Arch Environ Contam Toxicol DOI 10.1007/s00244-014-0102-y

Mercury Exposure as a Function of Fish Consumption in Two Asian Communities in Coastal Virginia, USA Xiaoyu Xu • Michael C. Newman

Received: 15 July 2014 / Accepted: 27 October 2014 Ó Springer Science+Business Media New York 2014

Abstract Fish consumption and associated mercury exposure were explored for two Asian-dominated church communities in coastal Virginia and compared with that of two non-Asian church communities. Seafood-consumption rates for the Chinese (36.9 g/person/day) and Vietnamese (52.7 g/person/day) church communities were greater than the general United States fish-consumption rate (12.8 g/ person/day). Correspondingly, hair mercury concentrations for people from the Chinese (0.52 lg/g) and the Vietnamese church (1.46 lg/g) were greater than the overall level for United States women (0.20 lg/g) but lower than the published World Health Organization exposure threshold (14 lg/g). A conventional regression model indicated a positive relationship between seafood consumption rates and hair mercury concentrations suggesting the importance of mercury exposure through seafood consumption. The annual-average daily methylmercury intake rate for the studied communities calculated by Monte Carlo simulations followed the sequence: Vietnamese community [ Chinese community [ non-Asian communities. Regardless, their daily methylmercury intake rates were all lower than the United States Environmental Protection Agency reference dose of 0.1 lg/kg body weight-day. In conclusion, fish-consumption patterns differed among communities, which resulted in different levels of mercury exposure. The greater seafood and mercury ingestion rates of studied Asian groups compared Electronic supplementary material The online version of this article (doi:10.1007/s00244-014-0102-y) contains supplementary material, which is available to authorized users. X. Xu (&)  M. C. Newman Virginia Institute of Marine Science, College of William & Mary, P.O. Box 1346, Gloucester Point, VA 23062, USA e-mail: [email protected]

with non-Asian groups suggest the need for specific seafood consumption advice for ethnic communities in the United States. Otherwise the health benefits from fish consumption could be perceived as trivial compared with the ill-defined risk of mercury exposure.

Concern about environmental mercury continues to grow due to its increasing anthropogenic emissions (Lamborg et al. 2002; Manohar et al. 2002), wide atmospheric dispersion (Mason et al. 1994), propensity to biomagnify after methylation (Compeau and Bartha 1985; Newman et al. 2011a), and high toxicity (Ginsberg and Toal 2009; Oken et al. 2005). Methylmercury, the form of most concern that readily transfers through food webs, is a well-established neurotoxicant (Guallar et al. 2002) and a risk factor for cardiovascular disease (Oken et al. 2005). During the past 50 years, increased mercury concentrations have been observed in many fish species, including some exceeding human toxicological thresholds [United States Food and Drug Administration (USFDA) 2013; United States Environmental Protection Agency (USEPA) 2004, 2009]. The accumulation of mercury in finfish and shellfish species provides an exposure pathway to humans, thus creating concerns about public health and increasing discussions on the safety of seafood consumption. Although there is general agreement about methylmercury toxicity, key features of present day fish-consumption advisories still remain incompletely defined (Ginsberg and Toal 2009). A common misconception emerged in many public sectors is that pervasive and harmful levels of mercury exist in seafood that causes serious health consequences. This leads to low consumption of fish that are rich in nutrients of high-quality protein, vitamins, and beneficial omega-3 polyunsaturated fatty acids (Helland et al. 2003;

