Unit III Article Critique

Soil ingestion in children and adults in the same family

SCOTT DAVIS a,b

AND DANA K. MIRICK a

a Program in Epidemiology, Division of Public Health Sciences, Fred Hutchinson Cancer Research Center, Seattle, WA, USA

b Department of Epidemiology, School of Public Health and Community Medicine, University of Washington, Seattle, WA, USA

Ingestion of soil may be a potentially important pathway of exposure to environmental pollutants. Although several studies have estimated soil ingestion

in children, data on ingestion in adults are sparse. The purposes of this study were to estimate soil ingestion in children aged 3 to 8 years and their parents,

identify factors associated with increased ingestion, and compare ingestion rates within the same family. Food/liquid, excreta, and soil/dust samples were

collected for the mother, father, and participant child for 11 consecutive days in 19 families. Soil ingestion was estimated using a mass balance approach.

Soil ingestion levels in children were similar to those reported previously, whereas adult estimates were somewhat higher than previous estimates.

Children’s eating of dirt and parents’ occupational contact with soil were associated with increased ingestion. Within families, soil ingestion levels in

children and adults were not correlated, although this analysis was based on fewer than 19 participant families. Children’s mean soil ingestion rates ranged

from 37 to 207 mg/day depending on the tracer, with the highest values based on titanium as a tracer. Adult mean soil ingestion rates ranged from 23 to

625 mg/day depending on the tracer, with the highest value based on titanium as a tracer. Soil ingestion rate estimates were more variable in adults than in

children.

Journal of Exposure Science and Environmental Epidemiology (2006) 16, 63–75. doi:10.1038/sj.jea.7500438; published online 27 July 2005

Keywords: adult, children, environmental, exposure, soil ingestion.

Introduction

Ingestion of soil and dust particles may be a potentially

important pathway of exposure to environmental pollutants,

both in children and adults, via recreational activities,

gardening, consumption of locally grown foods, and

ingestion of airborne dust. While adults may also ingest soil

via occupational exposure, it is generally assumed that child

ingestion rates are higher than adult rates, due primarily to

children’s mouthing behaviors. A number of studies have

attempted to estimate soil ingestion in children (Binder et al.,

1986; Clausing et al., 1987; Wong et al., 1988; Calabrese

et al., 1989; Davis et al., 1990; van Wijnen et al., 1990;

Calabrese et al., 1991; Stanek and Calabrese, 1995;

Calabrese et al., 1997a, b), whereas estimates of adult

ingestion are limited and based on relatively few participants

(Calabrese et al., 1990; Stanek et al., 1997). No study to date

has attempted to estimate soil ingestion in both children and

their parents, which would allow for meaningful comparisons

between adult and child ingestion rates while controlling for

other factors that could affect ingestion estimates, such as

contaminants and pollutants around the home environment,

hygiene practices and food choices that are unique to a given

family, and recreational activities shared by family members.

This study had several primary aims: (1) to estimate the

rate of soil ingestion during normal activities in the mother,

father, and participant child of 20 families; (2) to identify

behavioral and lifestyle factors potentially related to soil

ingestion values in children and adults; (3) to compare adult

and child soil ingestion estimates within the same family; and

(4) to compare ingestion estimates between the male and

female parent or guardian of the same family. The data

collected from the adults at the time of fieldwork constituted

the first obtained from a systematic investigation of this type

in the US. A secondary aim of the study was to substantially

refine existing field methods used to collect data of the type

required for ingestion estimates, potentially improving

existing estimates of soil ingestion values in young children.

Materials and methods

Participant families were a subset of those who completed a

previous study of soil ingestion in children conducted by the

investigative team 1 year prior to the current study (Davis

et al., 1990). Briefly, the previous study involved 104 children

between the ages of 2 and 7 years, selected randomly from

the tri-city area of Richland, Pasco, and Kennewick, located

in the arid, southeastern portion of Washington State. A

duplicate of all food items consumed, all feces excreted, Received 8 February 2005; revised 18 April 2005; accepted 18 April 2005;

published online 27 July 2005

1. Address all correspondence to: Dr S. Davis, Fred Hutchinson Cancer

Research Center, 1100 Fairview Avenue North, M4-A830, P.O. Box

19024, Seattle, WA 98109-1024, USA. Tel.: 206-667-2750.

Fax: 206 667 2683. E-mail: [email protected]

Journal of Exposure Science and Environmental Epidemiology (2006) 16, 63–75 r 2006 Nature Publishing Group All rights reserved 1559-0631/06/$30.00

www.nature.com/jes

twice-daily urine samples, soil, and house dust were collected

and analyzed for the presence of aluminum, silicon, and

titanium tracer elements. Soil ingestion was calculated using a

mass balance approach. Study methods and results are

described elsewhere in more detail (Davis et al., 1990). For

the present study, a subset of 20 families from the original

104 was chosen based on the following criteria: (1) they were

highly compliant with the previous study protocol, (2) they

expressed a willingness to participate in a future study, (3) the

participant child would be less than 8 years old at the time of

fieldwork, and (4) both a male and female parent or guardian

were living at home with the participant child. Thus, a

participant family consisted of the original child participant,

the female, and the male parent or guardian living in the

same house. The 20 families were selected without regard to

the results from the first study (i.e., selection was not based

on the child’s previous estimated soil ingestion). However,

study data were missing for one of the families. Therefore,

analysis was completed for the remaining 19 families.

Data Collection

Overview Data collection took place during the summer of

1988, and consisted of several components over a 14-day

period. Families were encouraged to avoid overnight, out-of-

town travel during their involvement. Each family received

$500 for completing the study. A short in-person interview

was administered on day 1 of participation. The purpose of

the interview was to collect information about the child’s and

parents’ food and dietary habits, personal habits,

occupational information, and family demographics.

Beginning on the second day of the study, a duplicate

sample of all food and non-food items consumed for 11

consecutive 24-h days were collected for all three family

members. Excreta was collected for 11 consecutive days,

beginning on day 3 of the study (offset by 24 h from the food

ingestion period). Parents completed a daily diary of

activities for themselves and the participant child for four

consecutive days. The field technician collected soil samples

based on information provided in the daily activity diary, as

well as household dust samples from the residence of each

family.

The Fred Hutchinson Cancer Research Center Institu-

tional Review Board approved the procedures for contacting

potential participants, obtaining informed consent, and all

data collection procedures, and both parents of each

participant child signed written informed consent before

participation.

Food and Liquid Collection and Processing Duplicate

samples of all food and non-food items (except for water)

consumed for 11 consecutive days were collected separately

for each of the three family members. A ‘‘day’’ was defined as

the period between midnight and 2359 hours, 24 h later.

Parents were given specific instructions for collecting certain

types of foods, such as frozen foods and candy, and were

instructed to not collect inedible parts of foods (e.g., banana

peels, chicken bones, food wrappers). A duplicate sample of

any meals taken away from home, such as in a restaurant or

at daycare, was to be included as well. A single sample of any

vitamin, food supplement, or medication (over-the-counter

or prescription) taken by mouth or as a suppository was

collected and labeled with the participant identification

number. Parents recorded the amount taken during each

24-h period. For each food collection day, parents were

instructed to list any food or liquid consumed that was not

contained in the duplicate sample (i.e., a missed item) and to

estimate the amount of the item consumed.

