population of more than 30,000 individuals between the ages of 2 months and 90
years, collected between 1988 and 1994. Our literature search identified no comparably large set of
consistently measured data on physical characteristics of the population. The NHANES IE data

10-1


-------
support classification not only by age, but also by sex, race (white, black, other), and ethnicity
(Mexican-American, other Hispanic, not Hispanic). NHANES III:

•	Reflects actual measurements under consistent conditions (as opposed to self-reported
values),

•	Includes data on a large number of individuals collected as a representative sample of the
U.S. population, and

•	Contains demographic data that have been confirmed by in-person interviews with survey
respondents (for other databases, these values may be either self-reported or inferred).

Although the NHANES IE data appear to be the best available, they are cross-sectional.
Inferences about the temporal relationships of growth can be made with these data, but actual
longitudinal data about height and weight are not available. A second problem is that the data are
limited in terms of the numbers of individuals in particular demographic subgroups (defined by sex,
race, and ethnicity). These limitations reflect the attempt to capture the relative demographic
composition of the U.S. population as a whole within a limited set of measurements. However, this
composition results in the prediction of changes in growth parameters that is more uncertain for
small demographic subgroups than for the majority groups. For example, NHANES III has data for
5,340 male, white, non-Hispanic individuals but only 256 male, "other" (includes racial groups other
than white or black), non-Hispanic individuals.

Setting up a growth reference as the relationship between age and weight can be
accomplished directly if age is a good predictor of weight. If age is not a good predictor of weight,
one can explore other options, such as weight as a function of height. If age is a good predictor of
height, and if height is a good predictor of weight, height to weight references may be employed.
Other population factors, such as gender and ethnicity, may influence these relationships.

In creating the growth charts, the CDC grouped data by month of age from 1 through 11
months, by 3-month intervals (bins) from 12 through 23 months, and by 6-month intervals from 2
through 19 years. Survey weights were applied to the empirical data, then empirical percentile data
points were calculated and plotted at the midpoint of each age group. Statistical smoothing
procedures were applied to the observed data to generate the reference curves. All procedures are
described in detail by CDC and reflect improvements collected over the 30 years of experience with
such tasks.

CDC has generated weight-for-age growth curves for age groups birth to 36 months and 2
to 20 years, considering boys and girls independently. The 50th percentile of the population of

10-2


-------
children ages 0 to 36 months, along with the variability described by the 3rd to 97th percentile of these
populations, show little difference by gender. Growth is described by a phase of steep weight gain
per unit time from 0 to 6 months of age. A second, less steep weight gain slope follows from 6 to
12 months. From 12 to 36 months, a third slope can be fitted to the weight gain curve. The
differences in the slopes during these time periods are less pronounced for girls than for boys. The
variability described between the 3rd and 97th percentile of the population increases with age, and
by 36 months variability is large (CDC, 2000; Figures 1 and 2).

Similar curves were fitted for the length-for-age relationships (CDC, 2000; Figures 3 and 4).
The curve shows less obvious regions of discrete slopes, and the variability across the population
is modest, even at 36 months, compared with the variability demonstrated in the weight-for-age
growth curves.

Weight-for-age percentiles were also constructed for boys and girls ages 2 to 20 years (CDC,
2000; Figures 9 and 10, respectively). The length-for-age growth charts—now called stature-for-age
percentiles—were constructed for these gender/age groups also (CDC, 2000; Figures 11 and 12).

Both the central tendency and variability of height and weight increase as age increases.
Changes in height and weight are clearly nonlinear, and the emerging plateaus reflect a transition
from childhood to adult patterns.

Height shows a clear biphasic pattern, with a nearly linear increase up to about 16 years of
age. The actual inflection point varies, primarily reflecting earlier attainment of adult height in
females. Weight shows a similar pattern, except that a broader band of variability is evident. Height
and weight are relatively highly correlated for heights less than 4.5 feet, and very poorly correlated
beyond that point.

10.3 STUDIES SELECTED FOR ANALYSIS TO OBTAIN NEW RECOMMENDATIONS

NHANES III provides data for constructing age-based bins of growth parameters, including
age-to-weight. It also provides data for which age-to-height measurements and age-to-body mass
indices can be constructed. Data are presented for population percentiles, by gender, and are also
graphically presented in the CDC growth charts. These data sets and graphs are convenient for
estimating exposure factors data for the proposed age bins.

10-3


-------
10.4 RECOMMENDATIONS FOR PROPOSED AGE BINS

The data sets for the CDC growth charts provide weight calculations for males and females
for each month between birth and 20 years of age. Midpoints are reported for each selected
percentile. In our analysis, the 50th percentile values were selected for each age grouping. For time
periods over several months, we summed the values and computed an average. One value for males
and one for females is thus calculated for each time category. These are summed and the average
reported. The recommended values for body weight are presented in Table 10-1. The confidence
in rating for the body weight recommendations is shown in Table 10-2. To address other percentiles,
further statistical analysis is needed which is beyond the scope of this issue paper (see Section 10.5).

