EPA
Trends in Blood Mercury Concentrations
and Fish Consumption
Among U.S. Women of Childbearing Age
NHANES, 1999-2010
Final Report
July 2013
EPA-823-R-13-002
Hilt
-------
ACKNOWLEDGMENTS
The U.S. Environmental Protection Agency (EPA) Project Manager for this project was Jeffrey D.
Bigler, who provided oversight and technical direction for the project. EPA was supported in this
effort by WESTAT, Inc under EPA Contract No. EP-C-10-023. The WESTAT project manager
was Rebecca Jeffries Birch who was responsible for the overall technical accuracy and quality of the
project. The methodology developed for the project was externally peer reviewed by the Eastern
Research Group, Inc. under EPA Contract No. EP-C-07-059.
-------
Table of Contents
Chapter Page
Executive Summary vi
1 Background and Purpose 1
2 Methods 4
2.1 Methods Overview 4
2.2 NHANES Data Overview 4
2.3 Blood Total and Inorganic Mercury Concentration Data 6
2.3.1 Adjustments for Non-Detects in Total and
Inorganic Mercury 8
2.3.2 Calculation of MeHg Concentration 8
2.4 30-Day Consumption Frequency Data 9
2.5 24-Hour Dietary Recall Data 9
2.6 Fish Tissue Mercury Data 10
2.7 Estimation of 30-Day Fish Consumption and Mercury
Intake 12
2.8 Statistical Analysis Methods 14
2.8.1 Calculation of Percentiles 15
2.8.2 Regression Analysis Predicting Blood Mercury
Concentrations 15
2.8.3 Logistic Regression Predicting MeHg > 5.8ug/L 18
2.8.4 Modeling Factors Contributing to Mercury
Intake 18
2.8.5 Trends 19
3 Results 20
3.1 Trends in Blood Mercury Concentrations 20
3.2 Trends in Fish Consumption 22
3.2.1 Trends in Frequency of Consumption 22
3.2.2 Trends in Estimated Amounts Consumed Over
the Previous 30 Days 24
3.3 Associations Between Fish Consumption Frequency and
Blood Mercury Concentrations 26
in
-------
Table of Contents
(continued)
Chapter Page
3.4 Regression Analysis Results: Associations between Blood
Mercury, Fish Consumption, Time, and Demographic
Factors 30
3.5 Regression Analysis Results: Associations between Fish
Consumption and Intake of Mercury with Time and
Demographic Factors 34
4 Discussion 43
5 Conclusions 47
6 References 48
Appendix A A-l
Table
1. Summary of trends in blood mercury and fish consumption,
women aged 16-49 years, NHANES 1999-2010 vm
2. Scope of data collection by data type and survey release 5
3. Number of participating women aged 16-49 by data file and
survey release 6
4. Laboratory analysis methodology for blood total mercury and
inorganic mercury 7
5. Percent of women aged 16 to 49 years with blood MeHg > 5.8
ug/L, by NHANES survey release 22
6. Parameter estimates and relative ratios from the non-linear model
predicting blood MeHg concentrations 31
7. Parameter estimates and odds ratios from the logistic model
predicting the probability of blood MeHg concentrations over 5.8
33
Figure
1. Distribution of blood THg (ug/L), by NHANES survey release,
women aged 16-49 years 20
2. Distribution of blood MeHg (ug/L), by NHANES survey release,
women aged 16-49 years 21
IV
-------
Table of Contents
(continued)
3. Percent of participants by 30-day fish consumption frequency, by
NHANES survey release, women aged 16-49 years 23
4. Percent of participants by 30-day fish consumption frequency, by
demographic characteristics, women aged 16-49 years, NHANES
1999-2010 24
5. Estimated mean and 90th percentile amounts of fish consumed in
30 days, women aged 16-49 years, NHANES 1999-2010 (with
95% confidence intervals) 25
6. Estimated mean and 90th percentile amounts of mercury ingested,
normed to body weight (ug/kg) in 30 days, women aged 16-49
years, NHANES 1999-2010 (with 95% confidence intervals) 26
7. Distribution of blood THg (ug/L), by reported frequency of fish
consumption in 30 days, women aged 16-49 years, NHANES
1999-2010 27
8. Distribution of blood MeHg (ug/L), by reported frequency of
fish consumption in 30 days, women aged 16-49 years, NHANES
1999-2010 27
9. Mean blood MeHg concentrations by reported frequency of fish
consumption in 30 days, woemn aged 16-49 years, NHANES
1999-2010 (with 95% confidence intervals) 29
10. Relative ratios and 95% confidence limits from the models
predicting fish consumption and mercury intake variables versus
race/ethnicity 36
11. Extract of the full plot of fish and mercury variables versus
race/ethincity 38
12. Relative ratios and 95% confidence limits from the models
predicting fish consumption and mercury intake variables versus
age 40
13. Relative ratios and 95% confidence limits from the models
predicting fish consumption and mercury intake variables versus
income 41
14. Relative ratios and 95% confidence limits from the models
predicting fish consumption and mercury intake variables versus
NHANES survey release 42
-------
Executive
Summary
This report documents an analysis of the 1999-2010 NHANES data on the distribution of blood
mercury concentration in women of childbearing age, their finfish/shellfish consumption, and
mercury intake and the association of these with time, age, race/ethnicity, income, and, for blood
mercury data, finfish/shellfish consumption. Note that, unless otherwise specified, "fish" refers to
both finfish and shellfish species. One goal of the EPA 2011-2015 Strategic Plan is to protect public
health by making fish and shellfish safer to eat. One of the primary risks of consumption of
contaminated fish and shellfish is exposure to methyl mercury (MeHg). Exposure to MeHg in
people in the U.S. is almost exclusively through the consumption offish and shellfish (NRC, 2000).
This report is in support of the goal of making fish and shellfish safer to eat. EPA's approach to
making fish safer to eat includes several key elements:
Encourage development of statewide mercury reduction strategies.
Reduce air deposition of mercury.
Improve public information and notification offish contamination risks.
One of the specific measures the Agency uses to estimate progress associated with this goal is the
measurement of blood mercury concentrations in women-of-childbearing age as reported by the
Centers for Disease Control and Prevention's (CDC) National Center for Health Statistics (NCHS)
in the National Health and Nutrition Examination Survey (NHANES).
For this analysis, EPA developed a methodology for assessing trends over time in NHANES blood
mercury data and used it to investigate whether there was a trend over time in the distribution of
blood mercury concentrations in women of childbearing age who participated in the 1999-2010
NHANES and their fish consumption, after statistically adjusting for the covariates known to
influence blood mercury concentrations. Each two-year NHANES survey (1999-2000, 2001-2002,
2003-2004, 2005-2006, 2007-2008, 2009-2010) consists of an independent, nationally representative
sample.
The analyses found blood mercury concentrations in NHANES survey release 1999-2000 to be
significantly higher than the mean of the subsequent releases for both blood THg and blood MeHg.
-------
The geometric mean blood THg in 1999-2000 was 1.21 times higher than the geometric mean across
the subsequent 10 years (2001-2010), representing an 18 percent decrease between 1999-2000 and
2001-2010. For blood MeHg, the geometric mean in 1999-2000 was 1.51 times higher than the
geometric mean across the subsequent 10 years. This represents a decrease of 34 percent between
1999-2000 and 2001-2010. Additionally, the percentwith THg >5.8 ug/L and MeHg >5.8 ug/L is
significantly higher in survey release 1999-2000. The percentage of women of reproductive age with
blood THg over 5.8 ug/L in 1999-2000 was 2.64 times that found in 2001-2010, a decrease of 62
percent between 1999-2000 and 2001-2010. For blood MeHg, the percent of women of
reproductive age over 5.8 ug/L in 1999-2000 was 2.86 times higher than the percent of women in
2001-2010, representing a 65 percent decrease between 1999-2000 and 2001-2010. The analysis also
found a significant quadratic trend in blood MeHg concentration since 1999-2000. This quadratic
trend indicates decreasing blood MeHg concentrations between NHANES survey release 2001-2002
and 2003-2004, followed by relatively small changes and a slight increase in the last years.
There was a significant relationship between mercury intake from fish consumption and blood
mercury, although mercury intake did not fully explain the differences observed across the survey
releases. The analysis showed few changes in fish consumption and mercury intake over the study
period. There was a marginally statistically significant decreasing trend across NHANES survey
releases in the ratio of mercury intake to fish consumed that is consistent with women shifting their
consumption to fish with lower mercury concentrations; however, other studies are needed to
determine 1) if there is a link between changing consumption patterns and blood mercury and 2) if
fish advisories have led to the changing consumption patterns
Demographic characteristics were associated with blood mercury as expected: higher concentrations
observed with increasing age and income and higher concentrations observed in the "other" race
category and lower concentrations observed in Mexican Americans. Similar patterns between fish
consumption and demographic characteristics were found. Table 1 presents a summary of results.
Vll
-------
Table 1.
Summary of trends in blood mercury and fish consumption, women aged 16-49 years, NHANES 1999-2010
NHANES Survey Release
1999-2000
2001-2002
2003-2004
2005-2006
2007-2008
2009-2010
1999-2000
2001-2002
2003-2004
2005-2006
2007-2008
2009-2010
1999-2000
2001-2002
2003-2004
2005-2006
2007-2008
2009-2010
Blood MeHg, ug/L
Blood THg, ug/L
Geometric Mean (95% Cl)
0.94(0.74,1.19)
0.71(0.57,0.90)
0.56 (0.40,0.78)
0.60 (0.44,0.82)
0.55(0.40,0.75)
0.69(0.56,0.86)
1.01 (0.84,1.23)
0.83 (0.74,0.93)
0.82 (0.72,0.93)
0.89 (0.80,0.99)
0.79 (0.70,0.87)
0.86 (0.77,0.95)
90* %ile (95% Cl)
4.56 (3.48,5.97)
2.84(2.48,3.25)
2.61(2.08,3.26)
2.70(2.31,3.16)
2.40 (1.88,3.07)
2.75(2.49,3.04)
4.81 (3.79,6.10)
3.05 (2.69,3.46)
3.10 (2.57,3.73)
3.14 (2.81,3.51)
2.73 (2.24,3.33)
3.11 (2.85,3.39)
Fish consumed in 30
days,g
MeHg intake in 30
days per unit body
weight, Mi/kg
Blood MeHg for
women who ate
Blood MeHg for
women who ate
fish 6+ times in 30
days, ug/l_
Arithmetic Mean (95% Cl)
254.6 (213.4,295.8)
310.5 (275.0,345.9)
270.2 (235.3,305.2)
322.5(277.1,367.8)
259.0(228.5,289.6)
308.5(269.3,347.8)
0.45(0.36,0.54)
0.54(0.41,0.67)
0.44(0.37,0.51)
0.46 (0.39,0.52)
0.42 (0.34,0.49)
0.45(0.40,0.51)
0.61(0.50,0.72)
0.43 (0.33,0.54)
0.38(0.27,0.50)
0.37 (0.25,0.48)
0.36 (0.25,0.47)
0.50 (0.40,0.60)
3.36(2.76,3.97)
2.34(1.92,2.75)
2.07 (1.68,2.46)
1.84(1.61,2.08)
1.95(1.54,2.37)
2.11(1.87,2.35)
90* %ile (95% Cl)
663.3(567.8,802.6)
717.7 (639.4,789.7)
646.8(575.6,769.7)
792.0(672.9,960.9)
653.7 (567.8,796.6)
768.0 (672.8,868.2)
1.13 (0.95,1.43)
1.13 (1.04,1.30)
1.10 (0.90,1.28)
1.16 (0.99,1.30)
1.10 (0.93,1.43)
1.15(1.05,1.29)
Percent (SE) of Participants by Frequency of Fish Consumption
0 times
21.8 (2.3)
16.9 (1.1)
21.3 (1.6)
20.1(2.2)
24.7 (1.5)
20.2 (1.9)
Itime 2 times
14.7 (1.1) 13.6 (1.2)
13.3 (1.4) 13.5 (1.6)
12.5 (1.7) 13.0 (1.1)
10.7 (1.1) 11.9 (0.8)
14.8 (1.0) 11.4 (0.7)
14.5 (0.9) 10.5 (0.6)
3 times
8.9 (1.6)
10.4 (1.0)
9.8 (0.9)
9.4(0.9)
9.4(0.9)
7.8(0.8)
1.32 (1.09,1.61)
0.92 (0.77,1.09)
0.79(0.57,1.10)
0.77 (0.60,0.98)
0.67 (0.54,0.83)
1.01 (0.73,1.39)
in 30 days
4-5 times
14.6 (1.0)
16.9 (1.2)
14.4 (1.2)
13.8 (1.1)
12.9 (1.0)
15.2 (1.1)
8.32 (6.26,11.05)
5.69 (4.28,7.57)
4.36 (3.21,5.92)
4.07 (3.34,4.95)
4.21(2.96,5.99)
4.24(3.54,5.08)
6+ times
26.4(3.0)
28.9 (1.5)
29.0(2.0)
34.2 (3.0)
26.9 (1.5)
31.8 (1.6)
Vlll
-------
Table 1.
Summary of trends in blood mercury and fish consumption, women aged 16-49 years, NHANES 1999-2010 (continued)
Overall p-value
1999-2000 different from others2
Linear trend after 1999-20003
Quadratic trend after 1999-20004
1999-2000 diff. from quadratic
trend5
Linear Trend 1999-2010 p-value3
Parameter Estimates (p-value) from Regression Modeling
Predicting log-
transformed
blood MeHg
(Mg/L)
<.0001
0.4657
(<.0001)
-0.0051(0.71)
0.0492 (0.004)
0.1062 (0.35)
Predicting % with blood
MeHg over 5.8 Mg/L
<.0001
1.433 (<.0001)
-0.0865(0.25)
0.0525(0.45)
0.8055(0.18)
Predicting
frequency of fish
consumption in 30
days, among
consumers1
0.0052 (0.37)
Predicting amount
of fish consumed in
meal (g), among
consumers1
-0.0013 (0.46)
Predicting Hg
concentration of
fish consumed (ug),
among consumers1
-0.0110 (0.035)
Predicting est. 30-
day Hg intake
(ug/kg), among
consumers1
-0.0086 (0.35)
iConsumers are defined as any participant who reported consumption of any finfish or shellfish in the previous 30 days.
2The parameter estimate is the difference in the mean natural-log transformed concentration between the 1999-2000 release and later releases.
3The parameter estimate is the change in the mean natural-log transformed dependent variable from one survey release to the next.
4A positive value indicates a concave curve (higher in early and late survey releases and in lower in the middle releases).
5The parameter estimate is the difference in the mean natural-log transformed concentration for the 1999-2000 release and the predicted value extrapolated from the trend
across later releases.
IX
-------
Background and Purpose
The National Water Program Guidance for fiscal year (FY) 2012 describes how the U.S.
Environmental Protection Agency (EPA), states, and tribal governments will work together to
protect and improve the quality of the nation's water, including wetlands, and ensure safe drinking
water (U.S. EPA, 2011). The Guidance describes the key actions needed to accomplish the public
health and environmental goals proposed in the EPA 2011-2015 Strategic Plan (U.S. EPA, 2010).
One goal is to protect public health by making finfish and shellfish safer to eat. Note that, unless
otherwise specified, "fish" refers to both finfish and shellfish. One of the primary risks from eating
fish is exposure to methylmercury (MeHg). In the U.S., exposure to MeHg in humans is largely
through the consumption offish (NRC, 2000). Mercury released into the environment is converted
to MeHg in sediments and in the water column and bioaccumulates through aquatic food webs. This
bioaccumulation leads to increased levels of MeHg in larger, older, predatory fish; concentrations in
fish tissue may exceed a million fold the concentrations in water (NRC, 2000). MeHg exposure in
utero is associated with adverse health effects, e.g., neurodevelopmental deficits such as IQ and
motor function deficits in children (Mergler et al, 2007; NRC, 2000). In 2004 the FDA and EPA
issued consumer advice. The advisory offers advice on amounts of commercial fish and fish from
local water bodies that are safe to consume. This report investigates national trends over time in
both blood mercury concentrations and fish consumption for women 16-49 years of age.
The EPA's approach to making fish safer to eat includes several key elements:
Encourage development of statewide mercury reduction strategies.
Reduce air deposition of mercury.
Improve public information and notification offish contamination risks.
One of the specific measures the Agency uses to estimate progress associated with this goal is blood
mercury concentrations in women-of-childbearing age as reported by the Centers for Disease
Control and Prevention's (CDC) National Center for Health Statistics (NCHS) in the National
Health and Nutrition Examination Survey (NHANES). Blood total mercury (THg) concentrations
reflect exposure to organic mercury, predominantly MeHg, from consumption offish (Bjornberg et
al. 2003; Sanzo et al. 2001; Svensson et al. 1992). NHANES is a continuing survey designed to
-------
collect data on the health and nutritional status of the U.S. population. The NHANES reports
include information on chemicals or their metabolites as measured in blood samples collected from
a statistically representative sample of the U.S. population. CDC releases the NHANES data every
two years and reports environmental exposure results every two years in the National Report on
Human Exposure to Environmental Chemicals (CDC, 2010a).
EPA developed a methodology to assess trends over time in the NHANES blood mercury data.
This methodology was peer reviewed. Comments and concerns about the methodology received by
the peer reviewers were addressed and the methodology was revised. The analyses discussed in this
report use this revised methodology.
Previous work on blood mercury trends includes the publication by Mahaffey, Clickner, and Jeffries
(2009), who analyzed NHANES 1999-2004 data for evidence of trends in blood mercury levels in
women aged 16-49 years. The authors found no statistically significant difference among the three
sets of study years (1999-2000, 2001-2002, and 2003-2004) for blood Hg, estimated 30-day Hg
intake, or reported frequency of fish consumption. However, in multiple regression modeling,
adjusting for covariates including coastal/non-coastal residence, women aged 16-49 who
participated in the years 1999-2000 had significantly higher blood Hg levels compared with those
who participated in 2003-2004.Women who participated in 2001-2002 had significantly lower blood
Hg levels than those who participated in 2003-2004. Although the analyses did not support the
conclusion that there was a general downward trend in blood Hg concentrations over the 6-year
study period, there was a decline in the upper percentiles reflecting the most highly exposed women
with blood Hg concentrations greater than established levels of concern. They observed a decrease
in the 90th percentile of 30-day estimated intake of Hg through fish consumption across the six
study years even though there was no similar decrease in the 90th percentile of 30-day estimated
consumption of grams offish. This suggested a shift in consumption to fish containing less Hg.
They did not observe a similar pattern at the mean, suggesting that this shift in fish consumption
occurred mainly with the highest fish consumers.
Caldwell, et al., 2009, reported finding no differences across NHANES study years in blood total
mercury concentrations for the subpopulation of women 16-49 years in NHANES 1999-2006, after
adjusting for age and race/ethnicity (p=0.11).
The goals of this report are to investigate differences over time in blood mercury concentrations in
women of child-bearing age using NHANES 1999-2010 data and to investigate changes in fish
consumption and mercury intake over time. This report documents an analysis of the 1999-2010
-------
NHANES data on the distribution of blood mercury concentrations and the association of these
with time, age, sex, race/ethnicity, income, and fish consumption. Regression analysis was used to
assess whether or not there are significant differences in blood mercury concentrations across the six
study segments (1999-2000, 2001-2002, 2003-2004, 2005-2006, 2007-2008, 2009-2010) based on
national estimates, after adjustment for fish consumption and demographic covariates known to be
associated with blood mercury concentrations (Schober et al., 2003; Mahaffey et al., 2004; Mahaffey
et al., 2009; Caldwell et al., 2009). Additional analysis was done to assess whether or not there are
significant differences in fish consumption and mercury intake across the study period.
-------
Methods
2.1 Methods Overview
Previous research on blood mercury indicates that the predictors of blood mercury concentrations
include fish consumption, age, race/ethnicity, income, and other variables. There are two sources of
information on dietary fish intake in the NHANES data. One source is the 24-hour recall data with
information on the quantity of food consumed and the second source is the report of 30-day
frequency of consumption of selected fish species. These were combined to produce useful
estimates of the amounts offish consumed by a participant. These estimates were used to develop
population-based statistics on fish consumption
The analysis also combined NHANES measurements of blood mercury with NHANES
measurements of fish consumption and measurements of mercury concentrations in fish to
investigate the relationship between mercury intake from fish and blood MeHg concentrations
across six NHANES survey releases covering 12 years. The following sections describe the data and
the analysis procedures. All data processing and analyses were performed using the Statistical
Analysis System (SAS) version 9.2 (SAS Institute, 2010).
2.2 NHANES Data Overview
The required NHANES data files and variables were identified and downloaded from the NHANES
website (CDC 201 Oa). These files were then merged to create a dataset customized to the needs of
this project. For each NHANES survey release, the required data are in the following files:
Demographics - gender, age, race/ethnicity, education, income, sampling weights,
pseudo-stratum, and pseudo-primary sampling unit (PSU). The pseudo-stratum and
pseudo-PSU variables provide information on how participants were selected and are
needed to calculate standard errors and p-values. They are modified from the actual
NHANES strata and PSUs for disclosure control, and are thus prefixed "pseudo."
Laboratory results - blood total mercury concentration and blood inorganic mercury
concentration.
-------
Body measures - body weight.
Dietary intake, 24 hour recall - food codes, meal name, amount eaten; one record per
food item eaten.