The 24-h food and liquid samples were collected daily by

the field technician and delivered to Battelle Pacific North-

west Laboratories (Richland, Washington), where they were

stored in a laboratory refrigerator until processing. Proces-

sing occurred immediately upon completion of sample

collection. Food/liquid samples were combined into two

time periods, from the first 4 days (study days 2–5) and last 7

days of sample collection (days 6–12), and homogenized in a

stainless steel blender. Weighted aliquots from each of the

two time periods was dried in a forced air oven at 80–1001C

for 72 h and ashed in platinum crucibles in a furnace at

9001C–10001C for 16–24 h. Once cooled, the ashed sample

was weighed and pulverized to a fine powder in a mortar, and

was then pressed into a undiluted thin wafer of approxi-

mately 50 mg/cm 2

thickness before storing for chemical

analysis (described below). Prescription and over-the-counter

medications and vitamins were submitted for analysis in their

original form.

Excreta Collection and Processing Offset by 24 h from the

food ingestion period, excreta was collected beginning on day

3 of the study in the following manner: (1) all feces for 11

consecutive days, (2) first morning and last evening void of

urine, as well as all urine excreted during the night, for 9

consecutive days, and (3) complete 24-h urine volume for 2

consecutive days. Samples from each participant family

member and each day were kept separate from one another.

As for the food/liquid collection, a ‘‘day’’ was defined as the

period between midnight and 2359 hours, 24 h later. As

much as possible, urine and feces samples were to be collected

and kept separate from one another. Parents were provided

with toilet-mounted collection devices to facilitate excreta

collection. Each day, parents were instructed to indicate the

number of samples collected and the number of samples

missed on a daily excreta collection form for each participant

family member.

Urine and fecal samples were collected daily by the field

technician and delivered to Battelle, where they were weighed

and stored in a laboratory refrigerator until processing.

Processing occurred immediately upon completion of excreta

Soil ingestion in children and adults in the same familyDavis and Mirick

64 Journal of Exposure Science and Environmental Epidemiology (2006), 16(1)

collection. Since daily urine samples for each participant were

often in multiple containers, samples were combined prior to

weighing and storing. Once sample collection was complete,

urine samples for each participant were consolidated into

four time periods as follows: (1) the first 4 days of sample

collection (study days 3–6), (2) the next 5 days (days 7–11),

(3) the first 24-h collection (day 12), and (4) the last 24-h

urine collection (day 13). Once combined, urine samples were

homogenized by thoroughly mixing with a stir-bar. Each

combined urine sample was weighed and transferred to a

crystallizing dish, and evaporated to hygroscopic syrup in an

oven at 80–1001C. These semidry samples were ashed and

stored for chemical analysis (described below). Similarly,

daily fecal samples were combined into one container prior to

weighing if there was more than one sample for a given

participant and day. Once sample collection was complete,

fecal samples were consolidated into two time periods, from

the first 4 days (study days 3–6) and the last 7 days of sample

collection (days 7–13). As for the food/liquid samples,

weighed aliquots from each of the two time periods were

dried in a forced air oven at 801C–1001C for 72 h and ashed

in platinum crucibles in a furnace at 9001C–10001C for 16–

24 h. Once cooled, the ashed sample was weighed and

pulverized to a fine powder in a mortar, and was then pressed

into a undiluted thin wafer of approximately 50 mg/cm 2

thickness before storing for analysis.

Daily Activity Diary Participants completed a daily

activities form for 4 consecutive days, beginning on study

day 4. Both the child’s and parents’ forms collected

information about time spent indoors and outdoors at

home and away from home during the 24-h period of

interest (i.e., between midnight and 2359 hours, 24 h later).

The child’s form asked specific questions regarding contact

with dirt or soil in both the family yard as well as time spent

away from home, and whether the child ate or drank while

sitting on the ground. The adult’s form addressed time spent

indoors performing household chores (e.g., vacuuming,

dusting, etc.) and outside doing various house and yard

maintenance (e.g., gardening, sweeping, etc.). In addition to

these questions, parents were asked to assist the field

technician in drawing a diagram of the yard, in order to

collect soil samples from those areas where each study

participant played or spent the most time around the house

and yard (see below).

Household Dust Sampling On day 5 of participation, the

field technician collected household dust samples from the

residence of each family, using a standardized collection

procedure to ensure uniformity among the samples. A hand-

held vacuum cleaner fitted with filter paper was used to

collect dust samples from floors both with and without

carpeting, from four rooms of the house: (1) the participant

child’s bedroom, (2) the parents’ bedroom, (3) the principal

living area of the house, and (4) kitchen. Each room was

vacuumed for approximately 3 min. The field technician

vacuumed bare floors rather than carpeting whenever

possible, and focused on areas where the child was most

likely to be in direct contact with the surface. Once several

millimeters of dust had accumulated on the filter paper, it was

removed using rubber gloves and forceps and placed in a

clean glass vial. Additionally, a large sample of household

dust was obtained from the participant family’s own vacuum

cleaner. If the family did not own a vacuum cleaner, the

interviewer obtained a vacuum sample by using the same

hand-held vacuum without the filter paper.

Household dust samples were delivered by the field

technician to Battelle, where they were promptly washed

from the filter paper using a dilute 0.1 N nitric acid solution.

Samples were then dried in a ventilated drying oven for at

least 24 h and stored. Dust samples from the home vacuum

cleaner were passed successively through 20- and 60-mesh

stainless-steel sieves to remove foreign material before

storage.

Soil Sampling On day 12 of participation, soil samples

were collected from the areas around the house where each

individual family member spent time, as determined from the

4-day daily activity diary for each individual, and pooled in a

manner proportional to the amount of time spent in the

various locations that were sampled. Both surface and core

soil samples were collected. Surface samples were collected

using a hand-held vacuum cleaner. At each location

identified by the participant as having spent time, the

surface was vacuumed for 1 s per percentage point of time

spent at that location. A different vacuum bag was used for

each participant family member. Five core samples were

collected at each location identified by the participant as

having spent time, using a soil-coring device that was

hammered into the ground to remove a plug of the soil

1 in 3 in size at a depth of approximately 3 in. Samples were

collected at the four corners of a square area 3 yards by 3

yards, as well as the point where the diagonals of the area

intersect, and consolidated into one sample for that location.

Each sample was labeled with the percentage of time spent at

that location, so that all the samples for a given participant

could be pooled proportionally to the time spent at each

location. No samples were collected away from the home

area.

Soil samples were delivered by the field technician to

Battelle, where they were promptly dried in a ventilated

drying oven at approximately 1251C for several days. The

dried samples were then passed successively through 20- and

60-mesh stainless-steel sieves in order to remove pebbles,

grass, and roots before storage.

Chemical Analysis Upon completion of all data collection

and sample processing, food/liquid, medications, feces, urine,

Soil ingestion in children and adults in the same family Davis and Mirick

Journal of Exposure Science and Environmental Epidemiology (2006), 16(1) 65

soil, and house dust samples were stored at Battelle, Richland

until funds were made available for chemical analyses. In

2000, the samples were transferred to Battelle Marine

Sciences Laboratory (Sequim, Washington) for analysis.

After digestion with an acid mixture, all sample types were

analyzed for concentrations of silicon (Si), aluminum (Al),

and titanium (Ti) using inductively coupled plasma-atomic

emission spectrometry (ICP-AES) and inductively coupled

plasma-mass spectrometry (ICP-MS).

