Journal of Toxicology and Environmental Health, Part A, 66:817–828, 2003
Copyright© 2003 Taylor & Francis
1528-7394/03 $12.00 + .00
EFFECT OF DEEP-FRYING FISH ON RISK FROM MERCURY
Joanna Burger, Carline Dixon, C. Shane Boring
Division of Life Sciences, Consortium for Risk Evaluation with Stakeholder
Participation, and Occupational and Environmental Health Sciences Institute,
Piscataway, New Jersey, USA
Environmental and Community Medicine, Consortium for Risk Evaluation
with Stakeholder Participation, UMDNJ–Robert Wood Johnson Medical
School, Piscataway, NJ, 08854
The U.S. Environmental Protection Agency (EPA) and many states have issued advisories to limit
or avoid the consumption of certain fish or fish from certain waters, particularly by pregnant
women and young children or even women of childbearing age. Typically, risk is calculated by
multiplying contaminant concentrations in fish tissue, frequency of meals, and meal size,
compared to some criterion, usually the U.S. EPA reference dose (RfD). Site-specific data on
mercury concentrations, meal size, and consumption frequency by fishermen were used to
determine how frying fish affected risk estimates. In consumption studies fishermen typically
estimate the size of portions as they appear on the plate (i.e., cooked), yet assessors calculate
risk based on contaminant levels in uncooked fish. Largemouth bass (Micropterus salmoides,
n = 39) were collected from the contaminated L Lake on the Savannah River Site, South Carolina.
Fillets from the opposite sides of the same fish were divided and randomly assigned to a raw or
fried treatment (the commonly used local cooking method). The fried fillet was further divided
in half for a breaded or nonbreaded treatment. Mercury averaged 0.44µg/g (ppm, wet weight)
in raw fish, 0.63µg/g in fried and breaded fish, and 0.76µg/g in fried, unbreaded fish. The
maximum concentration was 1.5µ g/g in raw fish (1.9µ g/g in cooked fish). Deep-frying with
and without breading resulted in weight loss of 25% and 39%, while mercury levels increased
by 45% and 75%, perhaps due to the breading and absorption of oil. At the mean fish
Received 12 August 2002; sent for revision 18 September 2002; accepted 22 October 2002.
We thank C. Jeitner and S. Shukla for laboratory and computer assistance; T. Shukla and H. Jensen for
analytical assistance; K. F. Gaines, W. L. Stephens, Jr., J. Snodgrass, I. L. Brisbin, Jr., and J. W. Gibbons for
logistical support at SRS; and B. D. Goldstein, B. Friedlander, J. Nelsen, C. Powers, A. Upton, C. Warren,
W. Whitaker, and anonymous referees for valuable comments on the research and manuscript. This
research was also influenced by discussions with members of the SRS Citizen’s Advisory Board, the Center
for Disease Control’s SRS Health Effects Subcommittee, South Carolina Department of Health and Environ-
mental Control, Georgia Department of Natural Resources, and the U.S. Environmental Protection Agency.
This research was funded by the Consortium for Risk Evaluation with Stakeholder Participation (CRESP)
through the Department of Energy (AI numbers DE-FC01-95EW55084, DE-FG 26-00NT 40938), NIEHS
(ESO 5022, J. Burger, M. Gochfeld), and Financial Assistance Award DE-FC09-96SR18546 from the U.S.
Department of Energy to the University of Georgia Research Foundation (S. Boring). The results, conclusions,
and interpretations reported herein are the sole responsibility of the authors, and should not in any way be
interpreted as representing the views of the funding agencies.
Address correspondence to Joanna Burger, Division of Life Sciences, Nelson Biology Lab., Rutgers
University, 604 Allison Street, Piscataway, NJ 08854-8082, USA. E-mail: email@example.com
J. BURGER ET AL.
consumption rate of people fishing locally, mercury intake exceeded the U.S. EPA RfD of
0.1 µg/kg/d for all except white females. Thus consumption of fish from this lake would exceed
acceptable levels. Risk assessments should be conducted with site-specific data on contamin-
ants and consumption of cooked fish and consumption studies should specify whether
portion size was pre- or postpreparation. Fishermen estimate the amounts of fish they eat
based on a meal size (usually cooked), while risk assessors determine mercury levels in raw fish.