Dietary intake, 30 day frequency of consumption - number of times each of the
following species was reported consumed in the previous 30 days: clams, crabs, crayfish,
lobster, mussels, oysters, scallops, shrimp, other shellfish, other unknown shellfish,
breaded fish products, tuna (not differentiated by canned light, canned white or steaks),
bass, catfish, cod, flatfish, haddock, mackerel, perch, pike, pollock, porgy, salmon,
sardines, sea bass, shark, swordfish, trout, walleye, other fish, and other unknown fish.
Over the six two-year periods from 1999 to 2010, NHANES has changed the scope of participants
for some types of data collected. Table 2 summarizes these changes. Across survey releases,
consistent data are available for children aged 1 to 5 and women aged 16 to 49. This analysis uses the
data for women ages 16 to 49. Note there were no differences pertinent to this analysis between
2005 and 2010.
Table 2.
Scope of data collection by data type and survey release
Data Type
Blood mercury
Demographics
Body weight
Fish
consumed,
30-day
frequency
Diet, 24-hr
intake
NHANES Survey Release
1999-2000
Both sexes aged 1-
5 years, women
aged 16-49 years
All participants
All participants
Both sexes aged 1
year and older
All participants,
one day's intake
2001-2002
Both sexes aged 1-
5 years, women
aged 16-49 years
All participants
All participants
Both sexes aged 1-
5 years, women
aged 16-49 years
All participants,
one day's intake
2003-2004
Both sexes aged 1
year and older
All participants
All participants
Both sexes aged
1-5 years, women
aged 16-49 years
All participants,
two days' intake
(second day by
phone)
2005-2010
Both sexes aged 1
year and older
All participants
All participants
Both sexes aged 1
year and older
All participants,
two days' intake
(second day by
phone)
The relevant files were downloaded from the NHANES website. The unique identifier variable
"SEQN" was used to link the data on each participant from different data sets. The analysis was
performed following the NHANES Analytical Guidelines posted on the NHANES website (CDC
2010b). The sample sizes number of unique SEQN's for women aged 16-49 with data for this
study for each of these files are in Table 3. The sample sizes vary by NHANES survey release due
to sample design changes over time and because some participants failed to provide all of the
-------
requested information. The final row of the table indicates the sample size used in the analyses
contained in this report.
Table 3. Number of participating women aged 16-49 by data file and survey release
Number of
participants
(unique SEQN)
Demographic
file
Blood Hg1
Body weight
Diet (24-h)2
Diet(SO-d)2
All Data
Elements of
Interest
1999-00
1,944
1,707
1,809
1,732
1,731
1,637
2001-02
2,140
1,906
1,966
1,933
1,931
1,780
2003-04
1,900
1,704
1,802
1,722
1,721
1,599
2005-06
2,085
1,873
1,992
1,920
1,918
1,792
2007-08
1,749
1,583
1,681
1,625
1,625
1,493
2009-10
1,996
1,868
1,939
1,865
1,864
1,786
Total,
1999-2010
11,814
10,641
11,189
10,797
10,790
10,087
10ne outlier was removed from the analysis dataset
2Counts of participants whose dietary recall status was reliable and met minimum criteria by NHANES
2.3 Blood Total and Inorganic Mercury Concentration Data
The laboratory data files contain total mercury and inorganic mercury. Table 4 summarizes the
laboratory methods used to analyze the blood samples for mercury.
In blood samples with low levels of mercury, total and/or inorganic mercury concentration
measurements below the laboratory lower detection limit (LDL) are reported as less than the
detection limit without providing a measured concentration. In the data files, the unspecified
concentrations for these samples are replaced by a substitute value. For the analysis, the LDL was
assumed to equal the substitute value in the data file times the square root of 2. This approach
appears consistent with the stated procedure for 1999-2000 and with the distribution of the detected
values. However, for total mercury in the 2007-2008 file, the substitute value times the square root
of 2 is 0.28 and the smallest reported detected value is 0.33. To be consistent with the distribution of
the reported values and because there are detected values of 0.33, the LDL was set to 0.325. Also,
for the 1999-2000 data, the substitute value in the file is rounded to one digit after the decimal place;
the calculations used the reported substitute value of 0.097 for total and 0.032 for inorganic
mercury. The different and sometimes multiple substitute values suggest that different detection
limits were applicable for different years or samples, contrary to what is implied by the reported
LDL in Table 4.
-------
Table 4.
Laboratory analysis methodology for blood total mercury and inorganic mercury
Parameter
Laboratory method3
Data file
Reportable range3
Survey Release
1999-2000 2001-2002
Flow Injection Cold Vapor Atomic
Absorption (CVAA)
Method A/o..- 1190B/06-OD
Lab06, 106_b
LDLto50|jg/L
2003-2004
Blood total mercury:
ICPDRCMS
Method A/o: ITB001A
Blood inorganic mercury6:
FIMS CVAA (formerly Flow
Injection Cold Vapor Atomic
Absorption (CVAA)).
Method A/o: ITB003A
(formerly:1190B/06-OD)
L06bmt_c
Above LDL
2005-2006
Blood total mercury:
ICPDRCMS
Method A/o: ITB001A
Blood inorganic
mercury: FIMS CVAA
(formerly Flow
Injection Cold Vapor
Atomic Absorption
(CVAA)).
Method A/o: ITB003A
(formerly:1190B/06-
OD)
Thgigh_d
LDL to 50 ug/L
2007-2008
Blood total mercury:
ICPDRCMS
Method A/o: ITB001A
Blood inorganic mercury:
FIMS CVAA (formerly
Flow Injection Cold Vapor
Atomic Absorption
(CVAA)).
Method A/o: ITB003A
(formerly:1190B/06-OD)
Thgihg_e
LDL to 50 ug/L
2009-2010
Blood total mercury:
ICPDRCMS
Method A/o: ITB001A
Blood inorganic
mercury: FIMS CVAA
(formerly Flow Injection
Cold Vapor Atomic
Absorption (CVAA)).
Method A/o: ITB003A
(formerly:1190B/06-
OD)
ThgihgJ
LDL to 50 ug/L
Lower Detection Limits (LDL)a:
Total Hga
Inorganic Hga
Substitute value for results
below the LDL*:
BDL indicator*
No. of digits after the decimal
Eligible Participants
0.137 ug/L (3*std of 10 runs of a low
Hg level sample)
0.446 ug/L (3*std of 10 runs of a low
Hg level sample)
DL over square root DL over square
of 2 root of 2"
Total Hg: 0.097 Total Hg: 0.07
ug/L, Inorganic Hg: ug/L, Inorganic
0.32 ug/L Hg: 0.28 ug/L
NA
1
Both sexes ages 1-5, Females 16-49
0.2" ug/L (3*std of >20 run
of blood blank)
0.446C ug/L(3*std of 10
runs of a low Hg level
sample)
Total Hg: .1, .14 ug/L
Inorganic Hg: .3 ug/L
NA
1
0.33" (3*std of >20
runs of blood blank)
0.446 (3*std of 10
runs of a low Hg level
sample)
Total Hg: .14, .23 ug/L
Inorganic Hg: .25, .28
ug/L
LBDTHGLC, LBDIHGLC
2
Not specif led (3*std of
>20 runs of blood blank)
0.446 ug/L(3*std of 10
runs of a low Hg level
sample)
Total Hg: 0.20 ug/L
Inorganic: 0.25 ug/L
LBDTHGLC, LBDIHGLC
2
0.33 (3*std of >20 runs
of blood blank)
0.35ug/L(3*stdof 10
runs of a low Hg level
sample)
Total Hg: 0.23 ug/L
Inorganic: 0.25 ug/L
LBDTHGLC, LBDIHGLC
2
Both sexes ages 1 - 150
aLaboratory Method procedure documents are available on the NHANES website (http://www.cdc.gov/nchs/nhanes.htm).
"From Caldwell, K.L., et al., 2009
Analytical method together with DL for inorganic Hg was not mentioned clearly in the method file. Based on the description in NHANES documentation, it uses the method with the DL
cited.
dA variable that indicates if a measurement was below the limit of detection.
-------
2.3.1 Adjustments for Non-Detects in Total and Inorganic Mercury
The analysis variable, blood MeHg concentration, is calculated by subtracting the inorganic mercury
concentration from the total mercury concentration. However, 81 percent of the inorganic and 12
percent of the total mercury measurements are below the detection limit. Thus, for most participants
the calculation of MeHg depends on how the non-detects are handled. Some methods for handling
these values below the detection limit include substituting values equal to the detection limit over
the square-root of 2, adapting survival analysis, and multiple imputation of the non-detects. Based
on analysis of simulated data with characteristics similar to the NHANES data, multiple imputation
was selected for analysis. Although more complicated than other alternatives, multiple imputation is
the most flexible and yields significant improvement in the estimates of standard errors and p-values
over other methods.
Multiple imputation involves imputing concentrations for the non-detect total and inorganic
mercury measurements. Assuming the total and inorganic mercury concentrations have a lognormal
distribution, the imputed concentrations are simulated values that 1) are less than the detection limit,
2) have the same correlation with other variables as do the non-censored values, and 3) have a
lognormal distribution. The imputed MeHg was calculated from the imputed total and inorganic
mercury concentrations. The imputation process involved the following steps: 1) use survival
analysis (SAS LIFEREG procedure) to model the inorganic mercury concentrations as a function of
total mercury concentration and other predictors; 2) simulate the inorganic mercury concentrations
for non-detects based on the model results; 3) use survival analysis to model the total mercury
concentrations as a function of inorganic mercury concentration and other predictors; 4) simulate
the total mercury concentrations for non-detects, and 5) repeat the previous steps to create ten
versions of the data, each with a different set of imputed values for the non-detects. Each version of
the data was analyzed and the ten results were combined (using the SAS MIANALYZE procedure)
to obtain the final estimates and standard errors, adjusted for the uncertainty associated with the
imputation process.
2.3.2 Calculation of MeHg Concentration
Preliminary MeHg concentrations (M) are obtained by subtracting inorganic mercury concentrations
from total mercury concentrations. Whether based on imputed or observed values, the calculated
organic mercury concentration can be negative due to imprecision in the total and organic mercury
measurements. As it is biologically impossible to have negative blood MeHg concentrations, and
because biological measurements can often be described by a lognormal distribution, the calculated
-------
values were transformed to be positive, with an approximate lognormal distribution. The
transformation that was used makes the negative measurements positive by adjusting up the negative
concentrations the most, adjusting values near zero up somewhat, and leaving values much greater
than zero relatively unchanged. This transformation has the advantage that values much above zero
can be interpreted as MeHg concentrations and values closer to zero provide a plausible
approximation to the MeHg concentration. The transformed values can be analyzed to identify
predictors of MeHg levels even though the lower values provide only an approximation to the true
concentration. However, estimates of lower percentiles of the distribution of MeHg concentrations
are, at best, approximate.
In the following transformation, MeHg is the MeHg concentration used in the analysis and C is set
to achieve a desired distribution for the log of the transformed data (approximately symmetric with
skewness close to zero and roughly normally distributed).
(M + VM2 + CN
MeHg =
To incorporate the uncertainty in the selection of C into the analysis, a slightly different value of C
was used for each imputed dataset. The skewness of the log-transformed MeHg values varied
from -0.37 to 0.66 across the ten imputed data sets.
2.4 30-Day Consumption Frequency Data
The 30-day consumption frequency data include reports by participants concerning the number of
times she consumed each of 31 types offish, as listed in Section 2.1 of this report, in the previous 30
days. There are no data on the amounts eaten. About three-quarters of participants report eating
some fish over the 30 days prior to the interview. These data, together with the 24-hour recall data,
are used to develop estimates of 30-day fish consumption amounts. In later formulas, the
consumption frequency data are designated as EatFrcqpersonSpecies.
2.5 24-Hour Dietary Recall Data
The 24-hour recall data include the U.S. Department of Agriculture (USDA) food codes from the
Food and Nutrient Database for Dietary Studies (FNNDS) and amount consumed in grams for
-------
every item of food eaten by the participant in the 24 hours immediately preceding the interview.
These FNDDS files are available from the Agriculture Research Service of the USDA (USDA, 2004;
USDA, 2006; USDA, 2008; USDA 2010; Ahuja, et al, 2012). The recipes for the food codes were
searched to find all food codes that contain finfish or shellfish. All records in the 24-hour data file
for women aged 16-49 years that were for fish-containing food codes were extracted. Only about 15
percent of participants reported consuming fish on any one day. The recipe file and 24-hour recall
data were merged to calculate quantity of raw fish consumed per recipe.
We fit a regression model using the SAS GLM procedure predicting the log-transformed quantity of
raw fish consumed (grams, adjusted for cooking method and percentage of fish in the recipe) from
the 24-hour file, as a function of number of times each species was consumed in the last 30 days,
race/ethnicity, age, income, and NHANES survey release, keeping only significant predictors. The
significant predictors were species, race/ethnicity, and log-transformed 30-day frequency of
consumption as both a linear and squared term. The prediction equation from the fit was applied to
the 30-day recall data to predict the geometric mean quantity of fish in a meal for each subject and
species consumed in the last 30 days. This is designated as MealSizePerson>species. The predicted
value was used, even if a woman had reported grams from the 24-hour recall data for a species also
reported consumed in the last 30 days. The details concerning adjustments due to cooking and
preparation have been published in previous reports (U.S. EPA, 1997; Mahaffey et al., 2004).
For species with little or no data, the models combined data across species to predict the amount of
fish consumed in a meal. The meal size for "Other fish" and "Other shellfish" was calculated from
finfish or shellfish species that are present in the 24-hour file but not specifically reported on in the
30-day frequency of consumption file; these include species such as tilapia and eel for finfish, squid
for shellfish, and other finfish and shellfish infrequently reported. For some species in the 30 day file
there were no corresponding data in the 24-hour recall file. For those species a value from another
species was used. Specifically, bass, porgy, and walleye used the quantities for salmon, "Breaded
fish" used the quantities for pollock, shark used the quantities for swordfish, and "Shellfish not
specified" used the quantities for scallop.
2.6 Fish Tissue Mercury Data
In order to estimate mercury intake, data on mercury concentrations in fish tissue are needed.
Mercury concentration in fish varies greatly among species and within species, with older and larger
fish having higher concentrations (U.S. EPA, 1997). Previous analyses (Mahaffey et al., 2004 and
10
-------
Mahaffey et al., 2009) used the fish tissue concentration data reported in the 1997 Mercury Study
Report to Congress (U.S. EPA, 1997). Much of the data in that report are from the National Marine
Fisheries Service 1978 database. As it is possible that mercury concentrations in fish tissue have
changed over the past few decades, we updated these values. For most species consumed by women
aged 16-49 years in the NHANES data 1999-2010, we were able to find data that corresponds to
this time period. However, for two species we were not able to locate more recent data and used
data from before 1998. These species are abalone and crayfish. The data we obtained represented
sampling of over 26,000 fish.
We obtained data on mercury concentration in fish tissue from the following sources:
Alaska Department of Environmental Conservation, Fish Tissue Testing Program;
Surface Water Ambient Monitoring Program, State of California;
Florida Marine Research Institute, Florida Fish and Wildlife Conservation Commission;
Department of Environmental Conservation, New York State;
State of Tennessee;
State of Virginia;
State of Massachusetts;
Micro Analytical Systems, Inc.;
Burger and Gochfeld, 2006 (data from fish in Chicago, Illinois supermarkets);
U.S. FDA;
State of Arkansas, Department of Environmental Quality;
Lake Michigan Mass Balance Study;
Contaminants in Fish from California Lakes and Reservoirs, State of California Water
Resources Control Board;
McKelvey et al., 2010 (data from New York City Asian fish markets);
Tsuchiya et al., 2008 (data from Asian grocery stores in the Puget Sound area);
McBride, 2005 (data from store-bought fish in Washington State);
Report to Congress (U.S. EPA, 1997);
11
-------
Health Canada (Health Canada, 2008);
National Rivers and Streams Assessment (data from U.S. EPA); and
State of Louisiana Department of Environmental Quality.
To estimate the geometric mean mercury concentration for each fish species, we used the SAS
MIXED procedure and modeled the log-transformed fish tissue mercury concentration by fish
species, treating the data source as a random effect. Some of the data sources reported average
concentrations for multiple fish samples and some sources reported mercury concentrations for
each individual fish sampled. In order to account for this in the modeling, we included a weighting
factor. The weighting factor allowed the modeling to take into account differing variances due to
both data source and number of individual fish samples contributing to each reported value,
modeling the error variance as a power function of the number of samples averaged to obtain the
reported value. The predicted values were converted to geometric mean fish mercury
concentrations, designated as FishHgSpecies. The average mercury concentration weighted by 30-
day consumption frequency was used for "Fish not specified" and "Shellfish not specified" species.
To the extent it could be tested, there were no consistent time trends in the fish mercury
concentration data in the sources that we used. The mercury concentrations used in the analyses by
species are in Table A-l.
2.7 Estimation of 30-Day Fish Consumption and Mercury Intake
Estimates of the amounts offish consumed over 30 days were calculated by combining the two
NHANES consumption data sets the 24-hour recall data and the 30-day frequency of
consumption data. Basically, the 24-hour data provide the amount consumed at one time. The 30-
day frequency data provide the number of times fish was consumed in past 30-days. Because many
women who reported consuming fish in the previous 30 days did not have data for fish
consumption in the 24-hour data (they did not consume fish in the past 24 hours), we needed to
estimate the amount they consume during a meal in order to calculate their estimated 30 day
consumption.
The predicted values of the amount of fish consumed at one time for each species from the
modeling were multiplied by the number of times the participant reported consuming that species in
the previous 30 days. The resulting values for each of the 31 species were then summed for each
participant to yield the estimated 30-day consumption of fish for each woman aged 16-49 years that
12
-------
completed the NHANES dietary data collection. In the formula below, FSSpecies corresponds to
the 31 fish species in the 30-day recall data. Similar calculations were made to obtain the grams of
finfish and grams of shellfish consumed.
GramsFSPerson = EatFrcclperSon,species x McalSizePerson>species
FSSpecies
To calculate the estimated 30-day mercury intake from fish, the predicted amount of fish consumed
at one time for each species from the modeling was also multiplied by the predicted mercury
concentration for the species and the number of times the participant reported consuming that
species in the previous 30 days.
HgIntakeFSPerson = } EatFreqPerson:Species x MealSizePerson>species x FishHgSpecies
FSSpecies
The reference dose (RfD) of MeHg of 0.1 ug/kg per day, is adjusted for body weight. Thus, in order
to present data that can be easily compared to the RfD, we adjusted the estimates of mercury intake
by body weight. We divided the 30-day estimate of mercury intake by body weight to get the body
weight adjusted estimates.
As demonstrated by the following equations, the mercury intake per body weight can be expressed
as the product of four components corresponding to: 1) frequency offish consumption; 2) weighted
average meal size, weighted by frequency of consumption; 3) weighted average fish tissue mercury
concentration, weighted by the quantity of fish consumed; and 4) inverse body weight.
HgIntakeFSPerKgPerson =
ies x MealSizePersori:Species x FishHgSpecies
BodyKGPerson
~ / | BO.tr 7
FSSpecies
^FSSpecies ctttr? eqPerson,Species ^ ^ea^lzePerson,species
FSSpecies ^ a*r ' eflPerson,Species
,ies x MealSizePerson>species X FishHgSpecies
t X
1
V
BodyKgPerson
13
-------
Logistic regression was used to model the probability of consuming fish in a 30-day period, and for
those that consumed fish (consumers), regression analysis was used to model these four
components, predicting the log-transformed values.
2.8 Statistical Analysis Methods
The distributions of blood mercury, frequency of fish consumption, 30-day estimates of fish
consumption, and 30-day estimates of mercury intake for the female NHANES participants aged
16-49 years were calculated. Analyses were carried out for both blood THg and blood MeHg
because while MeHg is the preferred analysis variable, due to the high percentage of samples with
inorganic mercury concentrations below the LDL, analysis of THg may provide more robust results.
Box plots were created to display the distributions of blood THg and blood MeHg. In these plots
the bottom and top edges of the box correspond to the 25th and 75th percentiles. The difference
between these is the intra-quartile range (IQR). The diamond inside the box indicates the geometric
mean value and the line inside the box indicates the median value. The whiskers that extend from
each box indicate the range of values that are within 1.5* IQR of the end of the box. Any points that
are a distance of more than 1.5* IQR from the box are indicated on the box plots as circles. The
width of the box is proportional to the number of participants in each category.
Blood THg and blood MeHg were both analyzed for evidence of trends over the period from 1999
to 2010. Detailed tables of blood THg and blood MeHg concentrations were generated, giving
sample sizes, arithmetic means, and percentiles (25th, 50th, 75th, 90th, and 95th), and their 95%
confidence intervals, by survey release, age, income, and race/ethnicity. These tables are in the
Appendix. Analytical extracts of these tabulations, generally graphic, are presented in the body of the
report. For presentation purposes, age was categorized into four groups, 16-19, 20-29, 30-39, and
40-49 years old. Race/ethnicity groups recorded in NHANES include Mexican American, Other
Hispanic, Non-Hispanic White, Non-Hispanic Black, and Other Race. Other Race consists of Asian,
Native American, Pacific and Caribbean Islander, Alaska Native, multiracial, and unknown race.