For all sample types, quality control samples were

analyzed with the study samples, including (1) laboratory

control samples (LCS) with known concentrations of tracer

elements of interest (food, feces, urine, medication) or

standard reference material (SRM; soil, house dust), (2)

matrix spike samples that were spiked in duplicate with

known concentrations of the tracer elements (MS/MSD),

and (3) duplicate digestions of some samples. Recoveries for

most analyses were within the quality control limits of

720% for the LCS or SRM and duplicate digestions, and 725% for the MS/MSD. Initial and continuing calibration verification standards of the ICP-AES and ICP-MS instru-

ments were within 715% of the true value of each calibration standard for all metals.

Statistical Methods A mass balance approach was used to

estimate daily soil ingestion values for each participant family

member. Ideally, this method requires that a person’s total

intake and output be collected in a complete fashion, and that

all collected samples be analyzed for tracer elements. Sample

data were more complete when limited to the last 7 days of

data collection (see below). Therefore, in order to maximize

the accuracy of the final soil ingestion determination, all soil

ingestion estimates and results are based on samples collected

on these final 7 days, which includes study days 6 through 12

for food/liquid samples and days 7 through 13 for excreta

samples. In this manner, the 24-h offset between collection of

food/liquid and excreta samples is preserved. Soil ingestion

was calculated in three different ways, by using core soil,

surface soil, or household dust as the source of the ingested

soil. For participant i and tracer element E, soil ingestion is

calculated as:

Si;E ¼

ððWtf �ADJf �Ef Þþ P

j

ðWtuj�ADJumj�ADJu24j �Euj ÞÞ

�ððWtfd �Efd ÞþðWtmd �Emd ÞÞ

Esoil=dust�10; 000 ð1Þ

where: Si,E ¼ soil ingested (g) for participant i based on tracer E; Wtf ¼feces dry weight (g), ADJf ¼adjustment for missing fecal samples, Ef ¼measured tracer concentration in feces (mg/g); Wtuj ¼urine dry weight (g) for urine collection period j, ADJumj ¼adjustment for missing urine samples in collection period j, ADJu24j ¼adjustment factor to estimate 24-h urine weight during twice-daily urine samples (days

7–11 only), Euj ¼measured tracer concentration in urine (mg/ g) in collection period j; Wtfd ¼food/liquid dry weight (g),

Efd ¼measured tracer concentration in food/liquid (mg/g), Wtmd ¼medication dry weight (g), Emd ¼measured tracer concentration in medication (mg/g), and Esoil/dust ¼measured tracer concentration in core soil, surface soil, or household

dust (% weight, multiplied by 10,000 to convert to units of

mg/g). For the 5 days of data collection in which urine samples were collected twice daily (days 7–11), an adjustment

factor was calculated to estimate 24-h urine weight during

these days (ADJu24j in the above equation), using actual 24-h

urine weight from the 24-h collection periods (days 12 and

13). In order to account for missing fecal samples (ADJf in

the above equation), the fecal dry weight was multiplied by a

ratio of the number of fecal samples that should have been

collected (i.e., the number collected þthe number missed) to the number that was actually collected. This adjustment

increased the fecal dry weight in proportion to the number of

samples missed. A comparable adjustment was made to

account for missing urine samples (ADJumj in the above

equation). Collection of food/liquid samples was complete,

and therefore no adjustment for missing samples was

necessary.

Soil ingestion was calculated and analyzed for the final 7

days of data collection only. Total soil ingestion values were

divided by seven to reflect average daily values, and units

were converted to mg/day. In some instances, the soil

ingestion calculation for an individual produced a negative

estimate (i.e., amount of tracer in food þmedication4 amount of tracer in feces þurine). For these cases, the ingestion estimate was set to 0 mg/day for tabulation and

analysis.

Results

Study data were missing for one of the original 20 participant

families. All of the remaining 19 families completed the in-

person interview, and the mother, father, and participant

child of each family completed the daily diary of activities for

all 4 days. Table 1 displays selected characteristics of

participant children and their parents, based on responses

to the in-person interview. Nearly two-thirds of participant

children were male and more than 70% were aged 5 years

and older. In 17 of 19 families, the responding parent (i.e.,

the person who completed the in-person interview) was the

mother. The majority of both mothers and fathers completed

at least some college. Of the mothers, most (13 of 19) were

homemakers, whereas the fathers’ occupations were varied

with no clear tendency towards blue- or white-collar

occupations. Slightly more than 60% had annual family

incomes under $30,000 (in 1988). One mother identified

herself as Native American; all other parents identified

themselves as Caucasian.

Responses to selected questions from the in-person inter-

view regarding hygiene and eating behaviors are presented in

Soil ingestion in children and adults in the same familyDavis and Mirick

66 Journal of Exposure Science and Environmental Epidemiology (2006), 16(1)

Table 2. Most study participants brushed their teeth at least

once per day and washed their hands, but not their faces,

before meals. None of the families reported eating food or

drinking liquids that were prepared, served, or stored in

homemade or imported clay pottery; the participant child of

one family consumed a food or liquid that had been stored in

a previously opened tin or aluminum container once during

the past week (data not shown). Few families reported

consuming unwashed vegetables, fruits, or berries either from

a family owned yard or garden or from another location such

as a neighbor’s yard or a store.

Responses to questions from the in-person interview

regarding the participant child’s mouthing behavior are

shown in Table 3. Most responding parents (16 of 19)

reported that their child does not suck his/her thumb or

fingers. Of 19 children, 11 carry around a favorite blanket or

toy; of these, only two take the item outside or put the item in

their mouths. All but one child take food or liquid outside.

Most participant children (16 of 19) do not mouth furniture

or windowsills, but more than half of the children (58%)

swallow dirt at least once per week.

Table 4 presents responses to questions regarding parents’

(respondent and spouse) occupational and recreational

behaviors that potentially involve contact with soil. More

mothers than fathers reported occupational contact with soil

for at least 1 h/week (47 versus 26%). However, four of 19

fathers reported contact with soil for at least 20 h/week

during work. Nearly all parents reported at least 1 h/week of

yard work (36 of 38), and most parents (30 of 38) reported

participating in outdoor recreational activities involving

contact with soil or dirt. Regarding indoor activities

potentially involving contact with soil, all 19 mothers

reported doing housework every week, and 21% (four of

19) reported engaging in some type of remodeling or indoor

carpentry. In contrast, 47% (nine of 19) of fathers did at

least some housework during the week and 42% (eight of 19)

did some remodeling or carpentry every week.

The completeness of collection and laboratory analyses of

food, excreta, soil, and dust samples for the final 7 days of

sample collection are presented in Table 5. All of the

participant children’s urine samples from the first 4 days of

data collection were unavailable for laboratory analysis, and

therefore all analyses were limited to the final 7 days of data

collection. Collection and chemical analyses of food/liquid

samples for the mother, father, and participant child was

complete for all 19 families. Overall, children’s urine

collection was less complete than that of mothers and

fathers. For urine collection on days 7–11 (twice-daily

collection), 10 of 19 children provided complete samples

(i.e., no collections missed and no incomplete collections).

Chemical analyses were conducted on some samples that

were missing at least part of the urine sample. Therefore,

laboratory results were available for two of the nine

incomplete children’s samples from days 7–11, for a total

of 12 samples. Similarly, for both mothers and fathers, 15 of

19 collected complete samples for days 7–11, but results of

chemical analyses were available for 18 of 19. The two 24-h

Table 1. Selected characteristics of participant children and respondent

parents, Soil Ingestion Study.