A conversion factor of about 2 for mercury increase during cooking is reasonable and conser-
In many rural and some urban regions of the United States fishing is
an important aspect of recreation, culture, and tradition (Fleming et al., 1995;
Toth & Brown, 1997; Burger, Stephens et al., 1999; Burger, Pflugh et al., 1999).
Yet contaminants, such as mercury and polychlorinated biphenyls (PCBs), are
sufficiently high in some fish and seafood to pose a risk to human consumers
(ATSDR, 1996; IOM, 1996; Ratcliffe et al., 1996; Weiss & Elsner, 1996; Weihe
et al., 1996; Kamrin & Fischer, 1999; Sweet & Zelikoff, 2001). Concern about
such health risks has led governmental agencies to issue consumption adviso-
ries for many waterbodies (U.S. EPA, 2002). The U.S. Environmental Protec-
tion Agency (1998) reported that 16% of the nation’s total lake acres and 7%
of the nation’s total river miles were under consumption advisories, as well as
all of the Great Lakes. Mercury accounts for most of the advisories, but PCBs,
chlordane, dioxins, and DDT are also important (U.S. EPA, 2002).
Fetuses, neonates, and young children are the group considered at highest
risk (Bakir et al., 1973; Jacobson et al., 1990; Davidson et al., 1995; Jacobson
& Jacobson, 1996; Weihe et al., 1996; Evens et al., 2001). Risk reduction for
fetuses and neonates involves clear and balanced risk communication to preg-
nant women, taking into account the fact that fish are a wholesome source of
nutrition during pregnancy (Sparks & Shepherd, 1994; Knuth, 1995; Ebert,
1996). Recognizing the preconception exposure contributes to maternal body
burden, agencies often expand their warnings to all women of childbearing
age. Risk from fish consumption involves contaminant levels in edible fish
tissue, meal frequency, and meal size. Yet other factors may affect the risk
from fish consumption, including preparation practices (Morgan et al., 1997;
Wilson et al., 1998).
This article examines the effect of deep-frying on mercury levels in large-
mouth bass (Micropterus salmoides), a preferred fish of people in South
Carolina and elsewhere in the south (Fleming et al., 1995; Toth & Brown,
1997). Over 80% of the people interviewed along the Savannah River deep-fry
their fish regularly (Burger, Stephens et al., 1999), making it critical to understand
whether this cooking method affects mercury levels in fish. Most of the fish-
frying is done with batter and breading, and the effect of breading on mercury
levels was also examined. Thus, the hypothesis was tested that the levels of
mercury in raw and deep-fried fish (with and without breading) do not differ. If
the levels of mercury differ in cooked and raw fish, this can have major impli-
cations for risk assessment because most consumption studies report consumption
EFFECT OF DEEP-FRYING FISH ON RISK FROM MERCURY
on the basis of cooked fish (Fleming et al., 1995; Ratcliffe et al., 1996; Burger,
Stephens et al., 1999; Kamrin & Fischer, 1999), yet contaminant data are
reported for raw fish (Knuth, 1995; Morgan et al., 1997; Burger, Gaines &
Gochfield, 2001; Burger, Gaines, Peles et al., 2001).
The overall objective was to determine whether deep-frying affects the
resultant risk assessment for fish consumption, including pregnant women.