123

Arch Environ Contam Toxicol

Oken et al. 2005, 2008; De Backer et al. 2003; Priori et al. 2003; Smith et al. 2006). Therefore, sound dietary decisions require a clear understanding by the public of both the risks and benefits. In addition, informing consumers (market, recreational, or tourism-related consumption) about the actual exposure levels could potentially remove people’s confusion and misperception about fish safety. The Chesapeake Bay is a large and biologically diverse estuary that yields a variety of seafood every year. Available fish-consumption advisories focused on mercury contamination for lower Chesapeake Bay (coastal Virginia) residents are provided by the Virginia Department of Health (VDH 2013) and the West Virginia Department of Health and Human Resources (DHHR 2014) including meal size and frequency for specific species from different waters based on metal and organic contamination scenarios. However, these advisories were based on the general public consumption rates, which can be very different from those of some specific subpopulations. Therefore, this general information is insufficient to adequately inform all lower Chesapeake Bay residents of their mercury risks from seafood consumption. For instance, women from a recently surveyed AfricanAmerican community in Newport News (Virginia, USA) showed a substantially greater seafood consumption rate of 147.8 g/day compared with the reported 1.8 g/day of United States women due to their distinct seafood consumption habits (Holloman and Newman 2010). In addition, people from ethnic groups tend to have distinctive seafood consumption patterns as a result of differences in choice of consumed species, meal size, consumption frequency, and parts of finfish consumed (Sechena et al. 2003). Such habitual and cultural differences will change mercury exposure from seafood consumption so that the generic consumption advisories (USEPA 2011; VDH 2013; DHHR 2014) would be misleading for members of such communities. So far, insufficient consumption advice is available for ethnic groups in coastal Virginia due to limited exploration of seafood consumption and mercury exposure. In this study, mercury exposure through seafood consumption for two Asian communities with notionally distinct consumption habits was studied. The purpose was to generate a better understanding of mercury exposure for such ethnic groups, formulate specific seafood consumption advice if needed, and explore the relationship between seafood consumption and mercury exposure. Two questions were proposed here: (1) How do dietary customs influence people’s mercury exposure for the recently immigrated Asians in this area? (2) If their mercury exposure is greater than that of the general population, is it high enough to require communication of additional advisory information?

123

Materials and Methods The Studied Communities Preliminary data of hair mercury in Chinese and Vietnamese communities along coastal Virginia were collected in 2008–2011. Mean mercury concentrations (0.30 lg/g for the Chinese community and 0.26 lg/g the Vietnamese community) were both [0.2 lg/g (McDowell et al. 2004), the mean of United States women age from 16 to 49 reported by National Health and Nutrition Examination Study (NHANES). These results suggested that mercury exposure levels in these communities were increased relative to the general United States population. Based on the preliminary findings, we performed follow-up studies of the typical Chinese and Vietnamese communities in coastal Virginia. The largest Chinese church [Peninsula Chinese Baptist Church, Yorktown, Virginia, USA (GPS 37.1386, -76.4562)], Vietnamese church [Our Lady of Vietnam Chapel, Hampton, Virginia, USA (GPS 37.0304, -76.3634)], and two reference (non-Asian) churches that served the general population [Unity Fellowship Church, Newport News, Virginia, USA (GPS 37.0921, -76.4736) and Gloucester Point Baptist Church, Gloucester Point, Virginia, USA (GPS 37.2561, -76.4923)] were selected. Church-based sampling through engagement of church leaders was adopted to provide access to a large number of individuals with common ethnic backgrounds. This approach is particularly appropriate for sampling ethnic groups in the study area where families might live in several distant neighborhoods.

Survey Design The Lower Chesapeake Bay Seafood Consumption Survey included a food frequency questionnaire to characterize seafood consumption patterns and hair sample collection for direct mercury exposure assessment. The questionnaire was drafted by modification of the Southeast Seafood Consumption Survey (Holloman and Newman 2010) and the Asian and Pacific Islander Seafood Consumption Study (Sechena et al. 2003). Information about seafood consumption patterns (such as species, consumption frequency, meal size, and consumed finfish parts), hair analysis, and demographic background (age, sex, and ethnic group) were solicited using both English and Mandarin or Vietnamese languages. Some questions about seafood consumption patterns (consumed items, consumption frequency, and meal size) were asked in two ways so that precision of answers could be checked. The relative percent difference was 25 and 17 % for questions on meal size and consumption frequency, respectively. Study participants