Household income categories are reported somewhat differently across the six NHANES survey
releases. In addition, income is elicited using two sets of questions, the first for less than or greater
than $20K and the second with a more detailed breakdown. For the analysis, the income categories
reflect the available data rather than a set of ordered categories. The income categories used for the
analysis are: less than $20K, $20-45K, $45-75K, greater than $75K, greater than $20K but not
14
-------
otherwise specified, Refused or Don't Know combined into one category due to small sample size,
and a category for multiple family household for which the household income was not calculated.
Similar analyses were performed for per capita frequency of fish consumption, estimated amounts of
fish consumed, and estimated mercury intake, to look for possible trends in fish consumption. Then,
blood MeHg levels were analyzed with respect to the 30-day frequencies of consumption, with
detailed tables in the Appendix and extracts in the body of the report.
All analyses were performed following the NHANES Analytical Guidelines posted on the
NHANES website (CDC 201 Ob). In particular, this means that all analyses were weighted using the
statistical weights recommended in the Analytical Guidelines, that is, the MEC weights. The
sampling design variables were used in calculating the variance of the estimates by 1) creating a set
of Balanced Repeated Replication (BRR) replicate weights with a Fay factor of 0.3; and 2) using the
SAS survey procedures (SURVEYMEANS, SURVEYREG, SURVEYFREQ) or equivalent
procedures with the replicate weights. The BRR weights were created to facilitate fitting the non-
linear model described below and, for consistency, used for all the estimates. For the analysis of
imputed concentrations (blood THg or blood MeHg), the estimates were calculated using each
imputed dataset and combined using the SAS MIANALYZE procedure to obtain standard errors
and p-values that account for the uncertainty associated with the imputation process.
2.8.1 Calculation of Percentiles
Percentiles were estimated using linear interpolation in the inverted sample cumulative probability
distribution such that bias was minimized (Hyndman and Fan, 1996; Rogers, 2003). The confidence
intervals around the percentile estimates were estimated using the Woodruff method (Sarndal,
Swenson, and Wretman, 1992).
2.8.2 Regression Analysis Predicting Blood Mercury Concentrations
Using the SAS NLIN procedure, non-linear regression analysis was used to model the relationship
between blood MeHg and mercury intake from fish, adjusting for differences by participant age,
race, income, and time across NHANES survey release. The analysis used the BRR replicate weights
and the imputed MeHg values. The resulting standard errors reflect the uncertainty due to
imputation.
15
-------
The model assumes a linear relationship between 30-day mercury intake from fish and the blood
MeHg concentration. Linear regression assumes the prediction errors have constant variance.
However, the variance of the MeHg concentrations increases as the mercury intake increases and the
variance is fairly constant for the log-transformed MeHg concentration. This suggests using the
following non-linear model where the error (n) is assumed to be normally distributed with constant
variance:
In(MeHg) = ln(lntercept + Slope x Hglntake) +
Since the MeHg concentration may vary by other factors, such as income, race, age, body weight,
and year of the NHANES survey release, the following model was fit:
(A (~lf>\
-^}
'Age\\ /Bodyweight
+ \ln[ -^- +ln[ + Di//1999 + YrTrend + YrQuadratic
\ \ 29 JJ \ 76 J
+ n
Income and race are represented by dummy variables, where the subscript indexes the income or
race category. Dummy variables can be formulated in different ways. The dummy variables for
income and race were created using deviations-from-the-mean coding (also referred to as effect level
coding). Using race with five categories as an example, assume the mean MeHg for the five
categories, after adjusting for other effects, is Meanl, Mean2, Mean3, Mean4, and MeanS. The
parameter for category l(for example) is the difference between Meanl and the mean of Meanl,
Mean2, Mean3, Mean4, and MeanS. As a result, the parameters sum to zero. The model is fit using
only four dummy variables and the parameter for the fifth category (the reference category) is equal
to the negative of the sum of the parameters for the other categories. The mean of the five
parameters (Meanl through MeanS) represents the mean for a population which is evenly
distributed across the five race categories. In the discussion of results, this will be referred to as the
response for a typical participant. The parameter estimates and standard errors can be used to assess
whether the mean or geometric mean for a selected race category and the mean or geometric mean
for a typical participant are significantly different. In the non-log measurement units, the income and
race parameters correspond to multiplicative differences in the MeHg concentration between
categories. The analysis tables show the estimate for the reference category and an overall F-test
across all levels of the categorical variables.
16
-------
The effect of age on ln(MeHg) was modeled using two variables representing a quadratic
relationship in In(age). In order to be able to interpret the parameter estimates and p-values for the
two age terms independently, age was scaled by dividing by 29 before taking the log. This makes the
linear and quadratic terms essentially uncorrelated. An overall F-test across the two age variables is
also calculated.
Although body weight is a component of mercury intake per body weight, there may be addition
differences associated with body weight. In the model, body weight is scaled by dividing by 76
before taking the log; 76 kilograms equals 168 pounds. Although not used in the model, scaling by
76 would make the correlation between log-transformed body weight and its square roughly zero.
In preliminary analyses, the blood THg and MeHg levels appear to decrease between the 1999-2000
NHANES release and subsequent releases. However, there is not a corresponding change in the
mercury intake or fish consumption estimates. As a result, the statistical modeling in this report
focuses on trends in blood mercury concentrations since the 1999-2000 release while including the
1999-2000 data in the analysis. The effect of time (Year of NHANES data release) on In(MeHg) is
represented by three terms for: a) a linear trend after 2000; b) a quadratic trend after 2000; and c) the
difference between the mean for the 1999-2000 NHANES release and the mean for releases after
2000. These three variables for assessing time differences were constructed to be essentially
uncorrelated so that the parameters and p-values could be interpreted independently. These variables
are centered to have a mean close to zero. An overall F-test across all three time variables is also
calculated. Also, the quadratic relationship from data after 2000 was extrapolated backwards to
assess if the average from the 1999-2000 release is different from what would be expected based on
the trend across later years.
The interpretation of the model parameters is complicated due to the complexity of the model, how
the dummy variables were created, and measurement error. With the provision that MeHg levels
vary by race and income, the intercept and slope can be interpreted as an approximate estimate of
geometric mean MeHg concentration and the increase in MeHg per unit increase in mercury intake
per body weight for a typical women 29 years old weighing 76 kilograms. Note however, that the
parameters are affected by the selected imputation procedures. Also, if essentially all MeHg comes
from fish consumption, then we would expect the terms for differences by income, race, and age to
be insignificant. However, the available measures of meal size, fish mercury concentration, and
frequency of fish consumption are imprecise. These quantities are related to income, race, and age.
As a result, it is likely that the income, race, and age differences found in the model are, at least in
17
-------
part, explaining differences in blood MeHg and THg that are not adequately explained by the
imprecise estimate of mercury intake.
2.8.3 Logistic Regression Predicting MeHg > 5.8
Logistic regression was used to predict the probability that the blood MeHg concentration is greater
than 5.8 ug/L and evaluate whether the probability of high MeHg values has decreased over the 12
years covered by the data. Using a safety factor of 10, the blood MeHg value of 5.8 ug/L is one
tenth of the lower 95 percent confidence limit of the cord blood mercury concentration associated
with neurologic effects on the fetus (NRC, 2000). Additionally, a blood MeHg value of 5.8 ug/L is
the concentration that forms the basis for the EPA RfD for MeHg (Rice et al., 2000). The model
used the same predictors as the nonlinear model described above except that the non-linear term
involving two parameters (ln(Intercept + Slope X HglntakeJ; was replaced by an intercept and
the transformed fish mercury intake per body weight derived from the nonlinear model described in
Section 2.8.2 above. The transformation used is the corresponding term from the non-linear fit
resulting in the following logistic model where b is a binomial error:
logit(MeHg) = Intercept + Slopel X Jn(0.4388 + 1.049'ZHgIntake) + Income^ + Racer
2
Aqe\\ /Bodyweiqht\
+ Diff1999
29 76
+ Yr Quadratic + b
2.8.4 Modeling Factors Contributing to Mercury Intake
As noted in Section 2.7, the mercury intake per body weight from fish can be expressed as the
product of four components described briefly as: frequency of consumption, meal size, fish mercury
concentration, and inverse body weight. For those that consumed fish, the log-transformed mercury
intake per body weight is the sum of the log-transformed components. In order to assess a linear
time trend in each of these components (represented by Y) the following model was used:
/n(V) = Intercept + Income t + Racer + Agea + YrTrend +
These results are presented in the appendix. However, to facilitate presentation of the results, the
parameters from the following model (treating all predictors as categorical) are shown in the plots:
18
-------
ln(Y) = Intercept + Income^ + Racer + Agea + Yeary +
As described above, dummy variables for the categorical levels were created using deviations-from-
the-mean coding. In addition, logistic regression was used to predict the probability of reporting any
fish consumption in the previous 30 days (using the variable AteFish30). These models used the
same predictors as above for predicting Logit(AteFish30).
2.8.5 Trends
When not otherwise specified, trends in continuous variables were assessed using regression with
only one predictor, year of NHANES survey release. Trends in percentiles were assessed using
logistic regression to test if the proportion above or below the overall percentile varies linearly by
year. This approach provides a general assessment of trends. However, regression adjustments for
other possible differences over time, as when using the non-linear model, can provide a more
precise assessment of trends.
19
-------
Results
3.1 Trends in Blood Mercury Concentrations
The distribution of blood THg and blood MeHg, both measured as ug Hg/L blood (ug/L) was first
examined for evidence of temporal trends. Figures 1 and 2 display the distributions for blood THg
ectivel usin boxlots.
.
and blood MeHg, respectively, using boxplots.
Figure 1. Distribution of blood THg (Mg/L), by NHANES survey release, women aged 16-49
years
IOC-
'S 10-
E :
ฃ I
ro
o
^^. .1
=5 I
01
"ro
0
1
o
m 0.1 T
0.01 -
c
<
J
J
5
L
o
\
|
4
i-
t
(i
t
j
)
9
1
I
0
I
I"
o
o
<
<
_S
9
>
i_
1
r
D
O
0
o
o
c
J
)
I
-1
ซ-
0
1999-2000 2001-2002 2003-2004 2005-2006 2007-2008 2009-2010
NHANES Data Release
20
-------
Figure 2. Distribution of blood MeHg (MgA), by NHANES survey release, women aged 16-49
years
10-
1 -
0.1 -
0.01 -
o
o
o
o
i
I
o
1999-2000 2001-2002 2003-2004 2005-2006
NHANES Data Release
2007-2008
2009-2010
The figures show similar patterns across time; however the MeHg plot shows larger differences in
the means between years. The geometric mean blood THg in 1999-2000 was 1.21 times higher than
the geometric mean across the subsequent 10 years (2001-2010), representing an 18 percent decrease
between 1999-2000 and 2001-2010. For blood MeHg, the geometric mean in 1999-2000 was 1.51
times higher than the geometric mean across the subsequent 10 years. This represents a decrease of
34 percent between 1999-2000 and 2001-2010. The linear time trend in the means of the log-
transformed data is not statistically significant for THg at the 5 percent level (p=0.11), but it is for
MeHg (p=0.0006). However, when survey release 1999-2000 is excluded from the analysis, there is
no statistically significant time trend for MeHg (p=0.74) from 2001-2002 to 2009-2010. There is a
statistically significant linear tend in the percent above the overall 90th percentile for both THg
(p=0.004) and MeHg (p0.003). Excluding survey release 1999-2000 from the analysis, there is no
statistically significant trend in the percent above the overall 90th percentile for either THg (p=0.72)
ofMeHg(p=0.58).
The percentage of participants with blood THg and blood MeHg over 5.8 ug/L in each of the
survey releases are shown in Table 5. The percentage of women of reproductive age with blood THg
21
-------
over 5.8 ug/L in 1999-2000 was 2.64 times that found in 2001-2010, a decrease of 62 percent
between 1999-2000 and 2001-2010. For blood MeHg, the percent of women of reproductive age
over 5.8 ug/L in 1999-2000 was 2.86 times higher than the percent of women in 2001-2010,
representing a 65 percent decrease between 1999-2000 and 2001-2010. There are significant
differences between the survey releases for both THg and MeHg (chi-square p-value <.0001), with
1999-2000 having approximately twice the number of women with levels over 5.8 ug/L compared to
the other sets of years. Excluding survey release 1999-2000 from the analysis, there are no significant
differences between the survey releases for either THg (chi-square p-value = 0.57) or MeHg (chi-
square p-value = 0.56).
Table 5. Percent of women aged 16 to 49 years with blood MeHg > 5.8 Mg/U by NHANES
survey release
Survey Release Percent THg >5.8 ug/L (SE)
Overall
1999-2000
2001-2002
2003-2004
2005-2006
2007-2008
2009-2010
3.45 (0.36;
7.13(1.78)
3.67(0.72)
2.40 (0.82)
2.67 (0.60)
2.49 (0.60)
2.30 (0.41)
Percent MeHg >5.8 ug/L (SE)
3.14 (0.34;
6.77(1.77;
3.14(0.71;
1.70(0.70;
2.33 (0.58)
2.42 (0.56)
2.14 (0.36)
More detailed tabulations of the blood mercury distributions, including sample sizes and 25th, 50th,
75th, 90th, and 95th percentiles, are in the Appendix, Tables A-2 and A-3. Table A-2 shows the
distributions of blood THg and Table A-3 shows the distributions for blood MeHg for all women
aged 16-49, by NHANES survey release, by race/ethnicity, by income, and by age.
3.2 Trends in Fish consumption
3.2.1 Trends in Frequency of Consumption
Figure 3 displays the percent of women aged 16-49 years in each of six categories of reported 30-day
fish frequency consumption. Detailed tabulations are in the Appendix, Table A-4. Note that the
percentages for consuming finfish only or shellfish only zero times in the past 30 days are greater
than the percent consuming total finfish/shellfish zero times in the past 30 days because some
participants only consume finfish or only consume shellfish. And the percent consuming total
finfish/shellfish six or more times in the past 30 days is greater than for either finfish or shellfish
22
-------
alone as individuals who consumed shellfish less than 6 times and finfish less than 6 times can have
a total greater than 6 when combined. While there are statistically significant differences in
consumption frequency between the survey releases (Rao-Scott Chi-Square p-values: p=0.03 for
total fish, p=0.02 for finfish, p=0.16 for shellfish), there is not a consistent trend over time.
Approximately 8 percent more women reported consuming fish 6 times or more in the previous 30
days in 2005-2006 (34.2%) compared to 1999-2000 (26.4%). The percentage of women reporting
this frequency of consumption in 2007-2008 (26.9%) drops back to what was observed in 1999-2000
data then increases again in 2009-2010 (31.8%).
Figure 3. Percent of participants by 30-day fish consumption frequency, by NHANES survey
release, women aged 16-49 years
Figure 4 summarizes the frequency of consumption by demographics of interest: income,
race/ethnicity, and age. The figure shows differences in frequency of fish consumption between
income group, race/ethnicity, and age (all have Rao-Scott Chi-Square p-values <.0001). Higher
income is associated with increased frequency of fish consumption as is older age. Women in the
"Other Race" category, which includes Asian, Pacific Islander, American Indian, Alaska Native,
multi-racial, eat fish more frequently than Hispanic, non-Hispanic white, and non-Hispanic black
23
-------
women. These findings agree with previous literature (Kudo et al., 2000; Sechena et al., 2003;
Mahaffey et al., 2004). Since these demographic characteristics are associated with fish consumption,
they were included in the multivariate analyses discussed in Section 3.4.
Figure 4. Percent of participants by 30-day fish consumption frequency, by demographic
characteristics, women aged 16-49 years, NHANES 1999-2008
I 6+times
4-5 times
3 times
D 2 times
D 1 time
DO times
*Uncalculated indicates that the participant is residing in a multi-family dwelling and one or more of the families only reported a range for
their family income, either <$20,000 or >$20,000. Thus NCHS did not calculate household income for these participants.
3.2.2 Trends in Estimated Amounts Consumed Over the Previous 30 Days
Detailed tables on the amounts offish consumed are in the Appendix, Tables A-5 through A-8.
Tables A-5, A-6, and A-7 present estimates of amounts offish eaten (g), mercury intake (jag), and
mercury intake per unit body weight (ug/kg) over the previous 30 days, by NHANES survey release.
Table A-8 presents the same statistics by income, race/ethnicity, and age. The estimates are
presented for shellfish consumption, finfish consumption, and combined fish/shellfish
consumption. The presented statistics include number of sampled women age 16-49, arithmetic
means and their 95% confidence intervals, 25th, 50th, 75th, 90th, and 95th percentiles and their 95%
24
-------
confidence intervals. Presented here are summaries from Tables A-5 and A-6.
Figure 5 displays the estimated mean amounts of fish consumed over a 30-day period by women
aged 16-49 and the 90th percentiles. There is no evidence of a trend over time in the consumption of
fish in either the mean (p=0.31) or the percent over the overall 90th percentile (p=0.30).
Figure 5. Estimated mean and 90th percentile amounts offish consumed in 30 days, women
aged 16-49 years, NHANES 1999-2010 (with 95% confidence intervals)
Arith. Mean -*-90th %ile
1999-2000 2001-2002 2003-2004 2005-2006 2007-2008 2009-2010
NHANES Survey Release
Figure 6 displays the estimated mean amounts of mercury ingested by eating fish per unit body
weight. The patterns over time are different than those observed in Figure 5 for estimated 30-day
fish consumed. At both the mean and the 90th percentile there appears to be a decrease over time in
ug mercury ingested per kg body weight. However, this change is not statistically significant (p=0.35
for the mean and p=0.92 for the percent over the overall 90th percentile). The different patterns
observed in figures 5 and 6 indicate that women who are the highest fish consumers are possibly
shifting to lower mercury fish.
25
-------
Figure 6. Estimated mean and 90th percentile amounts of mercury ingested, normed to body
weight (Mg/kg) in 30 days, women aged 16-49 years, NHANES 1999-2010 (with
95% confidence intervals)
1.60
1.40
in
1 1.20
o
ro
i i.oo
00
01
-o 0.80
0
-Q
2
if ฐ-60
00
- 0.40
00
1
0.20
0.00
y-error bars = 95% Cl
T T T
>' \ '-,; ^
L-^ '^--4- 4 -r____-4
1 IT
--Arith. Mean -A-90th %ile
1 1 1 1 1 1
1999-2000 2001-2002 2003-2004 2005-2006 2007-2008 2009-2010
NHANES Survey Release
3.3 Associations between Fish Consumption Frequency and
Blood Mercury Concentrations
This section examines statistical associations between fish consumption and blood mercury
concentrations, especially with respect to changes over time.
Figures 7 and 8 display the concentration of blood THg and blood MeHg by frequency of fish
consumption in 30 days, using box plots. As expected, both blood THg (p<.0001) and blood MeHg
(p=0.003) increase with frequency offish consumption. These figures agree with previous findings
(Schober et al., 2003; Mahaffey et al., 2004; Mahaffey et al., 2009) that people who eat fish more
frequently tend to have higher blood mercury levels and, further, there is a dose-response gradient
observed in the mean.
26
-------
Figure 7. Distribution of blood THg (Mg/L), by reported frequency of fish consumption in 30
days, women aged 16-49 years, NHANES 1999-2010
100-
10-
1 -
0.1 -
0.01 -
o
-8-
0 times 1 time 2 times 3 times 4-5 times
Frequency of Fish Consumption in 30 days
6+times
Figure 8. Distribution of blood MeHg (MgA), by reported frequency of fish consumption in 30
days, women aged 16-49 years, NHANES 1999-2010
100-
ฃ 101
ฃ=
cn
o
_l
d .1
=3 1 .
CD
O
-E 0.1 -
m
0.01 -
r-i
0
ง
-
j
!
0
0
f
I
I
J
5
} E
i
]
1
8
o
a
0 ฐ 0
0 0
o
o
]
<
\
s
>
I
8
e
ง
0 times 1 time 2 times 3 times 4-5 times 6
Frequency of Fish Consumption in 30 days
O
I
S
+ times
27
-------
Figure 9 displays mean blood MeHg levels against the reported 30-day frequency of consumption of
fish, for each of the six two-year periods. Detailed tables are in the Appendix, Table A-9. This figure
also shows a statistically significant change over time. Women who ate fish more frequently in 2009-
2010 had lower blood MeHg levels than women who ate fish with the same frequency in 1999-2000.
For example, among women who ate fish 6 or more times, the arithmetic mean blood mercury level
was 3.36 (2.75,3.97) ug/L in 1999-2000; in 2009-2010, it had dropped to 2.11 (1.87,2.35) ug/L, a
statistically significant decrease. However, the lowest concentrations are observed in survey release
2005-2006, with an arithmetic mean of 1.84 (1.61,2.08) ug/L, with slight increase the two following
survey periods. Similarly, the 90th percentile of blood MeHg dropped from 8.33 (6.28,11.07) ug/L
to 4.24 (3.54,5.08) ug/L. These findings suggest that women who consume fish more often may be
shifting to fish with lower concentrations of mercury.