Characteristic No. (%) a

Participant child

Sex

Male 12 (63.2)

Female 7 (36.8)

Age (years)

3 2 (10.5)

4 3 (15.8)

5 4 (21.0)

6 5 (26.3)

7 5 (26.3)

Responding parent b

Mother 17 (89.5)

Father 2 (10.5)

Highest level of education (completed)

Mother

o12 years 2 (10.5) High school graduate 4 (21.0)

Some college 10 (52.6)

College graduate 2 (10.5)

Post-graduate study 1 (5.3)

Father

o12 years 0 F High school graduate 2 (10.5)

Some college 10 (52.6)

College graduate 3 (15.8)

Post-graduate study 4 (21.0)

Current occupation

Mother

Managerial/professional 3 (15.8)

Technical 2 (10.5)

Service 1 (5.3)

Farming 0 F Production/laborers 0 F Homemaker 13 (68.4)

Father

Managerial/professional 5 (26.3)

Technical 4 (21.0)

Service 2 (10.5)

Farming 1 (5.3)

Production/laborers 6 (31.6)

Homemaker 1 (5.3)

Family income (annual) c

o$15,000 3 (15.8) $15,000F$30,000 9 (47.4) $30,000–$45,000 6 (31.6)

4$45,000 1 (5.3)

a Percentage calculated from total number of participant families (n ¼ 19). b Parent who completed the in-person interview.

c For the year preceding data collection (1987).

Soil ingestion in children and adults in the same family Davis and Mirick

Journal of Exposure Science and Environmental Epidemiology (2006), 16(1) 67

urine collection periods were slightly more successful: 14 of

19 children and 17 of 19 mothers had no missing or

incomplete urine samples for either of the two 24-h periods.

Regarding fathers, 15 of 19 for the first 24-h collection, and

17 of 19 for the second 24-h collection had no missing or

incomplete urine samples. Regardless of the number of

incomplete 24-h urine samples, laboratory results were

available for nearly all participants.

Regarding fecal samples for study days 7–13, 84% (16 of

19) of participant children provided complete fecal samples,

and 16% (three of 19) of children provided fecal samples

with at least one missing or incomplete collection. In total,

74% of mothers and 89% of fathers provided complete fecal

samples, with the remainder providing samples with at least

one missing or incomplete collection. As with the urine,

chemical analyses were conducted on all samples, regardless

of completeness.

During the 4 days in which the daily diary was completed

for all study participants (study days 4–7), all 19 children

spent at least some time outside playing around the house

and yard; therefore, surface and core soil samples were

collected and analyzed for all 19 children. Two of 19 each of

Table 2. Responses to questions regarding hygiene and eating behaviors of participant families, from the in-person interview of the Soil Ingestion

Study.

Behavior Child Mother Father

No. (%) a

No. (%) a

No. (%) a

Frequency of tooth-brushing (times/day)

o1 3 (15.8) 0 F 0 F 1–2 8 (42.1) 5 (26.3) 10 (52.6)

2 or more 8 (42.1) 14 (73.7) 9 (47.4)

Washes hands before meals

Yes 14 (73.7) 16 (84.2) 13 (68.4)

No 5 (26.3) 3 (15.8) 6 (31.6)

Washes face before meals

Yes 1 (5.3) 1 (5.3) 0 F No 18 (94.7) 18 (94.7) 19 (100)

Bites or chews fingernails

Yes 4 (21.0) 6 (31.6) 5 (26.3)

No 15 (79.0) 13 (68.4) 14 (73.7)

Contact with pet

Yes 18 (94.7) 15 (79.0) 14 (73.7)

No 1 (5.3) 4 (21.0) 5 (26.3)

Consumes unwashed vegetables (times/week)

From family-owned garden

0 19 (100) 18 (94.7) 19 (100)

1 0 F 1 (5.3) 0 F From store, neighbor’s garden, etc.

0 18 (94.7) 18 (94.7) 18 (94.7)

1 1 (5.3) 1 (5.3) 1 (5.3)

Consumes unwashed fruits or berries (times/week)

From family yard

0 17 (89.5) 17 (89.5) 19 (100)

1–2 1 (5.3) 1 (5.3) 0 F 42 1 (5.3) 1 (5.3) 0 F

From store, neighbor’s yard, etc.

0 15 (79.0) 18 (94.7) 17 (89.5)

1–2 2 (10.5) 0 F 1 (5.3) 42 2 (10.5) 1 (5.3) 1 (5.3)

a Percentage calculated from total number of participant families (n ¼ 19).

Soil ingestion in children and adults in the same familyDavis and Mirick

68 Journal of Exposure Science and Environmental Epidemiology (2006), 16(1)

mothers and fathers did not spend any time outdoors around

the house and yard; hence, no surface or core soil samples

were collected for these individuals. Of the remaining 17

mothers and 17 fathers, one mother and two fathers spent

time outdoors on surfaces in which core soil samples could

not be obtained. Therefore, core soil samples were collected

and analyzed for 16 mothers and 15 fathers. Household dust

was collected and analyzed from the homes of all 19 families.

Table 6 presents the results of the chemical analyses for the

food/liquid, excreta, soil, and household dust samples.

Concentrations of aluminum, silicon, and titanium were

generally lowest in urine samples, and considerably more

concentrated in the fecal samples. Regardless of the source of

the sample, silicon was present in substantially higher

concentrations than either aluminum or titanium. Children

consumed the least amount of food and liquid and fathers

consumed the most. Aluminum concentrations in food/liquid

were similar across the three family members, but silicon and

titanium differed somewhat. While children’s and mothers’

food samples had similar concentrations of silicon, the silicon

concentration of fathers’ samples were considerably higher.

Titanium levels in food/liquid among the three family

members vary according to whether the mean or median

level is considered: while children’s mean titanium levels were

similar to mothers and higher than fathers, children’s median

levels were higher than both mothers and fathers. For all

family members, aluminum concentrations in urine were

lower for the 5-day collection than for either 24-h period.

Silicon concentrations were higher for the last 24-h period

than for the other two collection periods. Titanium

concentrations did not differ substantially over the three

time periods and were overall quite low. Within each urine

collection period, tracer concentrations were similar across all

three family members. Compared to parents, children had

higher concentrations of aluminum and silicon, but not

titanium in feces. Mothers had the lowest silicon concentra-

tions, but the highest titanium concentrations. Fathers had

the lowest titanium concentrations.

Regarding core and surface soil samples, concentrations of

the three tracer elements were nearly identical for all three

Table 3. Responses to questions regarding child’s mouthing behavior,

from the in-person interview of the Soil Ingestion Study.

Behavior No. of children (%) a

Sucks thumb or fingers

Yes 3 (15.8)

No 16 (84.2)

Carries around a favorite blanket, toy, or stuffed animal

Yes 11 (57.9)

No 8 (42.1)

Takes item outside

Yes 2 (10.5)

No 9 (47.4)

Puts item in mouth

Yes 2 (10.5)

No 9 (47.4)

Mouths furniture or licks window sills

Yes 3 (15.8)

No 16 (84.2)

Takes food or liquid outside

Yes 18 (94.7)

No 1 (5.3)

Swallows dirt (times/week)

Never 8 (42.1)

1–2 6 (31.6)

3–10 4 (21.0)

410 1 (5.3)

a Percentage calculated from total number of participant children (n ¼ 19).

Table 4. Responses to questions regarding parents’ occupational and

recreational behaviors involving contact with soil, from the in-person

interview of the Soil Ingestion Study.