Arch Environ Contam Toxicol

were age C18 years and had been residents of the study area for [1 year. The survey was administered after Sunday services during the spring and summer of 2011–2012. At their convenience, participants were sampled during a 4-week period for each church. Every respondent was required to answer a questionnaire and gave a hair clipping. Approximately 10–100 mg of scalp hair was collected from at least two areas of the head, placed in an envelope, and marked with the respondent’s name or initials. Once the survey was completed, hair mercury results were sent back to respondents through mail along with typical hair mercury levels of the United States population and the general information about mercury exposure. Survey Instrument We chose written questionnaires instead of face-to-face verbal interviews due to the church-based sampling strategy. Essential to the success of such a survey is the involvement of a community opinion leader, that is, the church ministers. One week before the survey, the church minister communicated the study goal to their congregations, encouraged participation, and explained the value of knowing one’s mercury exposure level regardless of whether it was low or high. In addition, trained College of William & Mary staff and church volunteers who speak both English and their ethnic language administered the survey, communicated with respondents, and took hair samples. Two types of visual aids were used to maximize recall reliability. First, a seafood brochure including the names and pictures of commonly eaten species was provided to each respondent. The English names were also translated into the ethnic language, and more than one name was provided for some species because people from different places use different common names. The second visual aid was a series of pictures of seafood meals with different serving sizes (1, 2, 4, 8, and 16 oz and 25, 50, 100, 200, and 400 g) on a plate. Weights of the uncanned seafood items were estimated according to the provided weights of the meals, but weights of canned seafood items were estimated according to the weights given on the can (Holloman and Newman 2010). Bilingual posters were also displayed to explain the project’s significance and encourage participation. The survey design, instruments, and implementation plan were approved by the College of William & Mary Human Subjects Committee, and the sampling method and mercury measurement process for hair samples protocols were approved by the College’s Institutional Biosafety Committee.

Hair Mercury Analysis A 2-cm length of each hair sample proximal to the scalp was analyzed for mercury with a Direct Mercury Analyzer80 (Milestone Company, Shelton, Connecticut, USA) and expressed as microgram per gram (lg/g). Batches of 15–50 samples were analyzed during each analytical session. Calibration curves were established with a certified standard reference material DORM-3 (Fish Protein, National Research Council of Canada) for every batch of samples, and linearity with an r2 [ 0.99 was required for each calibration curve. Precision and accuracy for the analytical system were quantified with a second certified standard reference material TORT-2 (Lobster Hepatopancreas, National Research Council of Canada) and 10 % replicated samples. The mean percent recovery of TORT-2 was 104.5 % (SD = 1.2 %, n = 32). Method precision expressed as relative percent difference for duplicate samples averaged 0.9 % (SD = 0.7 %, n = 16). Data Collection Seafood Items, Relative Proportion, Consumption Frequency, Meal Size, and Body Weight Names of the commonly eaten seafood items were reported by each respondent in the questionnaire as were the corresponding estimates of consumption frequency (number of meals/month) and meal size (g/meal). Reported seafood items from respondents of each church were divided into different seafood groups, and the number of items in each group was applied as the community-based consumption frequency of that item. The relative proportion of an item was calculated by dividing its consumption frequency by the number of all items reported by that community, which was used in later Monte Carlo simulations as the probability of person selecting a certain seafood item. Given that certain local fish, such as striped bass, oyster, and flounder, were only available seasonally, the consumption frequency (number of meals/month) of such items were estimated separately for each season (spring, summer, fall, and winter) by respondents. The ‘‘number of meals/month’’ for each season was multiplied by three and summed to obtain an annual consumption frequency (number of meals/year) of an item, which was divided by 365 to yield a daily consumption frequency (number of meals/day). When meal size was asked in the questionnaire, respondents were required to exclude weights of inedible parts and side dishes. In addition, different units of meal size (gram, kilogram, ounce, pound) and body weight (kilogram, pound) reported by respondents were converted to grams for meal size and to kilograms for body weight.

123

Arch Environ Contam Toxicol

Missing Data, Suspicious Values, and Outliers

Mercury Concentrations in Seafood

Two methods were applied for missing data of consumption frequency and meal size: (1) If the item with a missing value was reported by other respondents, the missing value was replaced by randomly selecting one from the answers of other respondents; and (2) If the item with a missing value was not reported by any other respondent, the value was replaced by the answer of that respondent’s general consumption frequency (i.e., ‘‘How often do you usually have a seafood meal?’’) or meal size (i.e., ‘‘How much do you usually eat during a seafood meal?’’). Such questions on general consumption pattern (not species-specific) were included to check the precision of a respondent’s answer. A few respondents reported suspiciously high consumption data that suggested unreliability. Arithmetic means and SDs of consumption frequency and meal size for each seafood group were calculated after preliminary examination of the associated distribution of observations. If an observation was greater than 3 SDs above the arithmetic mean, the observation was treated as a suspicious value (Sechena et al. 2003). Two suspicious values for meal size (e.g., 2 kg of shrimp per meal) were replaced by the largest value that was lower than the arithmetic mean plus 3 SDs in that seafood group (Sechena et al. 2003). One respondent in the general population church was Asian, and one respondent in the Chinese church was a non-Asian. The answers of these two questionnaires were deemed unrepresentative and excluded from further analysis. An item was also omitted from the data set if the species name was missing or vaguely expressed. Among the 390 seafood items reported, 18 (4.6 %) values of consumption frequency and 18 (4.6 %) values of meal size were missing; 2 (0.5 %) values of meal size were suspicious; and 11 (2.8 %) values were dropped because of missing or vague seafood names.