28
-------
Figure 9. Mean blood MeHg concentrations by reported frequency of fish consumption in 30 days, women aged 16-49 years,
NHANES 1999-2010 (with 95% confidence intervals, median, and 90th percentile)
8
7
CUD 6
H.
n
X 5
0)
| 4
o
o
CO 3
Arithmetic Mean
-Median
y-error bars = 95% Cl
I
8
8
8
8
8
0 times
8
8
8
8
8
Itime
8
8
8
2 times
8
8
8
8
3 times
8
8
8
8
4-5 times
8
8
8
8
8
8
6+times
Number of Times Finfish/Shellfish Consumed in Past 30 Days
29
-------
3.4 Regression Analysis Results: Associations between Blood
Mercury, Fish Consumption, Time, and Demographic
Factors
To better understand the relationship between blood MeHg concentration and time, adjusting for
factors known to be associated with fish consumption, a nonlinear model was fit to predict the log
transformed MeHg concentration from transformed mercury intake and demographic
characteristics. The model assumed that there was a linear relationship between blood MeHg
concentration and mercury intake from fish. Predicting the log-transformed blood MeHg has the
advantages that the prediction errors have relatively constant variance, as assumed by the model, and
the predicted values are always positive. Using the non-linear model has the advantage that the
intercept and slope for mercury intake have the same interpretation as when using linear models.
Table 6 shows the parameter estimates, standard errors, p-values, and relative ratios from the
nonlinear model. The intercept and slope are equivalent to the intercept and slope in a linear model
predicting blood MeHg from fish mercury intake. The other demographic parameters model
multiplicative differences in the blood MeHg concentrations. Multiplicative differences
corresponding to the parameters are in the Relative Ratio column.
The slope parameter is highly significant, indicating an increase in blood MeHg with an increase in
mercury intake per body weight (p < 0.0001). MeHg concentrations also depend on the participant's
income (p < 0.0001), age (p < 0.0001), and race (p < 0.0001). Blood MeHg concentrations increase
with increasing age; however, at older ages the increase diminishes. Blood MeHg concentrations
increase with increasing income such that those with household income above $75,000 have MeHg
concentrations about 1.5 times higher than those with income less than $20,000. However,
interpreting the trend is complicated by how the income data are reported. Blood MeHg
concentrations also vary by race category with non-Hispanic whites and Mexican Americans having
lower concentrations than non-Hispanic blacks and other Hispanics and races other than those
listed (other races) having the highest levels. The blood MeHg concentrations for other races are
about 1.8 times higher than for non-Hispanic whites. Differences among NHANES survey releases
are also statistically significant (p < 0.0001). Blood MeHg concentrations from survey release 1999-
2000 are significantly higher than the mean of the other years (p<.0001). The linear trend after 1999-
2000 is not statistically significant (p=0.72), but the quadratic trend after 1999-2000 is (p=0.004).
This corresponds to decreasing blood MeHg concentrations followed by relatively small changes
and a slight increase in the last years. When extrapolated backward, the quadratic trend fit to the data
30
-------
after 2000 and the geometric mean from the first NHANES survey release, 1999-2000, are not
significantly different (p=0.34).
Table 6. Parameter estimates and relative ratios from the non-linear model predicting blood
MeHg concentrations
Intercept
Hg/Wt Slope
Body Weight
Age, Overall
Age
Age2
Income, Overall
0 to 20K
20 to 45 K
45 to 75K
>75K
MultiHH
Refuse/DK
Over20K
Race, Overall
Non-Hispanic Black
Mexican Amer.
Other Hispanic
Other Race
Non-Hispanic White
NHANES Year, Overall
1999-2000 different from post-2000 mean
Linear trend after 1999-2000
Quadratic trend after 1999-2000
1999-2000 diff. from post-2000 quadratic trend
Parameter
0.44
1.05
-0.09
0.34
-0.28
-0.16
-0.07
0.00
0.25
-0.01
0.07
-0.09
0.14
-0.27
0.04
0.33
-0.24
0.47
-0.005
0.05
0.11
Std. Error
0.07
0.08
0.05
0.05
0.13
0.04
0.03
0.03
0.04
0.07
0.09
0.08
0.04
0.04
0.05
0.05
0.04
0.09
0.01
0.02
0.11
p-Value
0.0076
<0.0001
0.08
<0.0001
<0.0001
0.04
<0.0001
<0.0001
0.04
0.97
<0.0001
0.90
0.42
0.29
<0.0001
0.0012
<0.0001
0.45
<0.0001
<0.0001
<0.0001
<0.0001
0.71
0.0038
0.35
Relative
Ratio
0.85
0.93
1.00
1.29
0.99
1.08
0.92
1.15
0.76
1.04
1.39
0.79
The same model described above was fit for blood THg. The results were similar for all independent
variables except for the year terms. When extrapolated backward, the quadratic trend fit to the data
after 2000 corresponds to a significantly lower THg concentration for the first NHANES survey
release, 1999-2000, than observed (p=0.004).
The non-linear model predicts the mean of the log-transformed blood MeHg concentrations.
However, the upper percentiles may follow a somewhat different pattern. To test if there is a trend
31
-------
in the upper percentiles, logistic regression was used to model the probability of a blood MeHg
measurement over 5.8 ug/L. Overall 3.1% of concentrations are greater than 5.8 ug/L. The
predictors in the model include scaled age (as a linear and quadratic parameter), dummy variables for
income and race, quadratic trends across NHANES survey release (including the difference between
the 1999-200 release and later releases), and transformed fish mercury intake. The model used the
transformation of fish mercury intake derived from the non-linear model. The parameter estimates
shown in the equation below, are the parameter estimates from the nonlinear model (Table 6.).
Trans f or me d(Hg Intake) = /n(0.4388 + 1.0492//#Intake)
Table 7 shows the parameter estimates, standard errors, odds ratios, and p-values from the logistic
regression model. Differences among NHANES survey releases are statistically significant
(p<.0001). The probability of having blood mercury concentration greater than 5.8 ug/L is higher in
1999-2000 survey release (p<.0001). Neither the linear nor quadratic trend since 1999-2000 is
statistically significant at the 5 percent level. The probability of MeHg concentrations over 5.8 ug/L
also depends on the participant's income (p<.0001), age (p = 0.002), and race/ethnicity (p <.0001).
The probability generally increases with increasing age. Probabilities increase with increasing income.
However, interpreting the trend is complicated by how the income data are reported. Probabilities
also vary by race/ethnicity category with lower probabilities for Mexican Americans and higher
probabilities for the "Other race" category.
32
-------
Table 7.
Parameter estimates and odds ratios from the logistic model predicting the
probability of blood MeHg concentrations over 5.8
Intercept
Transformed Mercury Intake
Body Weight
Age, Overall
Age
Age2
Income, Overall
0 to 20K
20 to 45K
45 to 75K
>75K
MultiHH
Refuse/DK
Over 20K
Race, Overall
Non-Hispanic Black
Mexican Amer.
Other Hispanic
Other Race
Non-Hispanic White
NHANESYear, Overall
1999-2000 different from others
Linear trend after 1999-2000
Quadratic trend after 1999-2000
1999-2000 diff. from quadratic trend
Parameter
-3.80
1.42
-0.61
0.92
-1.91
-0.99
0.0007
0.04
0.56
-0.20
0.67
-0.09
0.37
-1.39
-0.33
1.34
0.01
1.43
-0.09
0.05
0.81
Std. Error
0.23
0.13
0.35
0.31
0.98
0.33
0.21
0.25
0.21
0.35
0.49
0.61
0.14
0.24
0.49
0.24
0.22
0.29
0.07
0.07
0.60
p-Value
<0.0001
<0.0001
0.09
0.0018
0.0035
0.06
<0.0001
0.0037
1.00
0.87
0.0095
0.56
0.17
0.89
<0.0001
0.0116
<0.0001
0.50
<0.0001
0.97
<0.0001
<0.0001
0.25
0.45
0.18
Odds
Ratio
0.37
1.00
1.04
1.75
0.82
1.96
0.92
1.45
0.25
0.72
3.82
1.01
33
-------
3.5 Regression Analysis Results: Associations between Fish
Consumption and Intake of Mercury with Time and
Demographic Factors
To better understand the relationship between fish consumption and time, logistic regression was
used to model the probability of a person reporting any fish consumption in the previous 30 days,
and for those with fish consumption, five regression models were fit to predict fish consumption
and mercury intake variables from demographic characteristics and NHANES survey release. These
variables were amount of fish consumed in a meal (meal size), number of meals in 30 days, the
mercury concentration in the fish consumed (calculated as the ratio of 30-day mercury intake to 30-
day fish consumption), the inverse of body weight, and the mercury intake per unit body weight.
Full model results to assess a linear time trend in each of these components are presented in the
appendix (Tables A-10 to A-15). The results from the models treating all predictors as categorical (to
facilitate presentation) are summarized in Figures 10 through 14.
Figure 10 presents the results from the regression models for the race/ethnicity categories. The
parenthetical percentages beside the race/ethnicity group, e.g., white, non-Hispanic (65%), are the
percent of the total study population that comprises that category. The grey filled star symbols and
error bars plotted on the second y-axis are the percent of that category that reported any fish
consumption in the previous 30 days. The remaining colored symbols and error bars are, for fish
consumers, the relative ratios (RR) and 95 percent confidence intervals from the regression models
predicting 1) the log-transformed frequency of fish consumption in the previous 30 days (red open
diamond symbol); 2) the log-transformed meal size (green filled circle symbol); 3) the log-
transformed mercury concentration in the fish consumed (orange filled diamond symbol); 4) the log-
transformed inverse of body weight (blue open circle symbol); and 5) the log-transformed mercury
intake per unit body weight (black filled square symbol). The horizontal line on the plot at RR = 1
represents the geometric mean response for a hypothetical population equally divided among
categories for race or other categorical variables. This will be referred to as the response for a typical
participant. If a symbol is above the line at RR = 1, then that racial/ethnic group is higher than the
geometric mean for a typical participant for that fish consumption or mercury intake variable. For
example, non-Hispanic white women consume fish with higher geometric mean mercury
concentrations than the typical women (orange filled diamond symbol). Correspondingly, if a
symbol is below the line at RR = 1, then that racial/ethnic group is lower than typical for that
variable. For example, Mexican American women reported fewer meals in 30 days (red open
diamond symbol) compared to the geometric mean for a typical participant. The blue open circle
34
-------
symbol is the relative ratio of the inverse of body weight, thus a RR greater than one indicates lower
body weight than typical and a RR less than one indicates higher body weight than typical.
35
-------
Figure 10. Relative ratios and 95% confidence limits from the models predicting fish consumption and mercury intake variables
versus race/ethnicity
Fish and Mercury Variables versus Race
2.25 -
2-
1.75-
1.5-
.0 1.25-
"co
^ 1 -
>
1 0.8-
QL
0.6-
0.4-
i
*
<
?
%
- i ' 1
_ ฑ__ 9 f T * i
^ *i ^sj 1 s i
i ij i -
i i i
%, r^% \
4b *, \
\ \
Race
ate fish last 30 days o # meals in 30 days
Hg cone, in fish consummed o Inverse body weight
r Tl lTj~l
1
'
r i
1
\
\
'* ^*
* Meal
size
f r
f-
i
-
(0
ra
T3
-35% o
-i ซ
W
-30% -
to
-75% .E
-i
ro
-70% ฃ
CD
O
-65% ฃ
-
V
/ft
vr
Hg intake per body weight
36
-------
The distance of the RR from the reference line of one, for a particular racial/ethnic group (e.g.,
other Hispanic), can be summed across four of the models (meal size, number of meals in 30 days,
the mercury concentration in the fish consumed, and the inverse of body weight) to approximately
equal the distance from one of the RR of the fifth model, mercury intake per kg body weight. Figure
11 presents an extract of the full plot displayed in Figure 10 to illustrate this point.
As seen in both Figure 10 and Figure 11, other Hispanics consume fish less frequently, their meal
sizes are smaller, the mercury concentration in the fish consumed is less, and their body weights are
less than typical. In Figure 11, the brackets to the right of each symbol show the distance of the RR
from one. These distances can be combined to equal the RR for the fifth model, mercury intake per
unit body weight, as shown by the brackets to the right of the black filled square symbol. Note that
the RR for the inverse of body weight is the only one for other Hispanics that is above one, thus it is
subtracted from the total of the other distances. For the race category, other/multi-race, all RRs
except for meal size are greater than one, thus the distances from the other three models are
summed and the distance from the model of meal size is subtracted, to equal the distance of the RR
of the fifth model.
There are statistically significant differences for all of the fish consumption and mercury intake
variables by race/ethnicity. The p-values testing the overall significance of race/ethnicity in all
models is p<.0001. The proportion of non-Hispanic black women who consumed fish in the
previous 30 days is higher than for other racial/ethnic categories while the proportion of non-
Hispanic white women is less (Figure 10). Other/Multi-race women consume fish the most
frequently, consume fish with higher concentrations of mercury, and have lower body weights,
resulting in the highest mercury intake per unit body weight of the racial/ethnic categories. Mexican
American women consume the largest meal sizes; however they eat fish less frequently compared to
the other racial/ethnic groups, resulting in low mercury intake per unit body weight. Non-Hispanic
black women consume larger meal sizes than typical; however, their body weights are greater than
typical, resulting in lower than typical mercury intake per unit body weight. Non-Hispanic white
women consume fish with higher concentration of mercury than typical; however they consume fish
less frequently than typical and have smaller meal sizes, resulting in a lower than typical mercury
intake per unit body weight.
37
-------
Figure 11. Extract of the full plot of fish and mercury variables versus race/ethnicity
2.25
2
1.75
1.5-
1.25-
_ro
(Li -I
ce '
0.8-
0.6-
Example
frl
-J- ^I
Other Hispanic (7%)
Other/Multi- Race ( 6%)
Race
*
*
o
^^^^^^^^H
% ate fish last 30 days
Meal size
Inverse body weight
- # meals in 30 days
- Hg cone, in fish consummed -
# meals in 30 days
^ Hg cone, in fish consummed
Hg intake per body weight
^^^ Meal size
Inverse bodyweight
Figure 12 displays the RRs of age groups for each of the fish consumption and mercury intake
variables. The p-values testing the overall significance of age in all models is p<.0001, except for the
model of mercury concentration of the fish consumed which has a p-value of 0.0002. The percent of
women who consumed fish in the previous 30 days increases with increasing age. All of the factors
that contribute to mercury intake per unit body weight increase with increasing age. Women aged 16
to 19 years have the lowest intakes of mercury per unit body weight: they consume fish less
frequently, eat smaller meal sizes, consume fish with low concentrations of mercury, and have the
lowest body weights of all age groups. Women aged 40 to 49 years have the highest intakes of
mercury per unit body weight: they consume fish most frequently, eat the largest meal sizes,
consume fish with higher concentrations of mercury, and have the highest body weights of all age
groups.
Figure 13 displays the RRs of income groups for each of the fish consumption and mercury intake
variables. The p-values testing the overall significance of income in the models are as follows; for the
38
-------
proportion that consumed fish in the previous 30 days, p<.0001; for the frequency offish
consumption in the previous 30 days, p<.0001; for the meal size, p=0.44; for mercury concentration
of the fish consumed, p=0.13, for body weight, p<.0001, and for mercury intake per unit body
weight, p<.0001. Due to how income data were collected, it is difficult to fully assess trends across
income level. However, mercury intake per unit body weight is highest in the income categories
$75K and up and $20k and up. In both cases, the frequency of fish consumed in 30 days, the meal
size, and the mercury concentration in the fish consumed are higher than other income groups.
Additionally, body weight decreases with increasing income. All of these factors contribute to the
increased mercury intake per unit body weight in the higher income groups.
Figure 14 displays the RRs of NHANES survey release for each of the fish consumption and
mercury intake variables. The p-values testing for a trend across time in the models are as follows;
for the proportion that consumed fish in the previous 30 days, p=0.21; for the frequency of fish
consumption in the previous 30 days, p=0.37; for the meal size, p=0.46; for mercury concentration
of the fish consumed, p=0.035; for body weight, p=0.20; and for mercury intake per unit body
weight, p=0.35. Women from NHANES 2001-2002 and 2005-2006 have the highest mercury intake
per unit body weight. In NHANES 2001-2002, women consumed fish with higher concentrations of
mercury and had larger than typical meal sizes. In NHANES 2005-2006, women consumed fish
more frequently than typical. NHANES 2007-2008 has the lowest mercury intake per unit body
weight. Women in this survey period ate fish with low mercury concentrations and consumed fish
the least frequently. The decreasing trend in mercury concentration in the fish consumed (the ratio
of mercury intake to fish consumed) is consistent with women shifting consumption to species with
lower concentrations of mercury over time.
39
-------
Figure 12. Relative ratios and 95% confidence limits from the models predicting fish consumption and mercury intake variables
versus age
Fish and Mercury Variables versus Age Group
2.25-
2-
1.75-
1.5-
1.25-
0
"ro
QL 1
OJ
-i
ro
m
a. 0.8-
0.6-
0.4-
T
i
W T
1
1 ~~i i T
-rTr n~ fl)>
7 tffiJ *^ i "*"
XT I * m ffi
^
_L
II
--
-
-85% ฃ
ro
T3
0
n
to
ra
-80% E
"to
*
-75% ~
O.1
-1
OJ
-70% g
D_
-65%
_
i i i i
16 to 19 (11%) 20 to 29 (23%) 30 to 39 (30%) 40 to 49 (32%)
Age Group
* % ate fish last 30 days o # meals in 30 days ป Meal size
+ Hg cone, in fish consummed o Inverse body weight Hg intake per body weight
40
-------
Figure 13. Relative ratios and 95% confidence limits from the models predicting fish consumption and mercury intake variables
versus income
Fish and Mercury Variables versus Income
2.25-
2-
1.75-
1.5-
o 1.25-
"ro
di '
"ro
-^ 0.8-
ct
0.6-
0.4-
^1 -r-QjT-r
? n '?- f
, -1- _l_
If
i i
^ &>
^ #-.
^S-i ' j'
^ 0|fr
/ff> '*(
JL
$^
i
\
"
'
2u
i
c
ฐ^
\
t
-p 1
Ifi
*L
i
\
1
1
j
<
%
T
i
7^
1
-%
v
>
i
i
i
;.
^S"
^)
<
i
--i-
,1.
L
JT^_
i
^
i
1
i.
r
' ..
^Ji
4P
L_L
i
\
i
-
tn
ro
-85% S
en
"55
ro
-80% =
"ง
rm
-75% .E
"ro
iii
-70% "E
o
-65% |
-
*v-
%
?-i Vc_ '>?_ O,,
^% ~J&- ^ft" "ง" '^ ^>
^V
%-
^
^
^
Household Income
if % ate fish last 30 days
<> # meals in 30 days
Meal
+ Hg cone, in fish consummed o Inverse body weight Hg
size
intake
per body weight
41
-------
Figure 14. Relative ratios and 95% confidence limits from the models predicting fish consumption and mercury intake variables
versus NHANES survey release
Fish and Mercury Variables
2.25-
2-
1.75-
1.5-
1.25-
o
"ro
CD
"ro
5 0.8-
0.6-
0.4-
i
<
> -L 1
1
_
1
t
T i
'
3^J
I
f
ffi
" I
r
Year
= i
i
k
4
X
i i i i i i
1999-2000 2001-2002 2003-2004 2005-2006 2007-2008 2009-201
1
0
-85% ฃ
ro
O
-80% ^
-75% =
CD
-i
C
-70% Jj;
D_
-65%
-
NHANES Year
* c
fc ate fish last 30 days
* Hg cone, in fish
consummed
O # meals in
30 days * Meal size
o Inverse bodyweight Hg intake per body weight
42
-------
Discussion
This analysis found statistically significant differences in blood MeHg and blood THg
concentrations across the study period in both the mean concentrations and the upper percentiles.
The nonlinear model predicting the log-transformed blood MeHg found survey release 1999-2000 to
be significantly higher than the mean of the other releases (p<.0001) and a significant quadratic
trend (p=0.004) from 2001-2002 to 2009-2010, indicating decreasing blood MeHg concentrations,
followed by relatively small changes and a slight increase in the last years. The nonlinear model
predicting the log-transformed blood THg found NHANES survey release 1999-2000 to be
significantly higher than the mean of the other sets of NHANES survey releases (p=0.0005) and no
statistically significant trend since 2000. The logistic model predicting the probability of MeHg >5.8
ug/L found that women from survey release 1999-2000 had higher probability of having blood
MeHg >5.8 ug/L (p<.0001) and no statistically significant trend since 2000. Demographic
characteristics found to be associated with blood mercury in other studies (Schober et al., 2003;
Mahaffey et al., 2004; Mahaffey et al., 2009; Caldwell et al., 2009) showed the same relationships in
this analysis. There was a significant relationship between mercury intake from fish consumption
and blood mercury (p<.0001) in all three models. However, after adjusting for mercury intake, there
were additional differences in blood mercury between the survey releases.
The analysis showed few changes in fish consumption and mercury intake over the study period.
Chi-square analysis found a statistically significant difference in the reported frequency of
consumption (p=0.03) across survey releases; however, the model predicting the frequency of
consumption among consumers found no significant trend over time. It did find significant
relationships between race, age, and income with frequency of consumption (p<.0001). There was
no evidence of a trend in either estimated 30-day fish consumption or mercury intake per unit body
weight. There was a marginally statistically significant decreasing trend across the NHANES survey
releases in the ratio of mercury intake to fish consumed (p=0.035) that is consistent with women
shifting their consumption to fish with lower mercury concentrations. The decrease observed in
blood mercury concentrations was between 1999-2000 and the subsequent releases. The decreasing
trend in the ratio of mercury intake to fish consumed is observed after 2000, between 2001-2002
and 2009-2010, as is shown in Figure 14. Thus this finding does not explain the change in blood
mercury.