Site-specific information on consumption patterns (meal frequency, meal size),
cooking methods (deep-frying), and contaminant levels in largemouth bass,
a popular, predatory fish specifically mentioned in consumption advisories for
the Savannah River, was used (SCDHEC, 1996, 2001; GDNR, 2001). Second,
the potential risk to consumers from eating these fish was examined, considering
the possibility that recreational fishing may someday be allowed in a currently
Several studies have examined the effect of fat trimming, skin removal, and
cooking on PCBs and other fat-soluble contaminants in fish (Morgan et al.,
1997), but much less attention has been devoted to mercury (Wilson et al.,
1998). Trimming and cooking fish by various means reduces the levels of
fat-soluble contaminants, including dioxins, PCBs, and organochlorine pesticides
(Reinert et al., 1972; Sanders & Haynes, 1988; Morgan et al., 1997). This has
led some state agencies to use a reduction factor in calculating the dose of
fat-soluble contaminants from raw fish in developing fish advisories (Wilson
et al., 1998).
Information about the effects of cooking on mercury is less clear. Two
studies showed no statistical difference in mercury in cooked and raw fish
(Armbruster et al., 1987; Gutenmann & Lisk, 1991), while another showed
higher levels in cooked fish on a wet weight basis (Morgan et al., 1997). Morgan
et al. (1997) examined the effect of cooking practices on mercury for two
commonly caught fish from Lake Wisconsin (walleye, lake trout). They found
that mercury concentrations in pan-fried, baked, and broiled fish fillets were
from 1.1 to 1.6 times higher than in corresponding raw portions. They found that
total mercury amounts were constant before and after cooking (i.e., mercury was
not driven off during cooking). However, they did not deep-fry fillets, and data
on walleye and trout may not be applicable to the fish commonly caught in
much of the eastern and southeastern United States. Further, without site-
specific consumption data, they did not examine the effect on risk assessment
and risk management.
Fish were collected from L Lake on the Department of Energy’s Savannah
River Site (SRS) (33.1 °N, 81.3 °W). SRS is a 780-km2 former nuclear weapons
production and research facility operated by the U.S. government since the
J. BURGER ET AL.
Fish Collection and Cooking
Thirty-nine largemouth bass were collected under appropriate state permits,
and with protocol approvals from the University of Georgia Institutional Animal
Care and Use Committee (A960205) and Rutgers University Institutional
Review Board (07-017). Fish were collected using rod and reel, and taken to
the Savannah River Laboratory for dissection into fillets. In the laboratory, after
skinning, a 2-cm-thick fillet (200 g) was removed from each side of the fish.
One fillet was randomly assigned to raw (only half was used in the analysis)
and one was assigned to the fry treatment. The 2-cm-thick fillet assigned to fry
was further divided into half; one to receive breading and one to be without
breading. The 100-g fillets were immediately frozen (?4 °C) and labeled by fish
number, date, collection location, and treatment.
For breading, the wet fillet was dipped in a commercial breading mix
(Zatarain’s Seasons Fish Fry) until the surface was lightly covered. Cooking was
done according to local customs using local deep frying vats (uncovered), and
with a commonly used cooking oil (Dukes peanut oil) as is the local custom. No
mercury was detectable in the cooking oil when analyzed alone. Fillets were
weighed before and after cooking. Fillets were submerged in hot oil (approxi-
mately 190 °C) until the fillet began to float (typically 5–7 min), indicating that
cooking was complete. Fillets were then patted dry of oil, weighed, and frozen for
transportation to the Environmental and Occupational Health Sciences Institute.
Chemical and Statistical Analysis
A 2-g sample was cut from the center of each 100-g fillet and completely
digested in 3 ml Ultrex ultrapure nitric acid in a microwave (MD 2000 CEM),
using a digestion protocol of 3 stages of 10 min each under 50, 100, and 150
pounds per square inch (3.5, 7.0, and 10.6 kg/cm2) at 70 × power (Burger,
Gaines & Gochfeld, 2001). Digested samples were diluted in 20 ml deionized
water. Mercury was analyzed by cold vapor technique. All concentrations are
expressed in parts per million (µg/g on wet weight).
Mercury was analyzed with a HGA4 mercury analyzer in the Environmental
and Occupational Health Sciences Institute. Detection limits were 0.2 ppb for
mercury (instrument detection limits). All specimens were run in batches that
included blanks, a standard calibration curve, and spiked specimens. The
accepted recovery levels for spikes was 85%. All spikes exceeded this value.
