Prediction and Evaluation of Dietary Arsenic Exposure
Estimates Using the SHEDS Model

Jianping Xue, Valerie Zartarian, Hal ilk Ozkaynak

U.S.EPA/GRD National Exposure Research Laboratory, RTP, NC, USA
Presented at ISEA 2006, Paris, France

6 Q %

I

Introduction

Method

Epidemiological studies reveal that arsenic can cause
human cancers and many other adverse health effects.
Ingestion of food especially seafood is a primary route for
human arsenic exposure. Foods from endemic areas have
higher arsenic concentrations. Sources of arsenic are plant
or animal intake and/or contamination from food
processing. An accurate probabilistic human exposure
model for arsenic ingestion can help reduce uncertainty in
arsenic risk assessments.

Table 1 As residue concentration (mg/kg) from selected food items

k, pan-cooked

2.336 5.643

p50 p5 p25 p75 p95

10.430 1 0.430

rice infant cereal, instant, prepared witl"

carrots, strained/junior

chicken, fried (breast, leg, and thigh) he

chicken, fried (breast, leg, and thigh), fa

granola cereal

oatrino cereal

0.024 0H20 0.023

1)	Collect and process available As residue and food consumption data

•	Continuing survey of food intake by individuals (CSFII) 1994-96/98 by USDA

•	Recipe files from OPP

•	Total diet survey (TDS) by FDA

2)	Apply EPA/ORD's Stochastic Human Exposure and Dose Simulation (SHEDS) dietary
module using As data

•	Process As residue data

•	Match CSFII data with residues by food items

•	Match CSFII data with residues by food commodities

•	Process measurements below limit of detection (LOD)

•	Apply Processing factors

•	Calculate food residue intake

3)	Conduct variability and uncertainty runs with SHEDS

•	Randomly draw subjects from CSFII

•	Randomly draw As residue

•	Repeat runs for stability

•	Bootstrap various portions of CSFII

•	Bootstrap various portion of residues

Table 2 Contribution of Major food categories to As dietary exposure

0.400 7.6923
0.086 73.077
0!083 76.923

79.3M> 16.7%	4.0%

84.7% 10.0%	5.4%

62^3J/o 35.5%	2.2%

66:7% 29.9%	3.4%

| CSFII |	| Residue |



1 food Hem in

„„l











1 Rtc*
|no |1—

.„.l



1







| Modified Matched PatH |
| Food Intake Calculation"]

Figure 3. Uncertainty of daily dietary exposure
of As with 1/8 for As residue and 1/30 CSFII
bootstrap

1.0 118.6
1.6 130.6
4.2 147.3

Uncertainty Percentile

SHEDS dietary module flowchart

Figure 1.

Table 3 As dietaiy intake from SHEDS evaluated with NHEXAS duplicate data

Figure 2. Dietary Model evaluation*

Table 4 Uncertainty Analyses on CSFII and As residue data

n mean

p50 p5 p25 p75 p95

NHEXAS

MODEL 5th by Mean
MODEL 50th by Mean
MODEL 95th by Mean

MODEL 5th by Mean
MODEL 50th by Mean
MODEL 95th by Mean

Results

156 0.185 0.300 0.095 0.019 0.049	0.174

Fill in no-detects with zero

154 0.080 0.305 0.000 0.000 0.000	0.031

154 0.152 0.406 0.009 0.000 0.000	0.066

154 0.260 1.168 0.007 0.000 0.000	0.055
Fill in no-detects with 1/2 LOD when As residue mean >

154 0.128 0.399 0.051 0.004 0.023	0.110

154 0.192 0.561 0.052 0.004 0.024	0.115

154 0.300 1.293 0.062 0.009 0.029	0.128

Uncertainly Ratio (95th vs 5 th)

Bootstrap

50th

95th

99th

CSFII 1/20 bootstrap

1.19

1.93

3.28

As 1 iA and CSFII 1/10 bootstrap

1.20

1.66

2.43

As 1 iA and CSFII 1 /20 bootstrap

1.24

2.03

3.40

CSFII 1/8 bootstrap

1.14

1.51

2.14

As 1 /8 bootstrap

1.20

1.31

1.73

As 1 /8 and CSFII 1/10 bootstrap

1.26

1.69

2.52

As 1 /8 and CSFII 1 /20 bootstrap

1.30

1.99

3.87

As 1 /8 and CSFII 1/30 bootstrap

1.39

2.22

4.47

Conclusions

Seafood has 100% detection rate with high As residue concentrations; Rice also has high As
residues (see Table 1)

Asian has higher As intake from dietary than other races (Table 2)

Seafood is dominant contributor; for Asian, rice accounts for 22% high percentile and 35% for
whole (T able 2)

200 repeated runs with SHEDS dietary matched with region, race, age, and gender show good fit
with NHEXAS data (Figure 2 and Table 3)

Model uncertainty ranged from 1.5 to 5 times for 50th, 95th and 99th by using ratio of 95th to 5th
percentiles. The higher percentile, the higher uncertainty and uncertainty is more sensitive to
CSFII (Figure 3 and Table 4)

Seafood and rice are major sources of
dietary exposure to As
SHEDS dietary model performs well in
evaluation of model against duplicate dietary
survey

Uncertainty analyses indicated dietary
survey data has more uncertainty

References:

J Xue et al, A probabilistic arsenic exposure assessment for children who contact chromated copper arsenate (CCA)-treated playests
And decks, part2: sensitivity and uncertainty analyses; Risk Analysis,2006, Vol. 26, No. 2 :533-541

P. Barry Ryan et al, Analysis of dietary intake of selected metals in the NHEXAS-Maryland investigation, Environmental Health
Perspectives, 2001, Vol. 109, No. 2: 122-128

The United States
Environmental Protection
Agency through its Office of
Research and Development
funded and managed the
research described here. It has
been subj ected to Agency
review and approved for
publication.


-------