Behavior Mother Father

No. (%) a

No. (%) a

Occupational contact with soil or dirt (hours/week)

None 10 (52.6) 14 (73.7)

1–10 8 (42.1) 1 (5.3)

10–20 1 (5.3) 0 F 20 or more 0 F 4 (21.0)

Outdoor activities

Yard work b (hours/week)

None 0 F 2 (10.5) 1�5 13 (68.4) 10 (52.6) 5�10 6 (31.6) 5 (26.3) 10 or more 0 F 2 (10.5)

Recreation c

None 5 (26.3) 3 (15.8)

1�5 10 (52.6) 7 (36.8) 5�10 2 (10.5) 4 (21.0) 10 or more 2 (10.5) 5 (26.3)

Indoor activities

Housework d

None 0 F 10 (52.6) 1�5 17 (89.5) 8 (42.1) 5�10 1 (5.3) 0 F 10 or more 1 (5.3) 1 (5.3)

Remodeling/indoor carpentry

None 15 (79.0) 11 (57.9)

1�5 4 (21.0) 7 (36.8) 5 or more 0 F 1 (5.3)

a Percentage calculated from total number of participant families (n ¼ 19). b Includes gardening, weeding, mowing, trimming bushes/hedges, sweep-

ing, and other yard maintenance. c Includes sports and/or games, camping, playing, water sports, and

attending outdoor auctions. d Includes cleaning, sweeping/vacuuming, dusting, laundry, and caring for

houseplants.

Soil ingestion in children and adults in the same family Davis and Mirick

Journal of Exposure Science and Environmental Epidemiology (2006), 16(1) 69

family members, and core and surface soil tracer concentra-

tions were significantly correlated within each tracer element

(correlation coefficients ranged from 0.38 to 0.75, Po0.01). The three family members shared a common composite

house dust sample (see above); however, across families,

house dust concentrations of all three tracers varied

approximately two-fold between the lowest and highest

concentrations (aluminum: range ¼ 2.6–5.8% wt; silicon: range ¼ 12.3–24.8% wt; and titanium: range ¼ 0.3–0.7% wt). House dust silicon concentration was correlated with

surface soil silicon concentration (correlation

coefficient ¼ 0.32, P ¼ 0.02), but not with core soil silicon concentration (correlation coefficient ¼ 0.15, P ¼ 0.28). Con- centrations of aluminum and titanium in house dust were not

correlated with those tracers found in core or surface soil.

Table 7 presents the results of the chemical analyses of

medications taken by the study participants during data

collection. Total amounts and tracer concentrations of

individual medications used during the final 7 days of data

collection are summarized across family members and

medication type (solid or liquid). Liquid medications in

particular had fairly high concentrations of tracer elements,

but the amounts taken were generally quite small.

Summary statistics of estimated soil ingestion for each of

the three family members is presented in Table 8. For

individual study participants, ingestion estimates did not

differ substantially using core or surface soil, or household

dust, in the ‘‘soil’’ component of Eq. (1) (above). Thus,

estimates are presented based on using core soil tracer

concentrations of aluminum, silicon, and titanium in the

ingestion calculations. Since the core soil tracer element

concentrations were nearly identical for each of the three

family members (see Table 6), the mean concentrations of

aluminum, silicon, and titanium were calculated for each

family, and these ‘‘family means’’ were used in ingestion

calculations whenever the core soil data were missing for a

given individual. Soil ingestion could not be estimated if

laboratory results for food or excreta data were missing;

Table 5. Completeness of data collection and laboratory analyses of food and excreta, by family participant, Soil Ingestion Study.

Sample type Sample days Participant Collected

Complete a

Incomplete b

Analyzed for tracer elements c

No. (% d ) No. (%

d ) No. (%

d )

Food/liquid Days 6 –12 Child 19 (100) 0 F 19 (100) Mother 19 (100) 0 F 19 (100) Father 19 (100) 0 F 19 (100)

Urine Days 7–11 e

Child 10 (52.6) 9 (47.4) 12 (63.2)

Mother 15 (79.0) 4 (21.0) 18 (94.7)

Father 15 (79.0) 4 (21.0) 18 (94.7)

Day 12: 24-hour sample Child 14 (73.7) 5 (26.3) 19 (100)

Mother 17 (89.5) 2 (10.5) 18 (94.7)

Father 15 (79.0) 4 (21.0) 19 (100)

Day 13: 24-hour sample Child 14 (73.7) 5 (26.3) 19 (100)

Mother 17 (89.5) 2 (10.5) 19 (100)

Father f

17 (89.5) 1 (5.3) 18 (94.7)

Feces Days 7–13 Child 16 (84.2) 3 (15.8) 19 (100)

Mother 14 (73.7) 5 (26.3) 19 (100)

Father 17 (89.5) 2 (10.5) 19 (100)

Surface Soil Day 12 Child 19 (100) 0 F 19 (100) Mother 17 (89.5) 2 (10.5) 17 (89.5)

Father 17 (89.5) 2 (10.5) 17 (89.5)

Core Soil Day 12 Child 19 (100) 0 F 19 (100) Mother 16 (84.2) 3 (15.8) 16 (84.2)

Father 15 (78.9) 4 (21.0) 15 (78.9)

Household Dust Day 5 Family 19 (100) 0 F 19 (100)

a No missing or incomplete samples. b Includes sample periods with at least one missing or incomplete sample.

c Aluminum, silicon, titanium. d Percentage calculated from total number of participant families (n ¼ 19).

e Urine was collected at first morning and last evening void for this 5-day period. f The father of one family left town on day 13 and did not provide any samples for this day.

Soil ingestion in children and adults in the same familyDavis and Mirick

70 Journal of Exposure Science and Environmental Epidemiology (2006), 16(1)

therefore, soil ingestion could only be calculated for 12 of 19

children since seven children had missing urine data for days

7–11. Similarly, 16 mothers and 17 fathers had complete

food and excreta data, and thus summary statistics of

ingestion estimates are based on these numbers.

Within each family member grouping (children, mothers,

fathers), the titanium tracer produced the highest estimates of

soil ingestion. Relative to mothers and fathers, children had

the lowest mean soil ingestion when estimates were based on

aluminum and titanium tracers, but had the highest mean

ingestion using silicon as the tracer element of interest. Using

Spearman Rank correlation coefficients, within-family cor-

relations of soil ingestion were investigated, and no clear

pattern emerged. Regardless of the tracer element used,

children’s soil ingestion does not appear to be associated with

either their mothers’ or their fathers’ ingestion, nor do

mothers’ and fathers’ soil ingestion appear to be correlated.

Mothers’ and fathers’ ingestion estimates were nonsignifi-

cantly positively correlated using aluminum and silicon

tracers (correlation coefficient ¼ 0.45, P ¼ 0.09; and 0.24, P ¼ 0.39, respectively), but negatively correlated using titanium tracer (correlation coefficient ¼ �0.11, P ¼ 0.70). Children’s and parents’ ingestion estimates were negatively

correlated using aluminum tracer (correlation

coefficient ¼ �0.27, P ¼ 0.39), but positively correlated using silicon and titanium (correlation coefficients ¼ 0.14, P ¼ 0.66; and 0.28, P ¼ 0.37, respectively).

Estimated soil ingestion was investigated according to

responses from selected behaviors reported in the in-person

interview that potentially could be related to soil ingestion

and which had sufficient numbers in behavior categories for

investigation. Among both parents and children, neither

eating unwashed fruits or vegetables (from any source), nor

nail biting was associated with increased soil ingestion.