43
-------
There are limitations of the analysis that might affect the observed relationships. The laboratory
method for measuring blood total mercury changed between survey releases 2001-2002 and 2003-
2004 (Table 4). Thus some of the change observed in the blood mercury data between the first four
years (1999-2002) and the last 8 years (2003-2010) could be attributed in part to the laboratory
method. Caldwell, et al, 2009, reported a small bias due to the change in laboratory methods. They
observed a THg difference between the methods (earlier method compared to the latter method) of
-17 percent in the low concentration (0.24 ug/L) quality control pool and 7 percent in the high
concentration (10.6 ug/L) quality control pool (Caldwell, et al. 2009). Most observed THg
concentrations are between these two test concentrations. Caldwell et al. considered a correction for
methodological differences but found it had little effect on the conclusions.
The data used in the analysis do not include the quantities consumed for each meal nor the mercury
content of the fish item that was consumed. Instead, the analysis uses geometric mean estimates of
these values. As a result, the actual variation of the estimated grams offish consumed and the 30-day
mercury intake is most likely greater than estimated. Thus, the upper percentiles presented in the
tables are likely to underestimate the actual quantities.
In order to calculate mercury intake, we used fish tissue mercury concentration data from the same
time-period, 1999-2010, for most species, but we did not have enough data to assign mercury
concentrations to fish species by NHANES survey release. Thus, if mercury in fish tissue was
changing over the study period, this would not be accounted for in the analysis. Furthermore,
mercury varies in water bodies across the United States and the world, but given the data limitations
we do not know the source of the fish consumed by individual participants. Another factor that
hampers the calculation of mercury intake is the possible changes in commercial fishing practices
that affect the amount of mercury that gets into the fish at markets and grocery stores. For example,
fisheries may have reduced fishing of larger fish which are generally known to contain more
mercury. One improvement that could be possible in this analysis would be to adjust the mercury
concentrations in fish by the market share of fish size.
An additional limitation is the NHANES survey design. Regional differences have been reported in
the consumption offish, especially between coastal and inland regions (Mahaffey et al., 2009).
NHANES does not control for these regional differences from year to year. It is possible that some
of the observed findings may be related in part to changes in the regional patterns in NHANES
from year to year.
44
-------
Another limitation lies in the use of dietary recall data. Both the 24-hour recall data and the 30-day
frequency data have measurement error associated with them. This measurement error is known to
reduce the power to detect relationships (Willett, 1998). The conclusion that fish consumption has
changed minimally over the course of the study period relies on dietary recall data, both 24-hour
recalls and food frequency questionnaires, which have known limitations such as recall bias.
However, the methodology in the data collection across years is identical, thus it would be unlikely
that the recall bias would change from year to year.
Finally, the statistical model assumes a linear relationship between the fish mercury intake and the
blood MeHg. A detailed analysis of the data suggests a slightly non-linear relationship such that for
very frequent consumers of fish, the predicted consumption is greater than the reported value. This
apparent relationship may be due to participant problems estimating the number of fish meals, a
nonlinear biological response to frequent fish consumption, or other factors. Assessing the
functional form of the relationship is complicated by the high percentage of imputed values.
Although other relationships might be modeled, the basic conclusions should not be affected.
Oken et al., 2003, found a decline in fish consumption by pregnant women between 1999-2000 and
2001-2002 in Massachusetts: they attributed this to national fish consumption advisories. This
analysis does not show the same result in women of child-bearing age. This analysis found no trend
over time in amount of fish consumed or frequency of fish consumed. As fish consumption did not
decrease after the issuance of either the 2001or 2004 national fish consumption advisories, we can
conclude that the issuance did not influence women on a national level to decrease their fish
consumption. The decrease in the ratio of mercury intake to fish consumed occurred between 2001-
2002 through 2009-2010. It is possible that fish advisories have led women to choose fish of lower
mercury concentrations; however, the NHANES survey is not designed to assess the effectiveness
offish advisories, thus we cannot draw this conclusion from this analysis.
The cause of the discrepancy observed in detecting a difference in blood MeHg levels between
1999-2000 and the subsequent survey releases but no corresponding difference in the frequency and
amounts of fish eaten, is not yet clear. The models predicting blood mercury from mercury intake
per body weight show a strong relationship between mercury intake and blood mercury
concentrations. The absence of a trend in mercury intake from fish would be consistent with no
significant change in blood mercury concentrations. However, other factors that affect the
relationship and the uncertainty in the blood mercury measurements (reflected in part by the
presence of non-detects) complicate assessment of the relationship between mercury intake and
blood mercury. Even though the mercury intake per body weight is strongly related to blood
45
-------
mercury concentrations, there are changes in blood mercury concentrations that are not predicted by
mercury intake, particularly the drop from the 1999-2000 NHANES release to the subsequent
NHANES releases. This drop may be due to changes in mercury intake not reflected in the fish
consumption data, changes in other factors affecting blood mercury concentrations, changes in the
procedures for measuring mercury concentrations in the blood samples, or to random factors
affecting the selection of NHANES subjects. The magnitude of the drop in blood mercury from the
1999-2000 NHANES release to subsequent releases relative to the uncertainty in the estimates
suggests that the drop in blood mercury concentrations is unlikely to be due to chance. At the same
time, none of the available predictors explain the drop, suggesting that higher levels in 1999-2000
may be due to chance or unidentified causes.
46
-------
Conclusions
The analyses found blood mercury concentrations in NHANES survey release 1999-2000 to be
significantly higher than the mean of the subsequent releases for both blood THg and blood MeHg.
The geometric mean blood THg in 1999-2000 was 1.21 times higher than the geometric mean across
the subsequent 10 years (2001-2010), representing an 18 percent decrease between 1999-2000 and
2001-2010. For blood MeHg, the geometric mean in 1999-2000 was 1.51 times higher than the
geometric mean across the subsequent 10 years. This represents a decrease of 34 percent between
1999-2000 and 2001-2010. Additionally, the percentwith THg >5.8 ug/L and MeHg >5.8 ug/L is
significantly higher in survey release 1999-2000. The percentage of women of reproductive age with
blood THg over 5.8 ug/L in 1999-2000 was 2.64 times that found in 2001-2010, a decrease of 62
percent between 1999-2000 and 2001-2010. For blood MeHg, the percent of women of
reproductive age over 5.8 ug/L in 1999-2000 was 2.86 times higher than the percent of women in
2001-2010, representing a 65 percent decrease between 1999-2000 and 2001-2010. The analysis also
found a significant quadratic trend in blood MeHg concentration since 1999-2000. This trend
indicates decreasing blood MeHg concentrations between the initial sets of NHANES survey
releases, followed by relatively small changes and a slight increase in the last years.
There was a significant relationship between mercury intake from fish consumption and blood
mercury, although mercury intake did not fully explain the differences observed across the survey
releases. The analysis showed few changes in fish consumption and mercury intake over the study
period. There was a marginally statistically significant decreasing trend across NHANES survey
releases in the ratio of mercury intake to fish consumed that is consistent with women shifting their
consumption to fish with lower mercury concentrations; however, other studies are needed to
determine 1) if there is a link between changing consumption patterns and blood mercury and 2) if
fish advisories have led to the changing consumption patterns.
Demographic characteristics were associated with blood mercury as expected: higher concentrations
observed with increasing age and income and higher concentrations observed in the other race
category while lower concentrations observed in Mexican Americans. Similar patterns between fish
consumption and demographic characteristics were found.
47
-------
References
Adams, D.H., McMichael Jr., R.H., and Henderson, G.E. 2003. Mercury levels in marine and
estuarine fishes of Florida 1989-2001. Florida Marine Research Institute. Technical Report
TR-9. 2nd ed. rev. 57 pp.
AhujaJKA, Montville JB, Omolewa-Tomobi G, Heendeniya KY, Martin CL, Steinfeldt LC, Anand
J, Adler ME, LaComb RP, and Moshfegh AJ. 2012. USDA Food and Nutrient Database for
Dietary Studies, 5.0. U.S. Department of Agriculture, Agricultural Research Service, Food
Surveys Research Group, Beltsville, MD.
Alaska State Department of Health and Social Services, Division of Public Health. 2007. Fish
consumption advice for Alaskans: a fish management strategy to optimize the public's
health. Epidemiology Bulletin. Recommendations and Reports. Oct. 11(4).
Arkansas State Department of Environmental Quality. 2010. Surface water quality and fish
monitoring data. Available at:
http://www.adeq.state.ar.us/techsvs/water_quality/monitors.asp.
Bahnick, D., Sauer, C., Butterworth, B., Kuehl, D. 1994. A national study of mercury contamination
offish. Chemosphere 29:537-546
Bjornberg, K.A., Vahter, M., Petersson-Grawe, K., Glynn, A., Cnattingius, S., Darnerud, P.O.,
Atuma, S., Aune, M., Becker, W., and Berglund, M. 2003. MeHg and inorganic mercury in
Swedish pregnant women and in cord blood: influence of fish consumption. Environ Health
Perspect 111:637-641.
Burger, J. and Gochfeld, M. 2006. Mercury in fish available in supermarkets in Illinois: Are there
regional differences? Sci Total Environ. 367:1010-1016.
Caldwell, K.L., Mortensen, M.E.Jones, R.L., Caudill, S.P., Osterloh, J.D. 2009. Total blood mercury
concentrations in the U.S. population: 1999-2006. Int J Hyg Environ Health.
Nov;212(6):588-98.
Centers for Disease Control and Prevention (CDC) 2010a. Data and Documentation. National
Center for Health Statistics (NCHS). National Health and Nutrition Examination Survey
Data. Hyattsville, MD: U.S. Department of Health and Human Services, Centers for Disease
Control and Prevention. http://www.cdc.gov/nchs/nhanes/nhanes_questionnaires.htm.
Accessed July 31, 2010.
Centers for Disease Control and Prevention (CDC) 201 Ob. National Center for Health Statistics
(NCHS). National Health and Nutrition Examination Survey Analytical Guidelines.
Hyattsville, MD: U.S. Department of Health and Human Services, Centers for Disease
48
-------
Control and Prevention, http://www.cdc.gov/nchs/nhanes/nhanes2003-
2004/analytical_guidelines.htm.
Davis, J.A., A.R. Melwani, S.N. Bezalel, J.A. Hunt, G. Ichikawa, A. Bonnema, W.A. Heim, D. Crane,
S. Swenson, C. Lamerdin, and M. Stephenson. 2010. Contaminants in fish from California
lakes and reservoirs: technical report on a two-year screening survey, a report of the surface
water ambient monitoring program (SWAMP). California State Water Resources Control
Board, Sacramento, CA.
Denton, G.M. 2007. Mercury levels in Tennessee fish. Tennessee Department of Environment and
Conservation. Nashville, TN. Available at: http://www.tn.gov/environment
/wpc/publications /pdf/fishmercurylevels.pdf.
Health Canada, 2008. Human Health Risk Assessment of Mercury in Fish and Health Benefits of
Fish Consumption. Bureau of Chemical Safety, Food Directorate, Health Products and
Food Branch, Health Canada. Available at: http://www.hc-sc.gc.ca/fn-
an/pubs/mercur/merc_fish_poisson-eng.php#appa
Hyndman, Rob J. and Fan, Yanan, 1996. Sample quantiles in statistical packages. The American
Statistician, 50, 361-365.
Kudo, Y., Falcigila, G.A., Couch, S.C. 2000. Evolution of meal patterns and food choices of
Japanese-American females born in the United States. Eur J Clin Nutr 54:665670.
Louisiana Department of Environmental Quality. 2012. Mercury Levels in fish. Available at:
http://www.deq.louisiana.gov/portal/tabid/2733/Default.aspx
Mahaffey, K., Clickner, R., Bodurow, C. 2004. Blood organic mercury and dietary mercury intake:
National Health and Nutrition Examination Survey, 1999 and 2000. Environmental Health
Perspectives, 112, 562-570
Mahaffey, K.R., Clickner, R.P., and Jeffries, R.A. 2009. Adult women's blood mercury
concentrations vary regionally in the United States: Association with patterns offish
consumption (NHANES 1999-2004). Environmental Health Perspectives, 117(1), 47-53.
Massachusetts Department of Environmental Protection. Fish Mercury Research Data Portal. 2008,
Available at: http://public.dep.state.ma.us/fish/.
McBride, D. 2005. Analysis of chemical contaminant levels in store-bought fish from Washington
State. Available at: http://epa.gov/waterscience/fish/forum/2005/.
McCarty, H.B., Miller, K., Brent, R.N., and Schofield, J. 2004. Results of the Lake Michigan mass
balance study: mercury data report. Prepared for U.S. EPA Great Lakes National Program
Office. Available at: http://www.epa.gov/glnpo/lmmb/results/mercury/lmmbhg.pdf.
McKelvey, W., Chang, M., Arnason, J., Jeffery, N., Kricheff, J., Kass, D. Mercury and
polychlorinated biphenyls in Asian market fish: a response to results from mercury
biomonitonng in New York City. Environ Res. 2010 Oct;110(7):650-7. Epub 2010 Aug 6.
49
-------
Mergler, D., Anderson, H.A., Chan, L.H., Mahaffey, K.R., Murray, M., Sakamoto, M., and Stern,
A.H. 2007. The Panel on Health Risks and Toxicological Effects of Methylmercury.
Methylmercury exposure and health effects in humans: a worldwide concern. Ambio 36:3
11.
Micro Analytical Systems, Inc. 2008. Safe Harbor News. Winter 2008.
NRC (National Research Council). Committee on the Toxicological Effects of Methylmercury.
2000. Washington, DC: National Academies Press.
Oken, E., Klemman, K.P., Berland, W.E., Simon, S.R., Rich-Edwards, J.W., and Gnllman, M.W.
2003. Decline in fish consumption among pregnant women after a national mercury
advisory. Obstet Gynecol. Aug;102(2):346-51.
Rice, G., Swartout, J., Mahaffey, K., Schoeny, R., 2000. Derivation of the U.S. EPA's oral reference
dose (RfD) for methylmercury. Drug Chem Toxicol. 23(l):41-54.
Rogers, John W., 2003. Estimating the variance of percentiles using replicate weights ASA
Proceedings of the Joint Statistical Meetings, 3525-3532. American Statistical Association
(Alexandria, VA)
Sanzo, J.M., Dorronsoro, M., Amiano, P., Amurrio, A., Aguinagalde, F.X., and Azpiri, M.A. 2001.
Estimation and validation of mercury intake associated with fish consumption in an EPIC
cohort of Spam. Public Health Nutr 4:981-988.
Sarndal, C.E., Swensson, B. and Wretman, J, 1992. Model assisted survey sampling. Springer-Verlag
Inc (Berlin; New York)
SAS Institute Inc. 2010. SAS/STATฎ 9.22 User's Guide. Gary, NC: SAS Institute Inc
Schober, S.E., Sinks, T.H.Jones, R.L., Bolger, P.M., McDowell, M., Osterloh, J., Garrett, E.S.,
Canady, R.A., Dillon, C.F., Sun, Y.Joseph, C.B., and Mahaffey, K.R. 2003. Blood mercury
levels in US children and women of childbearmg age, 1999-2000. JAMA. Apr;289(13):1667-
74.
Sechena R, Liao S, Lorenzana R, Nakano C, Polissar N, Fenske R. 2003. Asian American and Pacific
Islander fish consumptiona community-based study in King County, Washington. J Expo
Anal Environ Epidemiol 13:256266.
Simonin, H., LoukmasJ., Skinner, L. and Roy, K. 2008. Strategic monitoring of mercury in New
York state fish. Department of Environmental Conservation, New York State. For New
York State Energy Research and Development Authority. NYSERDA Report 08-11.
Stern, A.H. and Smith, A.E. 2003. An assessment of the cord blood: maternal blood methylmercury
ratio: implication for risk assessment. Environ Health Perspect 111:1465-70.
-------
Svensson, B.G., Schutz, A., Nilsson, A., Akesson, I., Akesson, B., and Skerfving S. 1992. Fish as a
source of exposure to mercury and selenium. Sci Total Environ 126:6174.
Tsuchiya, A., Hinners, T.A., Burbacher, T.M., Faustman, E.M., Marie'n, K. 2008. Mercury exposure
from fish consumption within the Japanese and Korean communities. J Toxicol Environ
Health. 71(15):1019-31.
USDA Food and Nutrient Database for Dietary Studies, 4.1. 2010. Beltsville, MD: Agricultural
Research Service, Food Surveys Research Group.
USDA Food and Nutrient Database for Dietary Studies, 3.0. 2008. Beltsville, MD: Agricultural
Research Service, Food Surveys Research Group.
USDA Food and Nutrient Database for Dietary Studies, 2.0. 2006. Beltsville, MD: Agricultural
Research Service, Food Surveys Research Group.
USDA Food and Nutrient Database for Dietary Studies, 1.0. 2004. Beltsville, MD: Agricultural
Research Service, Food Surveys Research Group.
U.S. Food and Drug Administration (FDA). 2010. Mercury levels in commercial fish and shellfish
(1990-2010). Available http://www.fda.gov/food/foodsafety/product-specificinformation/
seafood/foodbornepathogenscontaminants/methylmercury/ucml 15644.htm.
U.S. Food and Drug Administration (FDA). 2010. Mercury concentrations in fish: FDA monitoring
program. (1990-2010). Available: http://www.fda.gov/Food/FoodSafety/Product-
SpecificInformation/Seafood/FoodbornePathogensContaminants/Methylmercury/ucm
191007.htm.
U.S. Environmental Protection Agency. Office of Water. 2011. National Water Program Guidance
Fiscal Year 2012. Washington, DC:U.S. Environmental Protection Agency. EPA-850-K-11-
001. Available at: http://www.epa.gov/planandbudget/annualplan/
FY12_OW_NPM_Gdnce.pdf.
U.S. Environmental Protection Agency. 2010. Fiscal Year 2011-2015 EPA Strategic Plan.
Washington, DC:U.S. Environmental Protection Agency. Available at:
http://www.epa.gov/planandbudget/strategicplan.html.
U.S. Environmental Protection Agency. Mercury study report to Congress. 1997. In: An Assessment
of Exposure to Mercury in the United States, Vol 4. EPA-452/R-97-006. Washington, DC:
U.S. Environmental Protection Agency, Office of Air Quality Planning and Standards and
Office of Research and Development.
Virginia Department of Environmental Quality. 2009. Available at: http://www.deq.state.va.us/
fishtissue/fishtissue.html.
Willet, W. 1998. Nutritional Epidemiology. Second Edition. New York, NY: Oxford University
Press.
51
-------
Appendix A
Extended Data Tables
A-l
-------
Table A-l. Mercury concentrations applied to fish species (|jg Hg/g fresh weight)
Species
Bass
Breaded fish products
Catfish
Cod
Flatfish
Haddock
Mackerel
Perch
Pike
Pollock
Porgy
Salmon
Sardine
Sea bass
Shark
Swordfish
Trout
Tuna
Walleye
Other finfish
Finfish, not specified
Clam
Crab
Crayfish
Lobster
Mussel
Oyster
Scallop
Shrimp
Other shellfish
Shellfish, not specified
Hg concentration
(W5 Hg/g wet weight)
0.263
0.013
0.107
0.089
0.054
0.069
0.639
0.143
0.301
0.013
0.315
0.041
0.023
0.188
0.628
1.265
0.045
0.242
0.265
0.097
0.139
0.026
0.057
0.028
0.190
0.026
0.027
0.017
0.014
0.032
0.026
A-2
-------
Table A-2. Distribution of blood THg concentrations (ug/L), by NHANES survey release, age, income and race/ethnicity, worm
16-49 years, NHANES 1999-2010
All Women 16-49
1999-2000
2001-2002
2003-2004
2005-2006
2007-2008
2009-2010
Age
16 to 19 years
20 to 29 years
30 to 39 years
40 to 49 years
Income
<$ 20,000
$20,000 to<$45,000
$45,000 to <$75,000
$75,000 and over
$20,000 and over
Refused/Don't Know
Uncalculated*
Race
Mexican American
Non-Hispanic Black
Non-Hispanic White
Other Hispanic
Other Race
N
1637
1780
1599
1792
1493
1786
2439
2739
2495
2414
2216
2894
1950
2148
225
163
491
2589
2230
4043
751
474
1.
1.
1.
1.
1.
1.
0.
1.
1.
1.
1.
1.
1.
1.
1.
1.
1.
1.
1.
1.
1.
2.
Arith. Mean
(95% Cl)
.98(1.51,2.45)
.43(1.23,1.63)
.35(1.16,1.54)
.44(1.26,1.63)
.26(1.07,1.44)
.39(1.25,1.53)
.93 (0.85,1.01)
.26(1.16,1.36)
.67(1.46,1.88)
.67(1.53,1.81)
.12(0.95,1.28)
.35(1.21,1.49)
.38(1.25,1.51)
.81(1.65,1.97)
.44(1.10,1.79)
.67 (0.94,2.40)
.72(1.31,2.13)
.03(0.93,1.13)
.57(1.41,1.72)
.39(1.27,1.51)
.64(1.19,2.08)
.74(2.31,3.17)
Geometric Mean
(95% Cl)
1.01 (0.84,1.23)
0.83 (0.74,0.93)
0.82 (0.72,0.93)
0.89 (0.80,0.99)
0.79 (0.70,0.87)
0.86 (0.77,0.95)
0.55 (0.51,0.59)
0.75 (0.70,0.80)
0.95 (0.87,1.03)
1.05 (0.99,1.11)
0.68 (0.63,0.74)
0.78 (0.72,0.83)
0.83 (0.78,0.89)
1.10(1.02,1.19)
0.93 (0.76,1.13)
0.91 (0.66,1.25)
0.86 (0.69,1.09)
0.68 (0.64,0.73)
1.02 (0.94,1.10)
0.81 (0.76,0.87)
0.98 (0.85,1.14)
1.51 (1.30,1.74)
Selected percentiles (95% Cl)
25th
0.47 (0.39,0.58)
0.44 (0.38,0.50)
0.43 (0.38,0.50)
0.47 (0.40,0.55)
0.42 (0.38,0.47)
0.45 (0.39,0.52)
0.29 (0.26,0.32)
0.40 (0.38,0.42)
0.50 (0.47,0.53)
0.57 (0.53,0.62)
0.39 (0.36,0.43)
0.40 (0.39,0.41)
0.45 (0.41,0.50)
0.58 (0.53,0.63)
0.51 (0.42,0.63)
0.44 (0.30,0.66)
0.39 (0.30,0.49)
0.40 (0.38,0.42)
0.60 (0.56,0.64)
0.40 (0.39,0.42)
0.51 (0.42,0.62)
0.69 (0.54,0.89)
50th
0.99 (0.