The coefficient of variation on replicate samples ranged from 4 to 10%.
Further quality control included blind runs of duplicate samples during the
analysis for each metal (acceptable criterion ±15%). Twelve replicates were
also analyzed using the pyrolysis feature of the Lumex mercury analyzer and
yielded a high correlation (r = .94) between the two methods. Like most
analytic reports, these procedures analyze the total mercury content, while risk
assessments are based on methylmercury. Other studies have shown that in
most cases at least 90% of the mercury in fish is in the methyl form. Thus this
adds a slight additional protective value of up to 10% to the risk assessment,
since not all the measured mercury is methyl.
EFFECT OF DEEP-FRYING FISH ON RISK FROM MERCURY
Mercury concentrations were compared among treatments using the
Kruskal-Wallis nonparametric analysis of variance (SAS PROC NPAR1WAY
with Wilcoxon Option), which generated a ?2 value to examine differences
among treatments (SAS Institute, Inc., 1995). Before and after mercury con-
centrations for the 39 fish were compared with a Kendall tau. An a priori
significance level of <.05 was designated.
Risk Analysis Methods
Development of the nervous system in the fetus is the most sensitive end
point for organic mercury (Stern, 1993; Young et al., 1997) and is now used in
risk assessments to develop RfDs or their equivalent. The RfD is based on
chronic exposure, no observed effect levels, and various uncertainty factors,
including one to protect sensitive subgroups, hence it can apply to either adults
or children. The development of the U.S. Environmental Protection Agency
RfDs was originally based on the Iraq organomercury epidemic in the early
1970s (Bakir et al., 1973), although Stern (1993) computed a lower RfD of
0.07 µg/kg/body weight/d based on data from Iraq. Other suggested RfDs have
been based on recent prospective longitudinal studies in the Seychelles (Bakir
et al., 1973) and the Faroe Islands (Weihe et al., 1996). The National Research
Council recently reviewed the studies and suggested that the Faroe Island
results were a reasonable basis for a revised RfD, and the U.S. EPA recalcu-
lated the RfD, arriving again at a value of 0.1 µg/kg/d, which is documented in
the U.S. EPA IRIS database (U.S. EPA, 1997, 1998). However, the U.S. EPA
Division of Water has based its fish advisories on an oral RfD of 0.06 µg/kg/d.
Recently, the Agency for Toxic Substances and Disease Registry has proposed
a minimum risk level of 0.3 µg/kg/d (ATSDR, 1999) based on the Seychelles
neurodevelopmental study (Davidson et al., 1995).
The risk assessment in this study is based on the U.S. EPA oral RfD for
methylmercury of 0.1 µg/kg/d and on the minimum risk level (MRL) of 0.3 µg/
kg/d, which is sometimes used as an unofficial RfD for adults who are not
pregnant. The estimated daily intake of mercury, derived from consumption
studies (Burger, Stephens et al., 1999), was compared with the RfD values, using
the median, mean, 75%, and 95% of the consumption distributions reported
for males and females. Since there was a clear ethnic differences in consump-
tion rates among fishermen in the area (Burger, Gaines & Gochfeld, 2001),
black and white fishermen are reported separately (Table 1).
Mercury in Raw and Cooked Fish
Mercury levels were significantly higher in cooked fish (breaded and non-
breaded) compared to raw fish (Kruskal-Wallis ?2= 13.8, Table 2). For a given
fish, there was a significant correlation between the levels of mercury in the
cooked and uncooked fish (tau = 0.85, p < .001). On average, an uncooked
J. BURGER ET AL.
TABLE 1. Fish Consumption
Black males (g/d)
Black females (g/d)
White males (g/d)
White females (g/d)
Black males (kg/yr)
Black females (kg/yr)
White males (kg/yr)
White females (kg/yr)
Note: Data are based on interviews with fishermen along the Savannah River (Burger, Stephens et al., 1999).