Reported species were categorized as local fish (Kirkley 1997) which included the commonly eaten species caught from the lower Chesapeake Bay or market fish that were distributed and sold nationally. Mercury concentrations of the local fish were derived from a lower Chesapeake Bay finfish mercury database (Xu et al. 2013) and another recent mercury exposure study in a nearby community (Holloman and Newman 2012). Mercury concentrations of the market fish were derived from reports of the USEPA (1997, 2003); USFDA (2013); and Virginia State Water Control Board (1991), studies of mercury in United States seafood markets (Sunderland 2007), and other related literatures (Adams 2004; Andersen and Depledge 1997; Gobeil et al. 1997; Harding et al. 2005; Legrand et al. 2005; Wang et al. 2013). General mercury concentrations of fresh tuna were used instead of those for canned albacore and light tuna because most respondents could not recollect which kind of tuna was consumed. Details of mercury concentrations are included in the Supplementary Material.

Seafood Consumption Rate Seafood consumption rate (g/day) of a reported item was calculated by multiplying its consumption frequency (number of meals/day) and meal size (g/meal). Because all reported items were divided into several seafood groups as previously mentioned, the group seafood consumption rate (g/person/day) was calculated by summing all seafood consumption rates of that item in the group and dividing the number of eligible respondents in that community. Seafood consumption rates of different items reported by the same respondent were also summed to obtain an individual seafood consumption rate (g/person/day). Geometric means, SDs, and distributions of the individual seafood consumption rates for each community were determined and compared.

123

Mercury Intake Rate Mercury intake rate (lg/day) of each reported item was calculated by multiplying its seafood consumption rate (g/ day) and total mercury concentrations (lg/gram). Because all reported items were divided into several seafood groups as mentioned previously, the group mercury intake rate (lg/person/day) was calculated by summing all mercury intake rates of that item in the group and dividing by the number of eligible respondents in that community. Statistical Analysis SAS 9.2 software (SAS Cary, North Carolina, USA) was used for data analysis. A linear model between hair total mercury concentrations and mean individual seafood consumption rates of the studied communities was built using the SAS PROC REG procedure. To avoid the current debate about interpreting P values from conventional statistical tests (Altman 2000), general discussion of significant differences for data from the communities applied Cumming’s general rules-of-thumb: Nonoverlap of 95 % confidence interval (CI) indicated a P value \ 0.01, and a proportion of overlap of 95 % CIs\0.5 indicated a P value of \0.05 (Cumming 2012; Cumming and Finch 2005). Exposure Modeling Equation for Calculating Daily Methylmercury Intake Rate Daily methylmercury intake rate (DMIR) was estimated using the following equation (USEPA 2011). The required

Arch Environ Contam Toxicol

consumption frequency, meal size for each reported item, and body weight for each respondent were derived from questionnaire answers as discussed in 2.6.1. DMIR ¼

CF  MS  THg  a  b ; BW

ð1Þ

where DMIR = daily methylmercury intake rate (lg/kg BW-d); CF = consumption frequency (number of meals/ day); MS = meal size (g/meal); THg = total mercury concentration (lg/gram); BW = body weight (kg); a = methylmercury conversion factor (unitless); and b = food preparation/cooking adjustment factor (g/g). Current mercury risk assessment is based on methylmercury exposure (USEPA 2011), so total mercury concentrations applied in this equation needed to be converted to methylmercury concentrations. Methylmercury concentrations of catfish, croaker, eel, flounder, perch, sea trout, spot, and striped bass were taken from the lower Chesapeake Bay finfish mercury database (Xu et al. 2013). For the species lacking methylmercury information, a methylmercury conversion factor (a) of 0.95 was adopted according to the national-scale assessment of mercury risk by USEPA (2011). In addition, food preparation and cooking will primarily remove moisture and the nonmuscle tissue (Burger et al. 2003), thus potentially increasing mercury concentration per unit of fish mass. So mercury concentrations of raw muscle must be adjusted with a food preparation/cooking adjustment factor (b): Mercury concentrations in raw item/mercury concentrations in cooked item (Morgan et al. 1997; Burger et al. 2003). The adjustment factors of 1.6 for finfish and 1.5 for shellfish in Holloman and Newman’s study (2012) was adopted because their sampling locations were close to target area of the current study, and the highest adjustment factors would avoid the potentially underestimating human exposure (Burger et al. 2003). Set-Up Distributions of Parameters A cumulative probability distribution of DMIR was generated by Monte Carlo simulation (Crystal Ball 11.1.1.1.00 package, Oracle, Redwood Shores, California, USA). Estimated distributions of parameters (consumption frequency, meal size, body weight, and mercury concentrations) in Eq. (1) needed to be defined before running the model. Customized distributions of consumption frequency and meal size of a seafood group were built directly with the data due to the small sample size, which excluded the fitting of any conventional distribution. Data on body weight for each community were fit to beta distributions. Mercury concentrations were fit to two-parameter lognormal distributions if the original data were available; if not, normal distributions were assumed for concentration data