However, washing hands before meals was associated with

increased soil ingestion, contrary to the expectation that

hand-washing would decrease soil ingestion. Among chil-

dren, thumb-sucking, carrying around a blanket or toy, or

Table 6. Mean and median measured concentrations of tracer elements, by sample type and family participant, Soil Ingestion Study.

Sample type Participant N a

Dry weight (g/day)

Tracer element (mg/g dry weight)b

Aluminum Silicon Titanium

Mean Median Mean Median Mean Median Mean Median

Food/liquid Child 19 285.6 287.2 27.3 18.5 50.0 39.5 23.5 9.9

Mother 19 395.3 367.7 23.3 14.1 53.8 46.5 24.0 4.3

Father 19 520.6 460.6 24.3 19.5 101.6 62.5 15.4 6.3

Urine

Days 7–11 Child 12 9.2 8.9 0.76 0.63 24.8 23.6 0.62 0.55

Mother 18 14.5 13.8 0.96 0.63 32.2 32.6 0.72 0.72

Father 18 20.1 18.7 0.79 0.63 33.4 28.7 0.76 0.57

Day 12 Child 19 15.4 16.7 1.87 1.50 29.3 25.4 0.78 0.80

Mother 18 31.4 31.2 2.08 1.50 25.6 24.2 0.69 0.67

Father 19 43.5 44.6 1.85 1.50 24.9 22.5 0.75 0.75

Day 13 Child 19 17.0 17.7 1.56 1.24 72.0 51.4 0.84 0.82

Mother 19 34.6 36.7 1.43 1.03 53.0 43.3 0.66 0.66

Father 18 52.7 51.9 1.32 1.03 61.7 59.8 0.78 0.79

Feces Child 19 13.4 13.6 767.5 440.0 1701.3 1450.0 516.3 279.0

Mother 18 18.5 15.6 610.8 333.0 1272.0 1040.0 602.2 385.0

Father 19 28.0 28.9 565.5 386.0 1537.8 1210.0 318.7 213.0

Core soil Child 19 Fc 6.47 6.52 27.4 27.8 0.62 0.56 Mother 16 6.51 6.57 27.4 27.7 0.62 0.59

Father 15 6.40 6.42 27.4 27.6 0.60 0.58

Surface soil Child 19 Fc 6.16 6.31 26.4 26.8 0.60 0.59 Mother 17 6.34 6.32 27.2 27.5 0.62 0.58

Father 17 6.26 6.22 27.2 27.1 0.59 0.56

House dust Family 19 F c 4.75 4.96 20.1 20.2 0.51 0.54

a Number of participants with laboratory results for the given sample type. b Soil and dust concentrations are presented in % weight.

c Soil samples were analyzed in aliquots weighing approximately 0.25 g.

Soil ingestion in children and adults in the same family Davis and Mirick

Journal of Exposure Science and Environmental Epidemiology (2006), 16(1) 71

furniture licking was not associated with increased soil

ingestion (data not shown) However, children’s reported

eating of dirt was associated with increased estimated soil

ingestion using aluminum or silicon, but not titanium

(aluminum: 51.2 versus 16.5 mg/day, silicon: 45.0 versus

28.5 mg/day, titanium: 129.0 versus 315.8 mg/day, ever

versus never eat dirt). Among parents, doing indoor

carpentry, outside yard work, or participating in outdoor

recreational activities was not associated with increased

estimated soil ingestion (data not shown). However, fathers’

occupational contact with soil was associated with an

increase in soil ingestion, particularly when the aluminum

and titanium tracers are considered (aluminum: 136.8 versus

39.9 mg/day, silicon: 29.2 versus 24.8 mg/day, titanium:

929.7 versus 497.9 mg/day, any hours/week versus no

hours/week). Similar results were observed for mothers using

aluminum and titanium tracers, but not silicon (aluminum:

146.6 versus 37.6, silicon 22.7 versus 24.7, titanium: 503.2

versus 214.8, any hours/week versus no hours/week).

Soil ingestion could not be calculated for a number of

participants due to missing urine data (seven children, two

mothers, and two fathers). As tracer concentrations in urine

were quite low and therefore have limited impact on soil

ingestion estimates, alternate estimates were calculated

disregarding the urine component of Eq. (1), and the

analyses described above were repeated. As expected,

ingestion estimates were slightly reduced among those

participants with complete data, since the total estimated

Table 7. Average daily quantity and tracer concentrations of medications used in the Soil Ingestion Study, by medication type (solid or liquid) and

family participant.

Medication type Participant N a

Average daily quantity b

Tracer element c

Aluminum Silicon Titanium

Mean Median Mean Median Mean Median

Solid Child 8 0.71 1231.1 866.4 3506.6 2996.4 3.82 0.96

Mother 16 0.49 6781.8 405.8 10633.1 2935.9 2880.6 2884.9

Father 11 0.79 398.9 432.0 3841.0 2421.2 618.6 2.74

Liquid Child 2 9.2 0.30 0.30 13.4 13.4 0.07 0.07

Mother 6 6.8 237.1 0.60 836.9 4.3 0.06 0.03

Father 5 12.4 42.2 0.30 44.2 3.8 0.28 0.02

a Number of participants who used the given medication type during the final 7 days of data collection. b Solids: g/day; liquids: ml/day.

c Medication tracer concentrations are presented in mg/g dry weight for solids and mg/ml for liquids.

Table 8. Mean values of estimates of soil ingestion (mg/day) based on aluminum, silicon, or titanium tracer and core soil sample, by family

participant, Soil Ingestion Study.

Participant Tracer element Estimated soil ingestion a (mg/day)

Mean Median Std Maximum

Child b

Aluminum 36.7 33.3 35.4 107.9

Silicon 38.1 26.4 31.4 95.0

Titanium 206.9 46.7 277.5 808.3

Mother c

Aluminum 92.1 0 218.3 813.6

Silicon 23.2 5.2 37.0 138.1

Titanium 359.0 259.5 421.5 1394.3

Father d

Aluminum 68.4 23.2 129.9 537.4

Silicon 26.1 0.2 49.0 196.8

Titanium 624.9 198.7 835.0 2899.1

a For some study participants, estimation of soil ingestion resulted in a negative value. These estimates have been set to 0 mg/day for tabulation and analysis. b Results based on 12 children with complete food, excreta, and soil data.

c Results based on 16 mothers with complete food, excreta, and soil data. d Results based on 17 fathers with complete food, excreta, and soil data.

Soil ingestion in children and adults in the same familyDavis and Mirick

72 Journal of Exposure Science and Environmental Epidemiology (2006), 16(1)

amount of tracer excreted is reduced when tracer in urine is

ignored. Overall, mean and median ingestion estimates did

not change appreciably with the addition of the participants

with missing urine data (e.g., children’s mean soil ingestion

was 33.2 mg/day, 27.8 mg/day, and 313.8 mg/day for

aluminum, silicon, and titanium tracers, respectively).

Furthermore, the relationships between soil ingestion and

the selected behaviors described above did not change.