0.85 (0.
0.81 (0.
0.90 (0.
.81,1.20)
.77,0.93)
.70,0.94)
.80,1.00)
0.77 (0.69,0.86)
0.84 (0.
.74,0.94)
0.54 (0.50,0.59)
0.73 (0.
0.90 (0.
1.01 (0.
0.70 (0.
0.78 (0.
0.81 (0.
1.11(1.
0.90 (0.
0.86 (0.
0.90 (0.
0.70 (0.
1.00 (0.
0.80 (0.
1.00 (0.
1.58(1.
.68,0.78)
.83,0.98)
.93,1.09)
.63,0.78)
.72,0.83)
.76,0.86)
.01,1.20)
.71,1.14)
.67,1.09)
.68,1.18)
.66,0.75)
.91,1.10)
.75,0.86)
.86,1.17)
.26,1.98)
75th
2.08(1.56,2.77)
1.65(1.46,1.87)
1.56(1.36,1.80)
1.66(1.42,1.94)
1.43(1.25,1.65)
1.63(1.43,1.87)
1.10(1.00,1.21)
1.46(1.36,1.58)
1.84(1.68,2.02)
1.89(1.74,2.06)
1.25(1.12,1.40)
1.50(1.38,1.62)
1.59(1.46,1.74)
2.20(1.96,2.48)
1.88(1.28,2.75)
1.59(1.03,2.46)
2.14(1.60,2.88)
1.20(1.11,1.30)
1.78(1.63,1.94)
1.60(1.48,1.72)
1.91(1.65,2.20)
3.54(2.89,4.33)
4.
3.
3.
3.
2.
3.
1.
2.
3.
3.
2.
2.
2.
4.
2.
4.
4.
1.
3.
3.
3.
6.
90th
.81 (3.79,6.10)
.05 (2.69,3.46)
.10(2.57,3.73)
.14(2.81,3.51)
.73(2.24,3.33)
.11(2.85,3.39)
.99(1.83,2.17)
.73 (2.47,3.01)
.61(3.16,4.12)
.73 (3.34,4.17)
.29(2.00,2.63)
.70(2.36,3.09)
.96(2.66,3.29)
.18(3.77,4.64)
.90(2.18,3.85)
.20(1.84,9.59)
.18(3.17,5.51)
.99(1.81,2.20)
.13(2.76,3.57)
.19(2.90,3.51)
.16(2.57,3.90)
.16(5.31,7.14)
95th
7.17 (4.93,10.44)
4.52(3.62,5.63)
4.35(3.55,5.33)
4.38(3.70,5.18)
3.82 (3.06,4.77)
4.27(3.86,4.71)
2.71(2.43,3.03)
4.05 (3.50,4.67)
5.53 (4.57,6.67)
5.08(4.48,5.75)
3.02 (2.65,3.46)
3.92(3.31,4.65)
4.68 (4.00,5.48)
6.00(5.19,6.93)
4.25 (2.47,7.32)
7.09 (4.03,12.48)
5.47(3.80,7.86)
2.87(2.58,3.20)
4.42 (3.80,5.15)
4.63(4.11,5.20)
4.32(3.56,5.23)
8.68(6.55,11.51)
A-3
-------
Table A-3. Distribution of blood MeHg concentrations (Mg/L), by NHANES survey release, age, income and race/ethnicity, women
aged 16-49 years, NHANES 1999-2010
All Women 16-49
1999-2000
2001-2002
2003-2004
2005-2006
2007-2008
2009-2010
Age
16 to 19 years
20 to 29 years
30 to 39 years
40 to 49 years
Income
<$ 20,000
$20,000 to<$45,000
$45,000 to <$75,000
$75,000 and over
$20,000 and over
Refused/Don't Know
Uncalculated*
Race
Mexican American
Non-Hispanic Black
Non-Hispanic White
Other Hispanic
Other Race
N
1,637
1,780
1,599
1,792
1,493
1,786
2,439
2,739
2,495
2,414
2,216
2,894
1,950
2,148
225
163
491
2,589
2,230
4,043
751
474
Arith. Mean Geometric Mean
(95% Cl) (95% Cl)
1.84(1.39,2.29)
1.28 (1.09,1.47)
1.08 (0.89,1.27)
1.14(0.98,1.31)
1.01 (0.82,1.19)
1.20(1.07,1.33)
0.78 (0.68,0.88)
1.08 (0.97,1.20)
1.43 (1.23,1.62)
1.42 (1.28,1.56)
0.92 (0.77,1.06)
1.13 (0.99,1.28)
1.16(1.03,1.30)
1.58(1.43,1.74)
1.21 (0.89,1.54)
1.44 (0.75,2.13)
1.54(1.14,1.95)
0.79 (0.70,0.88)
1.33 (1.17,1.48)
1.19(1.07,1.32)
1.39(1.01,1.77)
2.43 (2.05,2.82)
0.94 (0.74,1.19)
0.71 (0.57,0.90)
0.56 (0.40,0.78)
0.60 (0.44,0.82)
0.55 (0.40,0.75)
0.69 (0.56,0.86)
0.43 (0.28,0.67)
0.58 (0.43,0.80)
0.73 (0.58,0.93)
0.78 (0.64,0.97)
0.51 (0.37,0.71)
0.58 (0.43,0.79)
0.63 (0.48,0.83)
0.87 (0.72,1.06)
0.70 (0.52,0.93)
0.69 (0.47,1.02)
0.74 (0.54,1.03)
0.50 (0.36,0.70)
0.78 (0.65,0.94)
0.62 (0.46,0.84)
0.78 (0.61,0.98)
1.25 (1.03,1.52)
Selected percentiles (95% Cl)
25th
0.40 (0.
0.35 (0.
0.26 (0.
0.27 (0.
0.26 (0.
0.33 (0.
0.20 (0.
0.27 (0.
0.34 (0.
0.39 (0.
0.26 (0.
0.28 (0.
0.31 (0.
0.41 (0.
0.36 (0.
0.29 (0.
0.29 (0.
0.27 (0.
0.41 (0.
0.28 (0.
0.38 (0.
0.51 (0.
.31,0.53)
.26,0.46)
.17,0.41)
.18,0.40)
.17,0.40)
.25,0.45)
.08,0.50)
.18,0.41)
.26,0.45)
.31,0.49)
.17,0.39)
.18,0.41)
.22,0.42)
.33,0.50)
.25,0.50)
.17,0.48)
.18,0.46)
.17,0.43)
.33,0.50)
.19,0.42)
.28,0.50)
.39,0.67)
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
1.
50th
.88(0.71,1.08)
.70 (0.60,0.81)
.54 (0.43,0.68)
.63 (0.53,0.75)
.53 (0.44,0.64)
.66 (0.57,0.77)
.40 (0.33,0.49)
.57(0.50,0.65)
.72 (0.65,0.80)
.76 (0.69,0.83)
.51 (0.43,0.59)
.57(0.50,0.65)
.64(0.55,0.73)
.87 (0.78,0.97)
.64 (0.49,0.82)
.63 (0.44,0.90)
.75 (0.55,1.02)
.53 (0.46,0.61)
.79 (0.72,0.87)
.60 (0.53,0.68)
.80 (0.65,0.98)
.32(1.02,1.72)
75th
1.88(1.38,2.57)
1.44(1.27,1.65)
1.20(1.01,1.44)
1.33(1.08,1.63)
1.08(0.87,1.33)
1.40(1.20,1.64)
0.87 (0.76,0.99)
1.23(1.12,1.36)
1.56(1.39,1.75)
1.57(1.42,1.73)
1.00(0.89,1.12)
1.22(1.09,1.35)
1.27(1.13,1.41)
1.92(1.69,2.18)
1.42 (1.00,2.02)
1.36(0.85,2.19)
1.81(1.22,2.69)
0.95 (0.87,1.04)
1.46(1.31,1.63)
1.32(1.20,1.45)
1.65(1.40,1.93)
3.11(2.52,3.84)
4.
2.
2.
2.
2.
2.
1.
2.
3.
3.
1.
2.
2.
3.
2.
3.
3.
1.
2.
2.
2.
5.
90th
.56(3.48,5.97)
.84(2.48,3.25)
.61 (2.08,3.26)
.70(2.31,3.16)
.40(1.88,3.07)
.75 (2.49,3.04)
.71(1.53,1.90)
.48(2.22,2.77)
.34(2.88,3.87)
.32(2.92,3.78)
.95(1.66,2.28)
.41(2.11,2.75)
.58(2.21,3.01)
.82 (3.44,4.25)
.74(1.89,3.96)
.98(1.63,9.72)
.92 (2.93,5.24)
.60(1.45,1.76)
.77(2.38,3.22)
.85(2.55,3.19)
.76(2.25,3.39)
.84 (4.95,6.90)
95th
6.95 (4.73,10.20)
4.29(3.48,5.29)
3.83 (3.07,4.78)
3.96(3.16,4.98)
3.47(2.79,4.32)
4.02 (3.59,4.51)
2.41 (2.08,2.80)
3.82 (3.27,4.45)
4.99(4.06,6.14)
4.65 (4.09,5.29)
2.79(2.37,3.29)
3.54(2.97,4.20)
4.25(3.57,5.05)
5.72 (4.81,6.80)
4.09(2.17,7.71)
6.52(3.67,11.58)
5.27(3.66,7.58)
2.40(2.11,2.73)
4.04 (3.43,4.74)
4.28(3.78,4.83)
4.06(2.87,5.75)
8.48(6.42,11.21)
A-4
-------
Table A-4. Percentages and their standard errors for categorized reports of 30-day frequency of consumption of fish, by NHANES
survey release, income, race/ethnicity, and age for women aged 16-49 years, NHANES 1999-2010
Percent
Parameter
NHANES Survey Release
Total Fish
1999-2000
2001-2002
2003-2004
2005-2006
2007-2008
2009-2010
Finfish Only
1999-2000
2001-2002
2003-2004
2005-2006
2007-2008
2009-2010
Shellfish Only
1999-2000
2001-2002
2003-2004
2005-2006
2007-2008
2009-2010
N
1,637
1,780
1,599
1,792
1,493
1,786
1,637
1,780
1,599
1,792
1,493
1,786
1,637
1,780
1,599
1,792
1,493
1,786
0 times
21.8(2.
16.9(1.
21.3(1.
20.1 (2.
24.7(1.
20.2 (1.
32.9(1.
25.6(1.
31.2(2.
29.8(2.
35.9(1.
32.5(1.
.3)
.1)
.6)
.2)
.5)
.9)
.8)
.5)
.1)
.6)
.6)
.9)
48.3 (3.3)
48.3 (1.
48.9 (2.
44.4 (2.
49.9(1.
42.2(1.
.7)
.4)
.7)
.5)
.9)
1 time
14.7(1.1)
13.3 (1.4)
12.5(1.7)
10.7(1.1)
14.8 (1.0)
14.5 (0.9)
17.7(1.1)
17.2 (1.4)
16.8(1.3)
14.6 (1.0)
16.0(1.2)
15.1(1.0)
19.4(1.8)
17.9(1.7)
16.5(1.1)
17.1(1.7)
17.0 (1.0)
17.0 (0.9)
(Standard Error)
2 times
13.6(1.
13.5(1.
13.0(1.
11.9(0.
11.4 (0.
10.5 (0.
16.9 (0.
17.0(1.
14.6 (0.
12.8 (0.
13.5 (0.
14.3 (1.
9.4(1.
11.8(0.
11.0(1.
12.9(1.
11.1(1.
13.7(1.
.2)
.6)
.1)
.8)
.7)
.6)
.8)
.5)
.9)
.7)
.9)
.2)
.3)
.9)
.2)
.5)
.0)
.0)
3 times
8.9
10.4
9.8
9.4
9.4
7.8
7.5
11.5
10.3
10.1
9.1
8.6
7.0
7.8
8.6
6.9
6.9
9.2
(1.6)
(1.0)
(0.9)
(0.9)
(0.9)
(0.8)
(1.9)
(0.7)
(1.0)
(0.7)
(0.7)
(1.1)
(1.0)
(0.8)
(1.1)
(0.8)
(0.9)
(0.8)
4-5 times
14.6(1.0)
16.9(1.2)
14.4(1.2)
13.8(1.1)
12.9(1.0)
15.2(1.1)
11.4(1.2)
13.4(1.4)
12.0(1.1)
15.3(0.9)
11.6(1.0)
13.3(1.3)
9.4(1.8)
5.6(0.9)
7.8(0.9)
9.3(1.0)
7.3 (0.9)
8.2(1.0)
6+ times
26.
28.
29.
34.
26.
31.
13.
.4(3.0)
.9(1.5)
.0(2.0)
.2(3.0)
.9(1.5)
.8(1.6)
.6(1.9)
15.4(1.3)
15.
17.
14.
16.
6.
8.
7.
9.
7.
9.
.2(1.4)
.4(2.5)
.0(1.3)
.1(1.2)
.5(1.7)
.7(1.1)
.2(1.0)
.5(1.1)
.8(0.7)
.7(1.5)
A-5
-------
Table A-4. Percentages and their standard errors for categorized reports of 30-day frequency of consumption of fish, by NHANES
survey release, income, race/ethnicity, and age for women aged 16-49 years, NHANES 1999-2010 (continued)
Parameter
Income
Total Fish
Finfish Only
Shellfish Only
Percent (Standard Error)
<$20,000
$20,000 to<$45,000
$45,000 to <$75,000
$75,000 and over
$20,000 and over
Refused/Don't Know
Uncalculated*
<$20,000
$20,000 to<$45,000
$45,000 to <$75,000
$75,000 and over
$20,000 and over
Refused/Don't Know
Uncalculated*
<$20,000
$20,000 to<$45,000
$45,000 to <$75,000
$75,000 and over
$20,000 and over
Refused/Don't Know
Uncalculated*
N
2,216
2,894
1,950
2,148
225
163
491
2,216
2,894
1,950
2,148
225
163
491
2,216
2,894
1,950
2,148
225
163
491
0 times
24.8(1.5)
22.9(1.4)
21.6(1.5)
16.2(1.0)
21.3(3.6)
20.7(3.9)
21.5(2.5)
35.7(1.7)
33.6(1.6)
31.8(1.6)
26.2(1.2)
31.1(3.5)
39.0(6.8)
32.7(3.2)
55.2(1.4)
50.3(1.6)
45.8(1.8)
40.4(1.5)
46.6(4.5)
41.0(4.9)
49.2 (3.4)
Itime
16.1(0.9)
14.8(0.9)
13.7(0.9)
10.3 (0.8)
9.9(2.5)
20.9(4.4)
15.2(2.2)
19.7(1.0)
18.0(1.0)
16.2(1.1)
12.7(0.8)
10.6(2.3)
25.4(4.8)
16.6(2.0)
16.2(1.1)
16.7(0.9)
17.5(1.0)
18.3(1.2)
16.8(2.7)
24.8(4.4)
19.7(3.0)
2 times
14.5(1.1)
12.4(0.9)
11.0(0.8)
12.2(0.8)
10.8(3.3)
13.3(3.6)
10.7(2.3)
14.9(1.0)
15.4(1.0)
15.5(1.1)
14.7(0.9)
11.8(3.2)
9.0(2.2)
12.0(2.0)
9.8(0.8)
10.4(0.8)
12.8(1.0)
13.5(0.9)
10.4(2.7)
9.1(2.9)
8.9(1.6)
3 times
9.8(1.1)
9.5 (0.7)
9.7(0.8)
8.5 (0.8)
12.9(2.6)
9.0(2.8)
7.4(1.7)
7.5 (0.9)
8.6(0.6)
9.2 (0.8)
11.1(0.8)
12.9(2.5)
5.2(2.1)
12.6(3.4)
6.0(0.7)
7.7(0.7)
8.0(0.8)
8.8(0.7)
7.7(2.2)
5.4(1.8)
6.1(1.7)
4-5 times
11.8(0.8)
14.4 (0.8)
14.7 (1.0)
16.4 (0.9)
9.6 (2.4)
9.6(2.9)
17.7(2.7)
10.7 (1.0)
10.9 (0.8)
11.8(1.0)
16.9(1.1)
10.7 (2.8)
7.3 (2.4)
12.2(2.5)
5.9 (0.7)
6.9 (0.8)
8.8(1.0)
8.9(1.0)
8.9(2.7)
15.6(4.1)
7.4(1.8)
6+ times
23.0(1.4)
26.0(1.2)
29.4(1.6)
36.3(1.5)
35.5 (3.8)
26.4 (5.8)
27.4 (3.8)
11.6(1.3)
13.5(1.1)
15.5(1.3)
18.4(1.1)
23.0(3.3)
14.2 (5.2)
13.9(2.7)
6.9 (0.8)
8.0 (0.7)
7.1 (0.9)
10.1 (0.9)
9.7(2.9)
4.0(1.6)
8.6(2.3)
A-6
-------
Table A-4. Percentages and their standard errors for categorized reports of 30-day frequency of consumption of fish, by NHANES
survey release, income, race/ethnicity, and age for women aged 16-49 years, NHANES 1999-2010 (continued)
Percent (Standard
Parameter
Race/Ethnicity
Total Fish
Mexican American
Other Hispanic
Non-Hispanic White
Non-Hispanic Black
Other Race
Finfish Only
Mexican American
Other Hispanic
Non-Hispanic White
Non-Hispanic Black
Other Race
Shellfish Only
Mexican American
Other Hispanic
Non-Hispanic White
Non-Hispanic Black
Other Race
N
2,
4,
2,
2,
4,
2,
2,
4,
2,
589
751
043
230
474
589
751
043
230
474
589
751
043
230
474
0 times
22.9(1.
21.1(2.
21.4 (0.
17.1(1.
19.5 (2.
38.9(1.
34.7 (2.
31.5(1.
25.8(1.
25.7(2.
44.7 (1.
47.5 (3.
48.3 (1.
45.5(1.
38.4 (3.
.1)
.3)
.9)
.2)
.6)
.5)
.6)
.0)
.3)
.6)
.6)
.0)
.2)
.4)
.1)
Itime
16.9(1.
20.3 (1.
12.9 (0.
12.9 (0.
.0)
.9)
.7)
.8)
7.1(1.2)
21.4(1.
20.9 (2.
14.9 (0.
19.0(1.
11.9(1.
21.6(1.
19.5(1.
17.5 (0.
16.3(1.
10.6(1.
.0)
.4)
.7)
.1)
.9)
.1)
.7)
.8)
.1)
.7)
2 times
15.9(1.0)
10.2(1.6)
12.3(0.6)
13.0(0.8)
7.2(1.2)
15.0(0.8)
13.9(1.5)
15.1(0.6)
15.7(0.9)
10.7(1.6)
14.2 (0.9)
10.5(1.4)
11.3(0.6)
12.7(0.8)
10.6(1.6)
Error)
3 times
11.8(0
9.7(1.
9.2 (0.
8.6 (0.
7.0(1.
8.7 (0.
7.1(1.
9.6 (0.
10.3 (0
.8)
2)
6)
7)
4)
8)
1)
6)
.8)
10.3(1.4)
8.2 (0.
9.1(1.
7.2 (0.
7.4 (0.
11.8(1
5)
2)
5)
7)
.8)
4-5 times
14.2 (0.9)
12.3(1.5)
14.4 (0.6)
18.2(1.1)
12.6(1.8)
8.3 (0.6)
8.4(1.2)
13.4 (0.7)
13.6 (0.9)
17.2 (1.9)
6.5 (0.7)
5.8(1.1)
7.9(0.6)
8.4(0.6)
11.6(1.5)
6+ times
18.3(1.1)
26.5(2.8)
29.8(1.2)
30.2(1.2)
46.6(3.4)
7.8 (0.8)
15.0(2.2)
15.5(0.8)
15.6(0.9)
24.1(3.0)
4.7 (0.5)
7.7(1.2)
7.7 (0.7)
9.6 (0.8)
17.0(2.2)
A-7
-------
Table A-4. Percentages and their standard errors for categorized reports of 30-day frequency of consumption of fish, by NHANES
survey release, income, race/ethnicity, and age for women aged 16-49 years, NHANES 1999-2010 (continued)
Percent (Standard Error)
Parameter
Age
Total Fish
16 to
20 to
30 to
40 to
Finfish Only
16 to
20 to
30 to
40 to
Shellfish Only
16 to
20 to
30 to
40 to
19
29
39
49
19
29
39
49
19
29
39
49
years
years
years
years
years
years
years
years
years
years
years
years
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
2,
N
439
739
495
414
439
739
495
414
439
739
495
414
0 times
38.