with only arithmetic means and SDs, and customized distributions were produced for the total mercury concentrations of cobia and goby because only the means were provided in the literature. The individual seafood consumption rates were defined by fitting to two-parameter lognormal distributions. To avoid unrealistic values being generated during simulations, such as negative body weights or extremely high mercury concentrations, a minimum of zero and a maximum of three SDs above the mean were set as limits for each distribution (Wang and Newman 2013). Monte Carlo Simulation of DMIR Monte Carlo simulation of community-based DMIR was performed with Crystal Ball. A list of 100 possible seafood items were ranking by their relative proportions (2.6.1), and distributions of consumption frequency (number of meals/day), meal size (g/meal), mercury concentration (lg/ gram), seafood consumption rate [consumption frequency * meal size (g/day)], and mercury intake rate [seafood consumption rate * mercury concentrations (lg/ day)] for each item were tabulated on the same row. In each trial of the model, several seafood items from the data set were selected as were their corresponding values of seafood consumption rate and mercury intake rate from the parameter distributions. Meanwhile, one value of individual seafood consumption rate (g/person/day) and one value of body weight (kg) were selected from the lognormal and beta distributions, respectively. The simulation continued until the total amount of seafood consumed reached the selected individual seafood consumption rate. An adjusted amount of mercury ingested (lg/day) was then calculated and divided by the selected body weight (kg) to yield a DMIR (lg/kg BW-day). The designed procedure of each trial simulated a hypothetical community member’s daily seafood choice. The whole model included 10,000 trials that produced a distribution of 10,000 estimates of DMIR for each community. Monte Carlo Simulation of Annual-Average Daily Methylmercury Intake Rate (A-DMIR) The USEPA reference does (RfD 0.1 lg/kg BW-day) of dietary methylmercury was established based on three epidemiological studies in the Seychelles Islands (Myers et al. 1995a, b, c, 1997; Davidson et al. 1995, 1998), the Faroe Islands (Grandjean et al. 1997), and New Zealand (Kjellstrom et al. 1986, 1989), in which mercury concentrations in maternal hair and cord blood were monitored over the long term (USEPA 2001a). To compare the exposure levels of the studied communities with the EPA

123

Arch Environ Contam Toxicol

RfD, the 10,000 estimates of DMIR generated previously must also be modeled by Monte Carlo simulation to yield an annual-average DMIR (A-DMIR). In each trial of the model, 365 values (based on 365 days of a year) of DMIR were randomly selected from its lognormal distribution, and the arithmetic means of the 365 values were calculated as A-DMIR. The procedure of each trial simulated a hypothetical community member’s DMIR on an annualaverage level. The whole model included 10,000 trials that produced 10,000 estimates of A-DMIR for each community. The arithmetic mean, SD, and median of the 10,000 A-DMIRs were determined. Their cumulative probability distribution was then plotted and compared with the USEPA RfD, and the dose resulted in adverse health effects.