Discussion

This study was designed to meet several objectives, including

estimation of soil ingestion during normal activities in

children and their parents, identification of behavioral and

lifestyle factors potentially related to soil ingestion values,

comparison of adult and child soil ingestion estimates within

the same family, and investigation of gender differences in

soil ingestion by comparing estimates between the male and

female parent of the same family. The estimates of soil

ingestion for children in the present study are within the

ranges of estimates reported by previous studies (Binder

et al., 1986; Clausing et al., 1987; Calabrese et al., 1989; van

Wijnen et al., 1990; Stanek and Calabrese, 1995), and fall

within current federal guidelines which suggest the use of a

mean of 100 mg/day and upper confidence limit of 400 mg/

day (USEPA, 1996). Furthermore, these estimates are within

the range of those reported in the previous study of 104

children conducted by the investigative team (Davis et al.,

1990), of which these children were a subset. It should be

noted that the children in this study are somewhat older than

those in previous studies, which could result in lower soil

ingestion rates possibly due to decreased mouthing behavior

that might be expected with increasing age. With respect to

the aluminum and silicon tracers, soil ingestion estimates for

adults are in agreement with those from two smaller studies

of adult soil ingestion (Calabrese et al., 1990; Stanek et al.,

1997), as well as current federal guidelines which suggest a

mean of 50 mg/day (USEPA, 1996). No federal guidelines

have been set for an upper confidence limit of daily soil

ingestion in adults, due to the paucity of data in this area.

Interestingly, mothers’ and fathers’ mean estimated soil

ingestion was higher than the children’s, contrary to what

might be expected. However, median estimates are higher for

children than adults for aluminum and silicon, indicating that

adult estimates are somewhat skewed towards lower values.

Regardless of the tracer element used in the estimation,

estimated soil ingestion in the adults was considerably more

variable than for the children, possibly indicating an

important occupational contribution of soil ingestion in

some, but not all, of the adults.

Similar to previous studies, the highest ingestion estimates

resulted from using titanium as the tracer element of interest.

Toothpaste is a known source of titanium dioxide, and may

represent an important non-food, non-soil source of this

element. In the present study, every participant was provided

with a tube of toothpaste and instructed to use only this

toothpaste throughout the study period, and parents were

asked to monitor their child while brushing to minimize the

amount of toothpaste ingested. However, the titanium

concentration of the toothpaste provided is unknown,

making it difficult to determine whether the higher observed

soil ingestion estimates using titanium are the result of

toothpaste ingestion. A number of behaviors suspected to be

related to soil ingestion were investigated, and only two were

found to be associated with increased ingestion: among

children, reported eating of dirt was associated with the

amount of soil ingested, and for adults, occupational contact

with soil was associated with increased ingestion. In this

study, typical childhood behaviors thought to contribute to

soil ingestion, including thumb-sucking, furniture licking, or

carrying around a blanket or toy were not associated with

increased soil ingestion in the participant children. It should

be noted, however, that thumb-sucking and mouthing

furniture were each reported for only three of 19 children,

and the number was further reduced to two of 12 children for

thumb-sucking, and one of 12 children for furniture-

mouthing when results were restricted to those with complete

data for estimating soil ingestion. A behavior that one could

reasonably assume to be associated with soil ingestion, eating

unwashed fruits or vegetables, was not associated with soil

ingestion in either children or adults. However, the study

design required an equal amount of any food consumed,

including fruits and vegetables, to be included in the food

sample for analysis. Thus, assuming that the duplicate fruit

or vegetable sample was also unwashed, and since the

amount of a given tracer element in food is subtracted from

the amount of tracer in excreta in the ‘‘mass balance’’

approach to estimate soil ingestion (see Eq. (1)), consuming

unwashed fruits or vegetables would not have contributed to

an increase in estimated soil ingestion. Washing hands before

meals was associated with increased soil ingestion, although

these results are based on relatively few participants who do

not wash their hands before eating and for whom soil

ingestion could be estimated (two of 12 children, three of 16

mothers, and five of 17 fathers). It was initially thought that

hand-washing prior to meals would decrease soil ingestion,

but it is possible that those individuals who wash hands prior

to meals engage in activities involving contact with dirt

(presumably resulting in higher soil ingestion), necessitating

hand-washing before eating.

This study found that, within a given family, children’s soil

ingestion does not appear to be associated with either

parent’s ingestion, nor do mother’s and father’s soil ingestion

appear to be correlated. At the time of fieldwork, this was the

first study to investigate soil ingestion in both children and

adults, and at present, is the first such study to investigate soil

ingestion within a family. Given the relatively small number

Soil ingestion in children and adults in the same family Davis and Mirick

Journal of Exposure Science and Environmental Epidemiology (2006), 16(1) 73

of families for which within-family comparisons could be

made (10 families had complete ingestion data for all three

family members), it is difficult to conclude definitively

whether estimates of soil ingestion among family members

are independent from one another, or whether behaviors

shared among family members and also suspected to be

associated with soil ingestion make no contribution towards

estimates of ingestion within a family.

The ‘‘mass-balance’’ approach to estimate soil ingestion

relies on several underlying assumptions. First, it is assumed

that the tracer element used in the calculation is not well-

absorbed from the gastrointestinal tract, so that any tracer

consumed in food/liquid or medicines readily passes through

the digestive system and is excreted in the feces, but not urine,

of an individual (urinary excretion of a tracer indicates

absorption by the digestive tract). Thus, any remaining

amount of tracer is assumed to come from the ingestion of

soil. Further, an ideal tracer element should be present in

considerably greater concentrations in soil than in food/liquid

or medicines, in order to increase the likelihood that any

ingested tracer does indeed come from soil. In the present

study, the food/liquid samples had measurable amounts of all

three tracers of interest. Furthermore, some soil ingestion

estimates had negative values for some of the participants,

suggesting that some tracer may be absorbed in the

gastrointestinal tract. However, urinary concentrations of

the tracer elements (aluminum and titanium in particular)

were overall quite low, indicating that relatively little tracer

was absorbed in the digestive tract, and ingestion estimates

were similar when urine data were ignored in the calculation.

A second assumption is that children and parents ingest soil

and dust primarily from their own houses and yards, and that

the soil and house dust samples are representative of the

entire house and yard. Since sample collection occurred

during the summer months and participant children were not

in school, this may be a reasonable assumption for the

children. However, given that all but one of the fathers and

six of 19 mothers were employed outside the home, it seems

somewhat unlikely that the primary source of dust and soil

for these individuals would be around the house and yard,

and that any dust or soil ingested away from home would be

negligible. Soil samples were not collected beyond the house

and yard (i.e., no sample collection occurred at participants’

work places). It should be noted, however, that the range of

soil tracer element concentrations across families was quite

narrow, indicating that the soil composition around the area

is relatively uniform.

This was not the case for house dust: concentrations of

aluminum, silicon, and titanium varied widely across homes.

Although estimated soil ingestion did not differ significantly

for individuals when the primary source of ingested soil was

assumed to be house dust, slightly higher ingestion estimates

will result when house dust concentrations are used in the

Esoil/dust component of Eq. (1) (above), due to the overall

lower tracer concentrations relative to core and surface soil.

At the extreme, assuming 100% of the source of ingested soil

is house dust, ingestion estimates increased on average 24, 28,

and 14% for children using aluminum, silicon, and titanium

tracers, respectively. However, as discussed above, data

collection took place during the summer months, and it is

unlikely that children’s sole source of ingested soil came from

the indoors. Rather, it is more likely that actual ingestion

estimates fall somewhere between those for core (or surface)

soil and household dust.