23.
18.
.9(1.6)
.4(1.3)
.2(1.1)
1.05(1.0)
53.
36.
28.
22.
58.
47.
43.
45.
.7(1.7)
.1(1.4)
.7(1.2)
.1(1.1)
.8(1.4)
.9(1.4)
.7(1.5)
.3(1.5)
Itime
17.9(1.2)
14.2(1.0)
12.5(1.0)
12.1(0.8)
16.5(1.1)
16.9(1.0)
16.1(0.9)
15.6(0.9)
17.9(1.3)
18.2(1.0)
18.1(1.0)
16.0(1.0)
2 times
10.4(0.7)
13.2(0.9)
12.4(0.9)
12.0(0.7)
9.3 (0.8)
13.7(0.9)
15.4(0.9)
17.2(1.0)
9.4(0.8)
11.5(0.8)
11.4(0.8)
12.7(0.9)
3 times
7.2 (0.7)
9.8(0.9)
9.8(0.7)
9.0(0.7)
6.2 (0.7)
8.9(0.7)
10.3 (0.7)
10.4(0.8)
4.7(0.5)
7.6(0.7)
8.2 (0.7)
8.4(0.7)
4-5 times
10.6 (0.9)
13.6(1.0)
14.6 (0.8)
16.8 (0.8)
6.9 (0.8)
11.3 (0.8)
13.2 (0.8)
15.8 (0.9)
4.9 (0.5)
7.4 (0.7)
8.8 (0.8)
8.6 (0.9)
6+ times
15.1(1.1)
25.7(1.4)
32.4(1.4)
35.1(1.3)
7.4 (0.8)
13.1(1.1)
16.2 (0.9)
19.0(1.2)
4.3 (0.7)
7.4 (0.7)
9.7 (0.8)
8.9 (0.8)
-------
Table A-5. Estimated amount of fish consumed (g) in last 30 days, by NHANES survey release, women aged 16-49 years, NHANES
1999-2010
Parameter
Amount of
Shellfish
Finflsh
Total Fish
Years
Fish Eaten (gm)
1999-2000
2001-2002
2003-2004
2005-2006
2007-2008
2009-2010
1999-2000
2001-2002
2003-2004
2005-2006
2007-2008
2009-2010
1999-2000
2001-2002
2003-2004
2005-2006
2007-2008
2009-2010
N
1,637
1,780
1,599
1,792
1,493
1,786
1,637
1,780
1,599
1,792
1,493
1,786
1,637
1,780
1,599
1,792
1,493
1,786
Arith. Mean
(95% Cl)
69.3 (49.9,88.8)
73.2 (64.0,82.4)
66.9 (54.4,79.4)
79.7(70.2,89.1)
67.2 (58.7,75.8)
86.0(67.3,104.7)
185.2(159.8,210.6)
237.2 (203.7,270.8)
203.3(177.6,229.1)
242.8 (204.7,280.9)
191.8(167.7,215.9)
222.5(193.7,251.3)
254.6(213.4,295.8)
310.5 (275.0,345.9)
270.2 (235.3,305.2)
322.5 (277.1,367.8)
259.0(228.5,289.6)
308.5 (269.3,347.8)
Selected percentiles (95% Cl)
25th
0.8 (0.0,7.2)
0.8 (0.0,4.1)
0.5 (0.0,5.2)
2.9 (0.0,8.6)
0.0 (0.0,2.9)
4.3(0.1,8.5)
21.4(12.4,30.4)
39.3 (30.1,43.2)
25.2(13.8,36.5)
28.1(13.5,42.0)
16.4(9.2,23.5)
22.4(13.0,31.8)
27.2 (22.2,43.8)
58.7(43.9,68.3)
41.3(22.7,58.5)
43.5 (23.0,69.2)
24.1(19.0,27.4)
38.2 (24.4,50.0)
50th
24.1(18.8,25.4)
24.1(21.8,24.8)
24.1(19.9,25.0)
24.9 (23.7,30.8)
23.4 (20.6,24.4)
27.0(24.7,36.5)
98.9 (81.4,102.9)
121.4(103.0,149.4)
101.3 (95.0,128.8)
129.3(100.7,159.0)
96.5 (70.9,104.1)
103.0(94.6,137.9)
136.0(117.7,162.4)
169.5(148.9,204.0)
159.8(128.2,186.7)
186.1(147.7,236.9)
137.5(108.2,158.2)
170.1(143.1,204.3)
75th
88.2(54.3,135.5)
81.9 (68.4,92.1)
88.5 (69.2,96.5)
101.8 (86.8,124.8)
78.0(66.3,91.2)
98.2 (85.7,126.6)
238.6(217.4,286.9)
276.8 (255.5,317.3)
276.6(246.1,318.5)
327.5 (277.3,382.3)
255.5(225.5,292.5)
315.1(260.5,366.1)
329.7 (279.7,400.6)
378.9 (343.4,411.1)
374.8 (321.7,414.4)
440.5 (377.6,524.7)
357.5 (312.5,398.6)
424.4 (376.0,474.7)
90th
185.8(148.5,284.5)
217.5(185.3,251.1)
191.8(158.2,228.9)
233.4(199.7,255.4)
195.9(165.7,221.5)
231.2(187.3,303.0)
497.5 (412.9,612.4)
513.5 (467.7,588.5)
524.8 (439.7,610.0)
616.1(526.8,744.2)
522.1(466.2,572.2)
543.0 (506.8,657.6)
663.3 (567.8,802.6)
717.7 (639.4,789.7)
646.8 (575.6,769.7)
792.0 (672.9,960.9)
653.7(567.8,796.6)
768.0 (672.8,868.2)
95th
284.7(217.9,419.4)
335.5(285.5,419.9)
301.8(236.2,382.7)
344.3 (299.3,420.2)
312.3(265.0,360.2)
371.0(287.6,530.2)
684.1(597.1,851.0)
790.4 (722.6,924.4)
724.2 (653.4,873.4)
863.2 (744.2,1087.7)
719.4 (637.9,868.8)
829.7(711.9,939.9)
875.9 (769.1,1073.7)
1012.8 (874.9,1205.2)
926.4(836.9,1044.9)
1085.5 (976.4,1254.7)
940.6(817.9,1157.5)
1111.8(950.5,1235.1)
A-9
-------
Table A-6. Estimated mercury intake (|jg) in last 30 days, by NHANES survey release, women aged 16-49 years, NHANES 1999-2010
Parameter
Years
N
Arith. Mean
(95% Cl)
Selected percentiles (95% Cl)
25th
50th
75th
90th
95th
Intake of MeHg (ng)
Shellfish
Finflsh
Total Fish
1999-2000
2001-2002
2003-2004
2005-2006
2007-2008
2009-2010
1999-2000
2001-2002
2003-2004
2005-2006
2007-2008
2009-2010
1999-2000
2001-2002
2003-2004
2005-2006
2007-2008
2009-2010
1,637
1,780
1,599
1,792
1,493
1,786
1,637
1,780
1,599
1,792
1,493
1,786
1,637
1,780
1,599
1,792
1,493
1,786
2.75(1.86,3.63)
2.71(2.40,3.02)
2.26(1.73,2.80)
2.85(2.38,3.32)
2.25(1.89,2.61)
2.77(2.11,3.42)
28.26(23.24,33.29)
34.68 (26.73,42.64)
27.44(23.68,31.20)
29.12 (25.58,32.67)
27.08 (22.15,32.01)
28.65 (25.24,32.06)
31.01(25.40,36.62)
37.40 (29.36,45.43)
29.70(25.60,33.81)
31.97(28.14,35.81)
29.33 (24.21,34.46)
31.42 (27.83,35.00)
0.01(0.00,0.10)
0.01 (0.00,0.06)
0.01 (0.00,0.07)
0.04(0.00,0.12)
0.00 (0.00,0.04)
0.06(0.00,0.12)
0.38 (0.22,0.53)
0.69 (0.53,0.87)
0.44(0.24,0.63)
0.49 (0.24,0.73)
0.28 (0.16,0.41)
0.39 (0.23,0.56)
0.39 (0.32,0.81)
1.72 (0.78,3.18)
0.74(0.33,1.74)
1.02 (0.32,3.10)
0.34 (0.27,0.39)
0.57(0.35,1.42)
0.34 (0.27,0.36)
0.34(0.31,0.35)
0.34(0.28,0.35)
0.35 (0.34,0.42)
0.33 (0.29,0.35)
0.38 (0.35,0.52)
10.48 (8.68,11.27)
13.81(11.16,16.68)
10.48 (8.85,13.32)
13.35(10.56,16.89)
10.12 (8.31,10.53)
11.91(10.45,15.20)
11.32(10.49,14.64)
17.51(13.76,20.04)
12.20(10.55,16.01)
15.82(13.69,19.13)
10.51(9.48,12.45)
14.54(11.42,17.40)
1.95 (0.88,3.47)
1.94(1.53,2.54)
1.81(1.30,2.49)
2.57(1.81,3.35)
1.77(1.31,2.44)
2.56(1.94,3.13)
29.62 (24.77,38.29)
38.25 (33.40,41.71)
35.16 (30.21,39.10)
38.57(35.91,45.03)
30.80 (26.46,35.89)
36.60(31.34,39.20)
35.94(27.50,41.22)
39.51(37.86,44.32)
37.79 (32.83,42.45)
43.46(38.36,49.22)
34.38 (29.83,38.31)
39.40(36.14,44.76)
8.88(4.92,13.47)
8.76 (6.69,9.70)
6.71(4.48,9.58)
9.32(6.87,12.21)
6.41(4.91,9.14)
8.48 (6.18,10.89)
70.33 (57.39,96.75)
72.02 (66.53,83.12)
71.25 (66.23,79.73)
71.39 (66.44,81.72)
67.72 (58.50,86.06)
72.39 (66.53,79.41)
79.07(64.07,100.06)
77.57(70.23,91.48)
78.40 (68.60,82.49)
78.88 (72.25,92.35)
73.86(61.59,97.82)
79.40 (72.04,85.56)
14.49(11.50,17.61)
13.20(12.42,14.43)
12.99(9.62,16.07)
13.60(12.61,15.98)
12.23(9.87,14.23)
13.62(10.11,18.61)
132.27(98.82,159.01)
125.18 (99.34,158.28)
104.80(87.23,122.35)
108.62 (93.63,129.74)
121.74(96.42,135.42)
109.16(95.39,122.49)
140.17(102.81,176.54)
125.40(105.30,165.19)
105.86(96.12,123.05)
119.79(102.70,132.83)
128.48(103.75,141.44)
118.68 (98.00,130.99)
A-10
-------
Table A-7. Estimated mercury intake per unit body weight (|jg Hg/kg bw) in last 30 days, by NHANES survey release, women aged 16-
49 years, NHANES 1999-2010
Parameter Years
Intake of MeHg per Unit Body weight (tig/kg)
Shellfish
1999-2000
2001-2002
2003-2004
2005-2006
2007-2008
2009-2010
Finfish
1999-2000
2001-2002
2003-2004
2005-2006
2007-2008
2009-2010
Total Fish
1999-2000
2001-2002
2003-2004
2005-2006
2007-2008
2009-2010
N
1,637
1,780
1,599
1,792
1,493
1,786
1,637
1,780
1,599
1,792
1,493
1,786
1,637
1,780
1,599
1,792
1,493
1,786
Arith. Mean
(95% Cl)
0.04 (0.03,0.05)
0.04 (0.03,0.04)
0.03 (0.02,0.04)
0.04 (0.04,0.05)
0.03 (0.03,0.04)
0.04 (0.03,0.05)
0.41 (0.33,0.49)
0.50 (0.38,0.63)
0.40 (0.34,0.47)
0.42 (0.36,0.47)
0.39 (0.31,0.46)
0.41 (0.36,0.47)
0.45 (0.36,0.54)
0.54 (0.41,0.67)
0.44(0.37,0.51)
0.46 (0.39,0.52)
0.42 (0.34,0.49)
0.45 (0.40,0.51)
Selected percentiles (95% Cl)
25th
0.00 (0.00,0.00)
0.00 (0.00,0.00)
0.00 (0.00,0.00)
0.00 (0.00,0.00)
0.00 (0.00,0.00)
0.00 (0.00,0.00)
0.00 (0.00,0.00)
0.00 (0.00,0.01)
0.00 (0.00,0.00)
0.00 (0.00,0.01)
0.00 (0.00,0.00)
0.00 (0.00,0.00)
0.01 (0.00,0.01)
0.02 (0.01,0.05)
0.01 (0.00,0.02)
0.02 (0.00,0.05)
0.00 (0.00,0.01)
0.01 (0.00,0.02)
50th
0.00 (0.00,0.01)
0.00 (0.00,0.00)
0.00 (0.00,0.01)
0.00 (0.00,0.01)
0.00 (0.00,0.00)
0.01 (0.00,0.01)
0.15 (0.12,0.17)
0.20(0.17,0.24)
0.15 (0.13,0.18)
0.19 (0.16,0.23)
0.13 (0.10,0.15)
0.17(0.14,0.21)
0.17(0.15,0.20)
0.24 (0.20,0.28)
0.17(0.15,0.21)
0.23 (0.19,0.28)
0.15 (0.12,0.17)
0.20(0.17,0.24)
75th
0.03 (0.01,0.05)
0.03 (0.02,0.04)
0.02 (0.02,0.03)
0.04 (0.03,0.05)
0.02 (0.02,0.03)
0.03 (0.03,0.05)
0.41(0.37,0.48)
0.55 (0.48,0.59)
0.47 (0.40,0.57)
0.56(0.51,0.63)
0.43 (0.38,0.47)
0.50(0.42,0.58)
0.46 (0.40,0.56)
0.59 (0.54,0.64)
0.52 (0.44,0.60)
0.61(0.55,0.67)
0.47(0.42,0.53)
0.55 (0.47,0.64)
90th
0.12 (0.07,0.18)
0.11(0.10,0.14)
0.10(0.07,0.14)
0.13 (0.10,0.16)
0.10(0.07,0.13)
0.11(0.09,0.16)
1.01(0.82,1.40)
1.07(0.97,1.20)
1.02 (0.87,1.20)
1.05 (0.90,1.22)
1.01(0.87,1.26)
1.06(0.97,1.19)
1.13 (0.95,1.43)
1.13(1.04,1.30)
1.10(0.90,1.28)
1.16(0.99,1.30)
1.10(0.93,1.43)
1.15(1.05,1.29)
95th
0.20(0.17,0.28)
0.20(0.17,0.24)
0.19(0.14,0.26)
0.22(0.18,0.26)
0.17(0.14,0.23)
0.21(0.16,0.28)
1.81(1.39,2.59)
1.65(1.45,2.22)
1.59(1.33,1.95)
1.60(1.31,1.99)
1.65(1.34,2.14)
1.57(1.39,1.89)
2.02(1.43,2.73)
1.68(1.52,2.27)
1.67(1.39,2.01)
1.75(1.47,2.10)
1.71(1.45,2.33)
1.69(1.47,2.00)
A-ll
-------
Table A-8. Estimated amounts consumed in last 30 days; amount offish consumed (g), mercury intake (|jg), and mercury intake
unit body weight (|jg/kg), by income, race/ethnicity, and age, women aged 16-49 years, NHANES 1999-2010
per
arameter
Amount of Fish
Income
Race/Ethnicity
Age
Eaten (gm)
<$20,000
$20,000 to<$45,000
$45,000 to <$7B,000
$75,000 and over
$20,000 and over
Refused/Don't Know
Uncalculated*
Mexican American
Other Hispanic
Non-Hispanic White
Non-Hispanic Black
Other Race
16 to 19 years
20 to 29 years
30 to 39 years
40 to 49 years
N
2,216
2,894
1,950
2,148
225
163
491
2,589
751
4,043
2,230
474
2,439
2,739
2,495
2,414
Arith. Mean
(95% Cl)
243.5 (216.5,270.5)
266.0(243.0,289.1)
288.3 (257.0,319.6)
326.6(305.2,347.9)
343.3 (276.5,410.0)
283.2(155.0,411.5)
285.6(228.6,342.7)
268.6(244.9,292.2)
282.3(235.1,329.5)
270.8 (252.0,289.5)
315.0(292.0,337.9)
448.8 (348.5,549.2)
160.3(141.0,179.5)
253.3(232.4,274.2)
310.0(286.7,333.3)
339.5(311.5,367.6)
Selected percentiles (95% Cl)
25th
23.8(18.1,39.8)
26.8(21.9,42.8)
28.3 (24.3,43.5)
63.7(44.7,69.2)
42.4(16.2,70.4)
27.5 (10.9,60.6)
26.5 (17.9,59.6)
34.1(28.8,41.5)
29.1(19.4,49.1)
34.4 (24.5,43.4)
59.1 (55.8,66.7)
59.3(22.9,99.7)
6.70(3.00,10.5)
24.8(21.7,37.9)
55.8 (42.5,60.9)
69.1(59.1,83.9)
50th
118.2 (100.7,134.1)
139.3 (126.9,159.0)
157.4 (137.5,181.0)
217.1 (194.0,229.1)
202.8 (125.4,274.3)
115.3 (73.2,182.6)
158.5 (122.2,195.4)
148.2 (136.7,150.2)
139.3 (106.9,175.1)
158.3 (138.6,162.5)
181.7(168.8,199.3)
268.9 (209.4,339.2)
53.6(42.5,64.9)
129.4(115.2,148.7)
179.8 (159.7,198.8)
215.9 (194.9,229.8)
75th
318.4(275.0,362.8)
338.5(310.3,375.8)
364.7(331.8,403.1)
462.1 (427.1,495.4)
509.9 (456.2,672.5)
347.9 (201.9,604.1)
373.1(286.1,496.8)
343.8(321.2,380.3)
377.2 (284.5,468.4)
370.9 (343.8,400.3)
396.0(373.0,436.9)
583.8(511.3,683.4)
197.5 (168.0,222.8)
340.7(297.3,380.5)
411.3(377.1,445.3)
456.2 (420.2,496.1)
90th
630.8 (546.1,688)
653.0(585.1,759)
701.4(632.0,779.9)
769.4(710.7,837.8)
784.5 (697.0,1180)
805.9 (469.6,1579)
690.3 (603.1,981.9)
649.9 (596.7,725.4)
741.5 (622.5,876.4)
673.6(628.4,731.0)
756.7(682.9,831.6)
1065 (885.0,1285)
434.6 (390.7,506.9)
653.5(583.3,733.1)
731.8 (667.2,804.3)
781.6(720.8,856.8)
95th
926.5(780.5,1103)
1025(884.7,1152)
929.1(846.8,1075)
1008(942.6,1106)
1167(853.8,1534)
1353 (702.4,2276)
1021(799.7,1506)
982.5(883.1,1155)
1040(862.2,1530)
923.6 (856.8,992.0)
1101(989.0,1212)
1412(1177,1786)
687.2 (609.0,763.6)
941.6(866.5,1058)
1031(952.3,1112)
1075(951.3,1212)
Intake of MeHg (ng)
Income
Race/Ethnicity
<$20,000
$20,000 to<$45,000
$45,000 to <$75,000
$75,000 and over
$20,000 and over
Refused/Don't Know
Uncalculated*
Mexican American
Other Hispanic
Non-Hispanic White
Non-Hispanic Black
Other Race
2,216
2,894
1,950
2,148
225
163
491
2,589
751
4,043
2,230
474
25.61(21.95,29.28)
29.27(25.90,32.64)
31.83(25.50,38.15)
36.40(33.60,39.20)
36.52 (30.20,42.83)
38.73(16.38,61.09)
34.37(22.25,46.49)
27.68 (24.75,30.61)
31.15(24.90,37.39)
30.98 (28.62,33.35)
31.09 (28.44,33.74)
49.47(27.83,71.10)
0.34(0.26,0.66)
0.38 (0.31,0.75)
0.40 (0.34,0.85)
2.80(1.67,3.62)
0.90(0.23,3.88)
0.39(0.15,1.58)
0.38(0.25,1.83)
0.48 (0.41,0.66)
0.42 (0.28,0.88)
0.62 (0.35,0.81)
1.92 (0.90,2.81)
1.94(0.32,5.86)
10.49 (8.95,11.39)
11.41 (10.50,13.76)
12.77(10.83,15.18)
19.70(17.10,22.33)
21.14(10.85,29.14)
7.07(1.67,13.40)
12.18 (10.35,16.09)
12.82 (9.52,14.13)
11.88 (9.84,13.00)
13.34(11.54,15.21)
13.68 (12.34,15.82)
21.06(16.29,27.28)
29.15 (25.02,34.43)
34.73 (30.62,37.86)
38.39 (34.57,41.69)
47.37(43.67,51.93)
52.97(41.46,66.04)
33.57(13.92,71.35)
39.88 (30.32,45.34)
35.99 (34.59,37.81)
34.08 (28.35,44.44)
38.42 (37.01,40.53)
37.07(32.70,41.60)
51.31(43.76,67.69)
66.09 (57.05,74.27)
72.69 (66.39,82.84)
76.57(69.68,83.72)
85.32 (79.45,92.51)
95.91(81.86,124.7)
107.0(56.98,372.4)
80.78(54.44,116.4)
71.95 (63.07,79.72)
87.74(76.42,95.78)
74.64(69.87,81.13)
76.19 (68.49,84.95)
105.7(86.12,138.9)
106.7(87.62,133.1)
123.2 (104.3,151.6)
110.5 (96.38,134.9)
128.5 (116.7,139.6)
126.6(113.6,202.3)
242.4 (89.55,512.8)