10.5 RECOMMENDATIONS FOR FURTHER ANALYSIS AND RESEARCH NEEDS

The age-to-weight bins in the CDC growth charts are adequate for estimations of exposure
and risk, normalized by averages of the population weights. However, this approach is only
minimally adequate in that many exposure and risk assessments use other body metrics as key
components. The dermal exposure assessments use surface area factors and are usually related to
some age groups and gender/age subpopulations. Increasingly, risk assessment considers
pharmakokinetic and pharmacodynamic relationships. In those cases, consideration of the fat
compartments of the population or subpopulation are useful. The NHANES III survey provides data
for all of these situations, and maybe as valuable as the age-to-weight factors.

The age-to-weight factor approach can be improved by computing age-to-length (age-to-
stature) relationships and then relating the average and variance of weight around each of those
categories. Similarly, a Body Mass Index value (BMI) may be presented in the same way.

Many risk assessments address some subpopulation of the U.S. population, especially by
gender and/or by ethnicity. The NHANES IE survey, as presented in the CDC growth tables, provide
the growth measurements of interest as a function of gender-specific percentiles. It would be an
improvement to present the age-to-weight estimates as gender-specific values. This is particularly
important for values representing ages greater than 2 years. As age increases, the differences
between boys and girls increase. The CDC summaries do not address ethnic differences, but the
NHANES IE data are available to conduct separate analyses for selected ethnic groups. These will
provide better accuracy when used with ethnic-specific exposure and risk assessments.

10-4


-------
Table 10-1. Recommended Body Weight Values for Proposed Age Bins (kg)

Age bin

Male: Mean at
50th percentile

Female: Mean at
50th percentile

Average Male/Female
Age Bin Value

0-1 month

4.00

3.80

3.90

1-2 months

4.88

4.54

4.71

3-5 months

6.72

6.15

6.43

6-11 months

9.04

8.28

8.66

1-2 years

11.53

10.80

11.16

3-5 years

16.27

15.83

16.05

6-10 years

25.86

25.95

25.90

11-15 years

45.77

45.41

45.59

16-17 years

62.84

54.54

58.69

10-5


-------
Table 10-2. Confidence in Recommendations for Body Weight









Rating (High, Medium, Low)















Age









Considerations

<1 Month

1-2 Mos

3-5 Mos

6-11 Mos

1 -2 Yrs

3-5 Yrs

6-10 Yrs

11-15 Yrs

16-17 Yrs

Study Elements



















• Level of peer review

High

High

High

High

High

High

High

High

High

• Accessibility

High

High

High

High

High

High

High

High

High

• Reproducibility

High

High

High

High

High

High

High

High

High

• Focus on factor of interest

High

High

High

High

High

High

High

High

High

• Data pertinent to U.S.

High

High

High

High

High

High

High

High

High

• Primary data

High

High

High

High

High

High

High

High

High

• Currency

High

High

High

High

High

High

High

High

High

• Adequacy of data collection period

High

High

High

High

High

High

High

High

High

• Validity of approach

High

High

Medium

Medium

Medium

Medium

Medium

Low

Low

• Representativeness of the

High

High

High

High

High

High

High

High

High

population



















• Characterization of variability in

High

High

Medium

Medium

Medium

Medium

Medium

Low

Low

the population



















• Lack of bias in study design

High

High

High

High

High

High

High

High

High

• Measurement error

High

High

High

High

High

High

High

High

High

Overall Rating

High

High

Medium

Medium

Medium

Medium

Medium

Low

Low

10-6


-------
10.6 REFERENCES

Cambridge encyclopedia of human growth and development. (1998) Ulijaszek, S.J., Johnston,
F.E., Preece, M.A., eds. Cambridge, UK: Cambridge University Press, pp.63-73.

CDC (Centers for Disease Control and Prevention). (1987) Second National Health and Nutrition
Examination Survey (NHANES II), 1976-1980. U.S. Department of Health and Human
Services, CDC, National Center for Health Statistics, Hyattsville, MD. Available online

at http://archive.nlm.nih.gov/proj/webmirs/doc/nhanes2/doc_search.html.

CDC (Centers for Disease Control and Prevention). (1999) Third National Health and Nutrition
Examination Survey (NHANES HI), 1988-1994. U.S. Department of Health and Human
Services, CDC, National Center for Health Statistics, Hyattsville, MD.

CDC (Centers for Disease Control and Prevention). (2000) CDC Growth Charts: United States.
Vital and Health Statistics # 314, December 4, 2000. U.S. Department of Health and
Human Services, National Center for Health Statistics, Hyattsville, MD. Available at

http://www.cdc.gov/nchs/about.major.nhanes/growthcharts.

10-7


-------