TABLE 2. Concentrations of Total Mercury (in µg/g [ppm], wet weight) in Raw and Fried Fillets from
Largemouth Bass from L Lake
of sample (g)
Arithmetic mean ± SE
Range of Hg
0.446 ± 0.04 0.16–1.56
0.63 ± 0.065 0.16–1.97
0.76 ± 0.069 0.16–1.56
Note: Data are based on n = 39.
fillet of 100 g weighed 80 g after cooking with breading, and 72 g after cooking
without breading. While deep-frying with and without breading resulted in
weight loss of 25% and 39%, mercury levels increased by 45% (breaded) and
Based on site-specific data on fish consumption and mercury concentra-
tions, risk was calculated for four subgroups (black males, black females, white
males, white females) at four levels of intake (median, mean, 75th and 95th
percentile) for raw and cooked fish, using the mean mercury concentrations
under each condition (Table 3). Mercury intake for those with mean consump-
tion rates eating nonbreaded fish at the mean mercury concentration of 0.76ppm
ranged from 0.19 µg/kg/d for white females to 0.44 µg/kg/d for black males
(Table 3). At the median consumption rate, only white females (mean intake of
0.09 µg/kg/d) would not exceed the U.S. EPA reference dose.
Another method of examining risk is to calculate how much fish a preg-
nant women could consume without increasing her risk. Translating this into
consumption patterns, a pregnant women could not eat an 8-ounce (227 g)
portion of cooked fish more often than about once a month without exceeding
the U.S. EPA RfD (Table 4).
EFFECT OF DEEP-FRYING FISH ON RISK FROM MERCURY
TABLE 3. Mercury Intake for Four Subgroups at Four Levels of Fish Consumption
Deep-fried and breaded
Deep-fried, not breaded
Note: Data are based on four demographic groups, for the mean mercury concentration in each of the
cooking categories (Burger, Stephens et al., 1999). Body weight of 70 kg for males and 60 kg for females
was assumed. All values are in micrograms of mercury per kilogram body weight per day (µg/kg/d).
TABLE 4. Frequency with Which a 60-kg Woman Could Consume an 8-Ounce Meal of Each Type of Fish
(Raw, Fried) and Corresponding Advisory Text
Advice for Advice
µg Hg in 8-ounce Dose/meal or Dose/RfD Dose/ pregnant for other
(µg/g wet)a (227-g) meal
intake/60 kgb (HQ)c
1 mo 1 wk
1 mo 2 wk
Note: Doses are divided by the U.S. EPA RfD (0.1 µg/kg/d) and by the CDC/ATSDR MRL (0.3 µg/kg/d).
aConcentration of mercury in muscle in ppm (wet weight basis).
bMercury in 8-ounce (227-g) meal divided by 60 kg body weight.
cHazard quotient (HQ) is obtained by dividing daily dose by RfD.
dText for fish advisory: Pregnant women should not eat an 8-ounce meal more often than every (week,
eText for other adults: Do not eat this more often than once every week
The source of the mercury in the fish was primarily industrial pollution.
L Lake was used as a source of cooling water for one of the nuclear reactors
when it was functioning (Kennamer et al., 1998). Prior to the construction of
J. BURGER ET AL.
the cooling ponds, there was some ecosystem contamination of streams and
the floodplain, and small quantities of radionuclides were released subsequently
(Ashley & Zeigler, 1980; Whicker et al., 1990; Kennamer et al., 1998). There
is some controversy about the source of mercury contamination in L Lake,
which appears to come both from on-site releases and from mercury released
upriver, which was then pumped into the lake during cooling activities and re-
released to contribute to the mercury load in the Savannah River (Kvartek
et al., 1994; Sugg et al., 1995). Atmospheric deposition also contributes con-
taminants to SRS. The U.S. EPA has determined that the Savannah River is in a
zone of above average atmospheric mercury deposition (>10 µg/m2/yr; U.S.