137.1 (92.39,405.5)
109.1 (93.60,137.3)
124.2 (98.13,157.1)
118.4(105.6,132.5)
119.3 (107.1,134.3)
148.3 (125.9,184.3)
A-12
-------
Table A-8. Estimated amounts consumed in last 30 days; amount offish consumed (g), mercury intake (|jg), and mercury intake per
unit body weight (|jg/kg), by income, race/ethnicity, and age, women aged 16-49 years, NHANES 1999-2010 (continued)
Parameter
N
Arith. Mean
(95% Cl)
Selected percentiles (95% Cl)
25th
50th
75th
90th
95th
Intake of MeHg (u.g) continued
Age
16 to 19 years
20 to 29 years
30 to 39 years
40 to 49 years
Intake of MeHg per Unit Body weight
Income
Race/Ethnicity
Age
<$20,000
$20,000 to<$45,000
$45,000 to <$7B,000
$75,000 and over
$20,000 and over
Refused/Don't Know
Uncalculated*
Mexican American
Other Hispanic
Non-Hispanic White
Non-Hispanic Black
Other Race
16 to 19 years
20 to 29 years
30 to 39 years
40 to 49 years
2,439
2,739
2,495
2,414
(ug/kg)
2,216
2,894
1,950
2,148
225
163
491
2,589
751
4,043
2,230
474
2,439
2,739
2,495
2,414
17.70(15.19,20.22)
27.90(25.09,30.72)
34.69 (31.09,38.29)
37.26(32.44,42.08)
0.36 (0.30,0.41)
0.42 (0.37,0.47)
0.46 (0.36,0.55)
0.54 (0.49,0.58)
0.52 (0.42,0.62)
0.61 (0.26,0.96)
0.52 (0.31,0.73)
0.40 (0.36,0.45)
0.47 (0.36,0.57)
0.45 (0.41,0.49)
0.40 (0.36,0.43)
0.81 (0.47,1.15)
0.29 (0.25,0.33)
0.42 (0.37,0.47)
0.50 (0.44,0.55)
0.52 (0.44,0.59)
0.10(0.04,0.15)
0.35 (0.31,0.57)
1.12(0.74,1.81)
3.72 (2.70,6.09)
0.00 (0.00,0.01)
0.01 (0.00,0.01)
0.01 (0.00,0.01)
0.04 (0.02,0.05)
0.01 (0.00,0.05)
0.01 (0.00,0.02)
0.01 (0.00,0.03)
0.01 (0.00,0.01)
0.01 (0.00,0.01)
0.01 (0.01,0.01)
0.02 (0.01,0.04)
0.03 (0.01,0.09)
0.00 (0.00,0.00)
0.00 (0.00,0.01)
0.01 (0.01,0.02)
0.05 (0.04,0.07)
1.49 (0.77,2.75)
11.24(10.50,12.56)
16.02(13.30,18.55)
18.89 (17.05,20.91)
0.13 (0.11,0.15)
0.17(0.15,0.19)
0.18 (0.15,0.21)
0.27 (0.25,0.31)
0.29 (0.18,0.34)
0.10 (0.02,0.21)
0.18 (0.15,0.23)
0.17(0.13,0.19)
0.16(0.12,0.22)
0.19 (0.18,0.21)
0.17 (0.15,0.20)
0.34 (0.25,0.44)
0.02 (0.01,0.04)
0.17(0.15,0.19)
0.22 (0.19,0.25)
0.25 (0.23,0.27)
19.19 (15.61,22.64)
35.79 (32.48,38.39)
40.63 (37.40,44.60)
44.29 (40.92,48.79)
0.41 (0.36,0.45)
0.46(0.41,0.52)
0.53 (0.47,0.59)
0.67(0.61,0.73)
0.80(0.55,1.20)
0.40(0.21,1.18)
0.56 (0.40,0.74)
0.50 (0.45,0.55)
0.53 (0.44,0.62)
0.54 (0.49,0.58)
0.45 (0.41,0.51)
0.86(0.74,1.05)
0.29 (0.23,0.37)
0.50 (0.45,0.55)
0.58 (0.52,0.63)
0.59 (0.56,0.65)
51.94 (43.43,61.30)
71.95 (66.32,81.72)
79.98 (71.63,91.55)
85.79 (79.80,93.22)
0.94(0.84,1.12)
1.05(0.90,1.21)
1.04(0.96,1.20)
1.27(1.14,1.42)
1.36(1.21,1.78)
2.02 (0.78,5.83)
1.08(0.81,1.61)
1.00(0.90,1.17)
1.21(1.07,1.44)
1.10(1.01,1.19)
0.98(0.87,1.09)
1.76(1.38,2.23)
0.81 (0.72,0.95)
1.13(1.01,1.25)
1.13(1.04,1.29)
1.20(1.09,1.35)
80.79(69.24,102.8)
106.4(96.85,126.4)
131.8(115.7,148.0)
131.4(118.2,145.3)
1.48 (1.26,2.10)
1.80 (1.57,2.00)
1.63 (1.33,1.93)
1.87(1.67,2.17)
1.69 (1.39,3.73)
3.83 (1.52,7.79)
1.66(1.28,6.61)
1.65 (1.47,2.03)
1.85 (1.44,2.73)
1.68 (1.56,1.84)
1.51 (1.41,1.77)
2.52 (2.05,3.64)
1.38 (1.15,1.69)
1.63 (1.52,1.80)
1.86(1.62,2.17)
1.86(1.64,2.16)
A-13
-------
Table A-9. Blood MeHg concentrations (ug/L), by frequency of consuming fish, by NHANES survey release, women aged 16-49 years,
NHANES 1999-2010
E'vey
sase
1999-2000
2001-2002
2003-2004
Times eaten
in 30 days
0
1
2
3
4-5
6 and up
0
1
2
3
4-5
6 and up
0
1
2
3
4-5
6 and up
N
428
279
223
154
227
326
401
248
250
188
274
419
365
237
205
162
241
389
Arith. Mean
(95% Cl)
0.60 (0.50,0.71)
1.01 (0.72,1.30)
1.17 (0.93,1.40)
2.06 (0.26,3.86)
2.27 (1.74,2.80)
3.36(2.75,3.97)
0.43 (0.33,0.54)
0.71 (0.53,0.90)
0.84 (0.67,1.01)
1.14 (0.78,1.50)
1.20 (0.99,1.42)
2.33 (1.92,2.75)
0.38 (0.27,0.50)
0.50 (0.36,0.65)
0.65 (0.52,0.78)
0.89 (0.69,1.08)
1.15 (1.00,1.29)
2.07 (1.68,2.46)
Selected percentiles (95% Cl)
25th
0.21 (0.12,0.39)
0.37 (0.25,0.53)
0.38 (0.27,0.53)
0.46 (0.16,1.33)
0.67 (0.54,0.82)
1.02 (0.81,1.30)
0.16 (0.05,0.49)
0.30 (0.19,0.46)
0.32 (0.22,0.46)
0.37 (0.23,0.59)
0.44 (0.34,0.58)
0.73 (0.62,0.86)
0.14 (0.03,0.57)
0.15 (0.07,0.34)
0.26 (0.15,0.46)
0.31 (0.20,0.46)
0.41 (0.29,0.58)
0.64 (0.51,0.81)
50th
0.36 (0.27,0.47)
0.70 (0.54,0.90)
0.78 (0.58,1.04)
0.90 (0.60,1.37)
1.22 (0.89,1.67)
1.91 (1.52,2.41)
0.29 (0.21,0.41)
0.48 (0.38,0.60)
0.57 (0.43,0.77)
0.77 (0.58,1.01)
0.85 (0.68,1.06)
1.41 (1.28,1.55)
0.25 (0.16,0.40)
0.33 (0.20,0.53)
0.47 (0.36,0.61)
0.54 (0.44,0.66)
0.76 (0.63,0.92)
1.21 (0.99,1.48)
75th
0.69(0.56,0.85)
1.18(0.79,1.75)
1.54(1.30,1.83)
1.40 (0.49,4.02)
2.83 (1.88,4.27)
4.38(3.60,5.34)
0.55 (0.42,0.72)
0.79(0.63,1.01)
1.10(0.85,1.42)
1.31(1.05,1.63)
1.60(1.24,2.06)
2.82(2.53,3.14)
0.44 (0.34,0.57)
0.57(0.38,0.85)
0.77 (0.64,0.93)
1.13(0.81,1.58)
1.33(1.12,1.57)
2.61(1.92,3.54)
90th
1.32(1.08,1.61)
2.12(1.27,3.53)
2.50(1.75,3.57)
3.77(0.86,16.40)
5.69(4.43,7.29)
8.33(6.28,11.07)
0.92 (0.77,1.09)
1.52 (0.95,2.43)
1.76(1.38,2.25)
2.50(1.33,4.70)
2.58(1.96,3.39)
5.69(4.27,7.59)
0.79(0.57,1.10)
1.07(0.70,1.63)
1.43 (1.03,2.00)
1.92(1.37,2.68)
2.57(1.81,3.65)
4.36(3.20,5.92)
95th
1.74(1.31,2.31)
3.11(1.69,5.71)
3.55 (2.59,4.86)
10.87 (3.09,38.27)
7.10 (5.50,9.17)
11.81 (10.29,13.54)
1.19 (0.97,1.46)
2.02 (1.41,2.89)
2.14(1.21,3.76)
3.22 (1.78,5.83)
3.44 (2.73,4.35)
7.16 (5.04,10.18)
1.19(0.76,1.85)
1.40 (0.80,2.46)
2.05 (1.29,3.26)
3.09 (2.19,4.35)
4.19 (2.63,6.67)
6.24 (3.62,10.77)
A-14
-------
Table A-9. Blood MeHg concentrations (ug/L), by frequency of consuming fish, by NHANES survey release, women aged 16-49 years,
NHANES 1999-2010 (continued)
E'vey
sase
2005-2006
2007-2008
2009-2010
Times eaten
in 30 days
0
1
2
3
4-5
6 and up
0
1
2
3
4-5
6 and up
0
1
2
3
4-5
6 and up
N
433
248
224
173
235
479
374
251
190
136
197
345
413
250
213
132
258
520
Arith. Mean
(95% Cl)
0.37 (0.25,0.48)
0.64 (0.50,0.78)
0.82 (0.63,1.01)
1.06 (0.68,1.45)
1.28(0.82,1.73)
1.84 (1.61,2.08)
0.36 (0.25,0.47)
0.69 (0.45,0.92)
0.69 (0.52,0.86)
0.82 (0.57,1.07)
1.05 (0.87,1.23)
1.95 (1.54,2.37)
0.50 (0.40,0.60)
0.58 (0.50,0.67)
0.81 (0.67,0.94)
0.87 (0.70,1.04)
1.27(1.05,1.50)
2.11(1.87,2.35)
Selected percentiles (95% Cl)
25th
0.11 (0.03,0.39)
0.20 (0.10,0.38)
0.26 (0.16,0.41)
0.32 (0.20,0.49)
0.40 (0.26,0.61)
0.64 (0.56,0.75)
0.15 (0.05,0.41)
0.23 (0.13,0.40)
0.27 (0.17,0.43)
0.31 (0.19,0.49)
0.45 (0.34,0.60)
0.64 (0.50,0.82)
0.19 (0.08,0.45)
0.25 (0.16,0.40)
0.33 (0.23,0.46)
0.34 (0.24,0.48)
0.47 (0.37,0.60)
0.77 (0.66,0.91)
50th
0.22 (0.12,0.38)
0.42 (0.32,0.57)
0.50 (0.39,0.64)
0.62 (0.47,0.81)
0.76 (0.62,0.93)
1.20 (1.01,1.44)
0.25 (0.16,0.39)
0.43 (0.31,0.60)
0.46 (0.35,0.60)
0.54 (0.42,0.69)
0.79 (0.62,1.01)
1.24 (0.94,1.62)
0.31 (0.22,0.42)
0.43 (0.34,0.54)
0.55 (0.46,0.67)
0.56 (0.43,0.73)
0.80 (0.61,1.06)
1.36(1.15,1.61)
75th
0.45 (0.33,0.61)
0.80(0.63,1.01)
0.97(0.80,1.17)
1.11(0.64,1.91)
1.45(1.14,1.84)
2.38(2.00,2.83)
0.43 (0.34,0.53)
0.72 (0.49,1.06)
0.72 (0.60,0.88)
0.84(0.66,1.05)
1.32(1.16,1.49)
2.61(2.10,3.24)
0.56 (0.48,0.66)
0.76(0.63,0.93)
1.05 (0.76,1.46)
1.00(0.74,1.33)
1.55 (1.20,2.00)
2.72 (2.49,2.98)
90th
0.77 (0.60,0.98)
1.33(1.00,1.77)
1.61(1.12,2.31)
2.72(1.28,5.79)
2.45(1.16,5.16)
4.07(3.34,4.95)
0.67(0.54,0.83)
1.43 (0.99,2.08)
1.14(0.72,1.82)
1.68(0.93,3.02)
1.83(1.51,2.21)
4.21(2.96,5.99)
1.01(0.73,1.39)
1.19(1.02,1.38)
1.68(1.45,1.93)
2.01(1.36,2.97)
2.60(2.23,3.03)
4.24(3.54,5.08)
95th
1.10 (0.83,1.46)
1.77(1.14,2.76)
2.65 (1.25,5.60)
3.22 (1.85,5.60)
3.86 (1.38,10.76)
5.77 (4.31,7.73)
0.96 (0.72,1.29)
1.92 (1.05,3.50)
1.89(1.21,2.96)
2.51 (1.18,5.34)
2.34 (1.22,4.49)
6.72 (4.49,10.06)
1.41(1.13,1.75)
1.51(1.20,1.91)
2.06 (1.50,2.82)
2.53 (1.96,3.26)
3.15 (2.29,4.35)
6.47 (5.61,7.47)
A-15
-------
Table A-10. Parameter estimates and odds ratios from the logistic model predicting the
probability of reporting any fish consumption in the previous 30 days
Intercept
Age, Overall
Age
Age2
Income, Overall
0 to 20K
20 to 45K
45 to 75K
>75K
MultiHH
Refuse/DK
Over 20K
Race, Overall
Non-Hispanic Black
Mexican Amer.
Other Hispanic
Other Race
Non-Hispanic White
NHANES Survey Release, linear trend
Parameter
1.3809
1.1679
-0.9182
-0.1915
-0.1201
-0.0676
0.3070
-0.0142
0.0683
0.0181
0.2510
-0.0926
-0.0261
0.0533
-0.1856
-0.0184
Std. Error
0.08
0.09
0.36
0.10
0.08
0.09
0.08
0.14
0.23
0.19
0.09
0.07
0.11
0.14
0.06
0.01
p-Value
<0.0001
<0.0001
<0.0001
0.0110
<0.0001
0.05
0.15
0.45
0.0001
0.92
0.77
0.92
0.0001
0.0060
0.20
0.82
0.70
0.0022
0.21
Odds
Ratio
0.83
0.89
0.93
1.36
0.99
1.07
1.02
1.29
0.91
0.97
1.05
0.83
A-16
-------
Table A-ll. Parameter estimates and relative ratios from the model predicting the frequency of
fish consumption in the previous 30 days (times)
Intercept
Age, Overall
Age
Age2
Income, Overall
0 to 20K
20 to 45K
45 to 75K
>75K
MultiHH
Refuse/DK
Over 20K
Race, Overall
Non-Hispanic Black
Mexican Amer.
Other Hispanic
Other Race
Non-Hispanic White
NHANES Survey Release, linear trend
Parameter
1.3581
0.3129
-0.4643
-0.1101
-0.0388
-0.0062
0.1220
0.0169
-0.1535
0.1698
0.0371
-0.2516
-0.1389
0.4057
-0.0524
0.0052
Std. Error
0.03
0.04
0.13
0.04
0.04
0.03
0.04
0.07
0.12
0.07
0.03
0.03
0.05
0.06
0.03
0.01
p-Value
<0.0001
<0.0001
<0.0001
0.0004
<0.0001
0.0026
0.28
0.85
0.0011
0.81
0.19
0.0136
<0.0001
0.22
<0.0001
0.0066
<0.0001
0.07
0.37
Odds
Ratio
0.90
0.96
0.99
1.13
1.02
0.86
1.19
1.04
0.78
0.87
1.50
0.95
A-17
-------
Table A-12. Parameter estimates and relative ratios from the model predicting the amount of
fish consumed in a meal (meal size) (g)
Intercept
Age, Overall
Age
Age2
Income, Overall
0 to 20K
20 to 45K
45 to 75K
>75K
MultiHH
Refuse/DK
Over 20K
Race, Overall
Non-Hispanic Black
Mexican Amer.
Other Hispanic
Other Race
Non-Hispanic White
NHANES Survey Release, linear trend
Parameter
4.0657
0.1365
-0.0338
0.0165
-0.0074
-0.0124
0.0088
0.0085
-0.0381
0.0241
-0.0407
0.1964
-0.0129
-0.0477
-0.0952
-0.0013
Std. Error
0.01
0.01
0.05
0.01
0.01
0.01
0.01
0.03
0.04
0.03
0.01
0.01
0.02
0.02
0.01
0.002
p-Value
<0.0001
<0.0001
<0.0001
0.48
0.39
0.23
0.49
0.31
0.52
0.74
0.36
0.35
<0.0001
0.0003
<0.0001
0.51
0.0022
<0.0001
0.46
Odds
Ratio
1.02
0.99
0.99
1.01
1.01
0.96
1.02
0.96
1.22
0.99
0.95
0.91
A-18
-------
Table A-13. Parameter estimates and relative ratios from the model predicting the mercury
concentration of the fish consumed (|jg)
Intercept
Age, Overall
Age
Age2
Income, Overall
0 to 20K
20 to 45K
45 to 75K
>75K
MultiHH
Refuse/DK
Over 20K
Race, Overall
Non-Hispanic Black
Mexican Amer.
Other Hispanic
Other Race
Non-Hispanic White
NHANES Survey Release, linear trend
Parameter
-2.6518
0.1574
-0.2172
-0.0047
-0.0183
-0.0002
0.0409
-0.0259
-0.1409
0.1491
-0.0015
-0.1032
-0.0441
0.0560
0.0928
-0.0110
Std. Error
0.03
0.04
0.13
0.04
0.03
0.04
0.04
0.07
0.12
0.07
0.03
0.03
0.04
0.04
0.02
0.01
p-Value
<0.0001
0.0006
0.0003
0.09
0.21
0.90
0.58
1.00
0.28
0.71
0.25
0.03
<0.0001
0.95
0.0002
0.31
0.17
<0.0001
0.0352
Odds
Ratio
1.00
0.98
1.00
1.04
0.97
0.87
1.16
1.00
0.90
0.96
1.06
1.10
A-19
-------
Table A-14. Parameter estimates and relative ratios from the model predicting the inverse of
body weight (I/kg)
Intercept
Age, Overall
Age
Age2
Income, Overall
0 to 20K
20 to 45K
45 to 75K
>75K
MultiHH
Refuse/DK
Over 20K
Race, Overall
Non-Hispanic Black
Mexican Amer.
Other Hispanic
Other Race
Non-Hispanic White
NHANES Survey Release, linear trend
Parameter
0.0950
-0.1655
0.1410
-0.0437
-0.0177
-0.0084
0.0378
0.0142
0.0271
-0.0093
-0.1309
0.0081
0.0232
0.1148
-0.0152
-0.0015
Std. Error
0.01
0.01
0.04
0.01
0.01
0.01
0.01
0.02
0.03
0.02
0.01
0.01
0.01
0.01
0.01
0.001
p-Value
<0.0001
<0.0001
<0.0001
0.0004
<0.0001
<0.0001
0.03
0.42
0.0002
0.41
0.29
0.64
<0.0001
<0.0001
0.21
0.03
<0.0001
0.02
0.20
Odds
Ratio
0.96
0.98
0.99
1.04
1.01
1.03
0.99
0.88
1.01
1.02
1.12
0.98
A-20
-------
Table A-15. Parameter estimates and relative ratios from the model predicting mercury intake
per unit body weight (|jg/kg)
Intercept
Age, Overall
Age
Age2
Income, Overall
0 to 20K
20 to 45K
45 to 75K
>75K
MultiHH
Refuse/DK
Over 20K
Race, Overall
Non-Hispanic Black
Mexican Amer.
Other Hispanic
Other Race
Non-Hispanic White
NHANES Survey Release, linear trend
Parameter
-1.4637
0.4413
-0.5742
-0.1420
-0.0821
-0.0273
0.2095
0.0137
-0.3055
0.3337
-0.1361
-0.1502
-0.1726
0.5289
-0.0699
-0.0086
Std. Error
0.06
0.07
0.23
0.07
0.06
0.06
0.07
0.13
0.24
0.10
0.05
0.05
0.09
0.09
0.04
0.01
p-Value
<0.0001
<0.0001
<0.0001
0.0141
<0.0001
0.04
0.18
0.67
0.0032
0.92
0.21
0.0016
<0.0001
0.0069
0.0065
0.06
<0.0001
0.09
0.35
Odds
Ratio
0.87
0.92
0.97
1.23
1.01
0.74
1.40
0.87
0.86
0.84
1.70
0.93
A-21
------- |