For some participants, estimated soil ingestion was a

negative value. A negative value will result when the amount

of tracer present in food and medicines exceeds the tracer

amount present in urine and fecal matter for a given

individual. There are several possible explanations for such

a result. The results of the chemical analyses of tracers in each

sample are subject to a certain amount of error; given that the

dried food/liquid samples were many times heavier than

either the urine or fecal matter (Table 6), any overestimation

of tracer amount in the food samples could not be

compensated for by a similar overestimation in the tracer

amounts found in urine and feces. A second possibility, noted

above, is that some tracer is absorbed by the body, so that

the amount consumed in food/liquid or medicines is, in fact,

greater than the amount excreted in feces. Indeed, the results

of a number of studies indicate that aluminum, silicon, and

titanium can be absorbed in small amounts from the digestive

tract in adults (Schroeder et al., 1963; Tipton et al., 1966;

Schroeder, 1973; Carlisle 1974; Sorenson et al., 1974; Bowen

1979; Kelsay et al., 1979; Greger and Baier, 1983), and

absorption of tracer elements may be significantly greater in

children than adults, based on studies of lead metabolism

(NRC, 1980). Furthermore, the results presented here

indicate that at least some small fraction of tracer was

absorbed in the digestive tract, as evidenced by the presence

of tracers in the urine (silicon, in particular). A third

possibility is that the assumption of a 24-h lag time between

consumption of a tracer element in food and its subsequent

excretion is not entirely correct (i.e., gastrointestinal transit

time may be greater or less than 24 h for a given individual),

resulting in a misalignment between the ‘‘input’’ and

‘‘output’’ components of Eq. (1). Lastly, a negative soil

ingestion estimate could occur if the adjustment factors for

missing urine and fecal samples are inappropriately applied.

Every study participant provided complete food/liquid

samples; however, there were a number of incomplete urine

and fecal samples and, as described above, an adjustment

factor was calculated that inflated the actual dry weight of the

urine or fecal sample in proportion to the number of samples

missed. If the missed sample was heavier than the collected

samples, this adjustment would underestimate total sample

weight which, in turn, would underestimate the total amount

of tracer output in the urine and feces and potentially result

in a negative estimate for soil ingestion.

Soil ingestion in children and adults in the same familyDavis and Mirick

74 Journal of Exposure Science and Environmental Epidemiology (2006), 16(1)

In conclusion, this study found increased soil ingestion

associated with reported eating of dirt in children and

occupational exposure in adults. Although a number of

samples were not available for laboratory analyses and

limited the number of participants (children in particular) for

which soil ingestion could be estimated, it remains the largest

study conducted to date on soil ingestion in adults and is the

first to examine soil ingestion within a family. The participant

families were chosen from a subset of those in a previous

study of soil ingestion due to their high compliance with the

study protocol and, therefore, they may not be representative

of the general population. However, they were chosen for

their high compliance in order to maximize the quality of

data obtained from what is undoubtedly a difficult protocol

to follow. The success of the fieldwork demonstrates the

feasibility of conducting studies with complex data collection

methods, necessary for obtaining reliable soil ingestion

estimates in free-living individuals.

As such, the ingestion estimates presented here could aid in

refining current federal guidelines for soil ingestion in adults,

and add further support for current estimation guidelines in

children.

Acknowledgements

This research was supported by funding under a cooperative

agreement (R-82959101-0) with the U.S. Environmental

Protection Agency. We thank Laurie Shields, Cathy Kirk-

wood, and Christy Callahan for their valuable contributions to

the considerable data-related activities required of this project.

They are indebted to Dr Jacqueline Moya for her helpful

advice and reviews of earlier versions of this manuscript.

References

Binder S., Sokal D., and Maughan D. Estimating soil ingestion: the use of tracer

elements in estimating the amount of soil ingested by young children. Arch

Environ Health 1986: 41(6): 341–345.

Bowen H.J.M. Environmental Chemistry of the Elements. Academy Press, New

York, 1979.

Calabrese E.J., Barnes R., Stanek III E.J., Pastides H., Gilbert CE, Veneman P.,

Wang XR, Lasztity A., and Kostecki P.T. How much soil do young children

ingest: an epidemiologic study. Regul Toxicol Pharmacol 1989: 10(2): 123–137.

Calabrese E.J., Stanek E.J., and Gilbert C.E. Evidence of soil-pica behavior and

quantification of soil ingested. Hum Exp Toxicol 1991: 10(4): 245–249.

Calabrese E.J., Stanek E.J., Gilbert C.E., and Barnes R.M. Preliminary adult soil

ingestion estimates: results of a pilot study. Regul Toxicol Pharmacol 1990:

12(1): 88–95.

Calabrese E.J., Stanek E.J., James R.C., and Roberts S.M. Soil ingestion: a

concern for acute toxicity in children. Environ Health Perspect 1997a: 105(12):

1354–1358.

Calabrese E.J., Stanek E.J., Pekow P., and Barnes R.M. Soil ingestion estimates

for children residing on a superfund site. Ecotoxicol Environ Saf 1997b: 36(3):

258–268.

Carlisle E.M. Proceedings: silicon as an essential element. Fed Proc 1974: 33(6):

1758–1766.

Clausing P., Brunekreef B., and Van Wijnen J.H. A method for estimating soil

ingestion by children. Int Arch Occup Environ Health 1987: 59(1): 73–82.

Davis S., Waller P., Buschbom R., Ballou J., and White P. Quantitative estimates

of soil ingestion in normal children between the ages of 2 and 7 years:

population-based estimates using aluminum, silicon, and titanium as soil tracer

elements. Arch Environ Health 1990: 45(2): 112–122.

Greger J.L., and Baier M.J. Excretion and retention of low or moderate levels of

aluminum by human subjects. Food Chem Toxicol 1983: 21: 473–477.

Kelsay J.L., Behall K.M., and Prather E.S. Effect of fiber from fruits and

vegetables on metabolic responses of human subjects: II. Calcium, magnesium,

iron, and silicon balances. Am J Clin Nutr 1979: 32: 1876–1880.

National Research Council. Lead in the Human Environment. National Research

Council, Washington, DC, 1980.

Schroeder H.A. The Trace Elements and Man. Devim Adair Co., Connecticut,

1973.

Schroeder H.A., Balassa J.J., and Tipton I.H. Abnormal trace elements in man:

titanium. J Chron Dis 1963: 16: 55–69.

Sorenson J.R., Campbell I.R., Tepper L.B., and Lingg R.D. Aluminum in the

environment and human health. Environ Health Perspect 1974: 8: 3–95.

Stanek III E.J., and Calabrese E.J. Daily estimates of soil ingestion in children.

Environ Health Perspect 1995: 103(3): 276–285.

Stanek III E.J., Calabrese E.J., Barnes R., and Pekow P. Soil ingestion in adults

F results of a second pilot study. Ecotoxicol Environ Saf 1997: 36(3): 249–257.

Tipton I.H., Stewart P.L., and Martin P.G. Trace elements in diets and excreta.

Health Phys 1966: 12: 1683–1689.

U.S. Environmental Protection Agency. Exposure Factors Handbook. Chapter 4:

Soil Ingestion and Pica. Office of Health and Environmental Assessment,

Washington, DC, 1996.

van Wijnen J.H., Clausing P., and Brunekreef B. Estimated soil ingestion by

children. Environ Res 1990: 51(2): 147–162.

Wong M.S., Bundy D.A., and Golden M.H. Quantitative assessment of

geophagous behaviour as a potential source of exposure to geohelminth

infection. Trans R Soc Trop Med Hyg 1988: 82(4): 621–625.

Soil ingestion in children and adults in the same family Davis and Mirick

Journal of Exposure Science and Environmental Epidemiology (2006), 16(1) 75