Mercury in Raw and Cooked Fish
Heavy metal levels in fish are usually calculated on a wet weight basis,
rather than a dry weight basis. For 11 species of Savannah River fish the dry
weight ranged from 23 to 33% of the corresponding wet weight (i.e., water
content of 67–77%). Thus for the same samples, concentrations expressed on
a wet weight basis are one-fourth to one-third of the same mercury content
expressed on a dry weight basis, although in some fish the ratio may be as high
as one-fifth (Burger, Gaines, Boring et al., 2001). While drying fish for dry
weight analysis results in complete moisture loss, cooking for human consump-
tion removes only some of the moisture.
Deep-frying of raw fish resulted in moisture losses. However, the fillets
deep-fried with breading weighed more than those without breading, no
doubt due to the weight of the breading itself. When fish is deep-fried, it loses
moisture, but gains weight from the breading and from the uptake of oil.
There was an apparent increase in mercury concentration in the deep-fried
fish (largely because when calculated on the basis of the weight of the sample,
the fried fish had lost moisture but retained the same amount of mercury). The
end result of moisture loss in fish on contaminant levels is that levels are higher
in cooked fish than fresh fish (on a wet weight basis of the portion itself). If
drying takes a 100-g fillet to 20 g, but retains the same level of mercury, then
the concentration of mercury will be five times as high in the dried fish as the
wet weight fish, even though the actual amount has not changed. For
example, if deep-frying results in a water loss of 50% (the weight of a fillet goes
from 100 to 50 g), the concentration of mercury in the average fillet would
increase from 0.45 µg/g to 0.9 µg/g while the mass would remain the same
(45 µg). If, during the process of deep frying, the fillet absorbed 9 g cooking oil,
the fillet would weigh 59 g, and the concentration of mercury would be
0.76 µg/g—the average for a fried fillet in this study.
Ecologists interested in mercury levels in fish regularly use dry weight/wet
weight conversion factors of 4 to 5 when determining the toxic dose for the
fish themselves. While the conversion factor between raw and cooked fish is
not as high as between laboratory dried and raw fish, it still is a factor that
should be considered.
EFFECT OF DEEP-FRYING FISH ON RISK FROM MERCURY
In largemouth bass, mercury levels were 45–75% higher in the cooked
fillets compared to uncooked fillets (on a wet weight basis), which is slightly
higher than the range reported by Morgan et al. (1997) for fish from Lake
Superior. In some of the samples, mercury concentrations were twice as high
in cooked as in uncooked fish.
Risk Assessment: Fish from L Lake
The mercury in the fish from L Lake, averaging 436 ± 43 ppb in raw fish,
can be used to compute the risk to fishermen if they ate these fish. While at
present fishing is not allowed on L Lake, the possibility exists for illegal fishing,
and future land use scenarios for SRS include recreational fishing. Fishermen
often catch more than a single fish and even if they fish infrequently, may take
home enough fish for several meals (Burger, Stephens et al., 1999). The data
from this study clearly indicate that at this time, there would be a risk to
people consuming largemouth bass from L Lake more frequently than once a
month. From a risk management perspective, it means that DOE officials
should be diligent about preventing fishing on this site, and aware of the risk in
any future land use decisions, until mercury levels in fish decline substantially.
Risk Assessment: Raw Versus Cooked Fish
The primary objective was to understand the effect of local cooking
methods on mercury in largemouth bass, one of the preferred local fishes
(Burger, 1998). The data from this study clearly indicate that the concentration
of mercury (on a wet weight basis) in uncooked fish is less than in cooked fish
(for the same portion size). The actual conversion factor will depend upon the
species of fish and the cooking method. While it is clear from these data that
a conversion factor should be used in risk assessment, there are several ques-
tions that need to be answered for adequate risk assessment and management,
including: (1) Are estimates of meal size or amount eaten based on cooked or
uncooked fish? (2) Are contaminant values based on cooked or uncooked fish?
(3) What are the conversion values for a specific fish and a specific cooking
method? (4) What are the relative contributions of the different fish species and
cooking methods to the total fish consumption of people eating wild-caught fish?
Morgan et al. (1997) suggested using food preparation factors in risk
assessment, but these have generally not been applied. Preparation factors
(mercury concentration in cooked fish/mercury concentration in raw fish) in
their study generally ranged from 1.3 to 1.6 for fillets from Great Lakes fish,
compared to 1.5 to 1.8 for largemouth bass in this study. These two studies
suggest that a preparation conversion factor of 2 would be a suitable, protective
default. Thus, risk assessors who do not take into account cooking method, but
use contaminants data from raw fish, may be overestimating safe consumption
levels. This factor should be considered by state agencies setting consumption
levels for high risk populations.
The risk assessment presented in this paper for largemouth bass from the
Savannah River Site indicates that at the mean, median, and 75th and 95th
J. BURGER ET AL.
percentile consumption levels, black and white men and women exceed the
U.S. EPA reference dose. Translated into consumption levels, it means that
people should not eat an 8-ounce portion of raw bass more than every 17 d,
or an 8-ounce portion of cooked fish every 24–29 d. The average meal size of
South Carolina fishermen interviewed was actually close to 10 ounces per
meal (Burger, Stephens et al., 1999). In Table 4 a suggested interval for fish
consumption is provided. If risk assessors are basing their risk estimates on
mercury in raw fish (the usual practice) and consumers estimate their meal size
from cooked fish (the usual practice), then the risk estimates are biased down-
ward. This is counterbalanced by the fact that the RfD is based on methylmercury,
while the analytic results are based on total mercury, and the risk calculations
assume 100% is methylmercury. Since methylmercury actually comprises
about 90–95% of the total mercury in fish, this adds an additional 5–10%
Calculations of the RfD and other criteria values incorporate one or more
safety or uncertainly factors to assure protection of sensitive individuals. Thus,
what appears unacceptably risky to a public health agency does not necessarily
mean that an individual’s health will be measurably impacted.
Finally, it is clear that the mass of mercury in the fillet itself has not
changed; what has changed between raw and cooked fish is the perception of
the quantity of fish consumed. The human health risk is based on the dose
(mass of mercury). However, if people estimate their consumption based on
cooked fish but risk assessors compute risk on raw fish, the estimates are
underestimates of the actual risk. Another implication of this research is that
when people are asked about fish consumption, whether their answers are
based on cooked or uncooked fish should be clearly stated. Unlike fishermen,
people who purchase fish in stores may have weighed fillets and may estimate
their intake on the basis of precooked weights.
Agency for Toxic Substances and Disease Registry. 1996. States issue a record number of health advisories.
Hazard. Subst. Public Health 6:1–2.
Agency for Toxic Substances and Disease Registry. 1999. Toxicological profile for mercury—Update. Atlanta,
GA: Centers for Disease Control.
Armbruster, G., Gerow, K. G., Gutenmann, W. H., Littman, C. B., and Lisk, D. J. 1987. The effect of several
methods of fish preparation on residues of polychlorinated biphenyls and sensory characteristics of
striped bass. J. Food Safety 8:235–243.
Ashley, C., and Zeigler, C. C. 1980. Releases of radioactivity at the Savannah River Plant, 1954 through
1978. E. I. DuPont de Nemours and Company Report DPSPU 75-25-1, Aiken, SC.
Bakir, F., Damluji, S. F., Amin-Zaki, L., Murtadha, M., Khalidi, A., al-Rawi, N. Y., Tikriti, S., Dahahir, H. I.,
Clarkson, T. W., Smith, J. C., and Doherty, R. A. 1973. Methylmercury poisoning in Iraq. Science
Burger, J. 1998. Fishing and risk along the Savannah River: Possible intervention. J. Toxicol. Environ. Health
Burger, J., Stephens, W., Boring, C. S., Kuklinski, M., Gibbons, J. W., and Gochfeld, M. 1999. Factors in
exposure assessment: Ethnic and socioeconomic differences in fishing and consumption of fish caught
along the Savannah River. Risk Anal. 19:421–431.