Results Participation Rates A total of 140 people in the 4 communities responded to the survey, and the average participation rate was 41 % (Table 1). The number of church members was counted on 4 consecutive Sundays and estimated as a mean. The survey participation rates differed among communities and ranged from 37 % in the Vietnamese church to 45 % in the Chinese church.

consumed items were local species in the reference and the Chinese church communities, but four of the top five items were local species in the Vietnamese church community. The summed relative proportions in the reference, Chinese, and Vietnamese churches were 52, 67, and 51 % for the consumed market fish and 48, 33, and 49 % for the consumed local species. In conclusion, market fish were consumed more frequently compared with the local species, especially for the Chinese church, with a high difference between relative proportions of consumed market and local species. The Vietnamese communities consumed market fish more frequently than local species although four of the top five commonly eaten species were from local waters. Consumption Rates of Specific Seafood Groups The consumption rates of a specific seafood group (category) were materially different among the studied communities (Fig. 1). For the market fish, all studied communities favored shrimp, especially the Vietnamese community, which had a shrimp consumption rate of approximately 1.3 g/day. The communities from the reference and Chinese churches both consumed large amounts of salmon. Tuna was only consumed at a high rate in the reference church communities. For the local fish, people attending the reference churches favored blue crab and flounder. The Vietnamese community tended to eat a large amount of diverse local species; in contrast, the Chinese community ate only a modest amount.

Seafood Consumption Patterns Mercury Intake Rates of Specific Seafood Groups Relative proportions of commonly consumed seafood items for the reference, Chinese, and Vietnamese churches are listed in Table 2, and data of the two reference churches were pooled for clear comparisons among communities. Shrimp purchased from grocery stores or seafood markets was the most frequently consumed items for all three communities, and blue crab was the most frequently consumed local species. Two of the top five commonly

Reference church-1

15

35

42.9

30.1–56.7

The mercury intake rates of a specific seafood group (category) were also different among communities (Fig. 2). Tuna, salmon, and crabs (snow crab and king crab) were the primary species that contributed to mercury intake by the communities of the reference church and the Chinese church; however, tuna, mackerel, and crabs contributed the major amount of mercury to the Vietnamese church. People from the reference and Chinese church took in most of their mercury from market fish; people from the Vietnamese church took in mercury from both the market and locally harvested species such as croaker, striped bass, and catfish. In addition, snapper and crayfish were species only consumed by Chinese church communities that also contributed materially to mercury intake.

Reference church-2

47

120

39.2

32.1–46.7

Hair Mercury Concentrations

Chinese church

45

100

45.0

37–53.2

Vietnamese church

33

90

36.7

28.8–45.3

140

345

40.6

36.3–45.0

Table 1 Survey participation rates by communities Community

Total

123

No. of respondents

No. of church members

Participation rates (%)

95 % CI

The overall geometric mean of hair total mercury in United States females was 0.20 lg/g (McDowell et al. 2004; Table 3). No statistical difference of mean hair mercury concentrations was identified between communities of the

Arch Environ Contam Toxicol Table 2 Consumption probability for reported seafood items in different communities

Reference churches

Chinese church

Species

Relative proportion (%)

Species

Relative proportion (%)

Species

Relative proportion (%)

Shrimp

14.6

Shrimp

18.7

Shrimp

20.5

Salmon

11.5

Salmon

17.3

Blue craba

Blue crab

a

a

Vietnamese church

a

11.6

11.5

Blue crab

10.7

Croaker

10.9

Tuna

8.0

Catfisha

7.1

Tilapia

5.3

Striped bassa

6.3

Local fish. All other species were market fish

Oystera

5.3

Fig. 1 Community-based consumption rates of major reported items (g/person/day). The distance between each concentric ring represents 2 g/person/day. Names in capital letters identify local fish, and lower-

case ones indicate market fish. Data from the two reference churches were combined as one reference group

reference church and the NHANES value of 0.20 lg/g (McDowell et al. 2004). However, the geometric mean of hair mercury concentrations in the Chinese and Vietnamese churches were both[0.20 lg/g (P \ 0.01). Frequencies with which hair mercury concentrations were C0.2 lg/g (McDowell et al. 2004) were 46, 90, and 100 % in the reference, Chinese, and Vietnamese churches, respectively (Fig. 3).

Chinese church, and the Vietnamese church were 33.0 g/ person/day (95 % CI = 5.3–60.8, n = 15), 28.3 g/person/ day (95 % CI = 15.2–41.3, n = 47), 36.9 g/person/day (95 % CI = 25.1–48.7, n = 45), and 52.7 g/person/day (95 % CI = 31.1–74.3, n = 33), respectively (Fig. 4). For the United States population (male and female C14 to 45 years), the daily consumption rate of fish (finfish and shellfish from freshwater, estuarine, and marine) was estimated to be 12.8 g/person/day (90 % CI = 12.05–13.61) (USEPA 2002). Except for reference church-1, individual seafood consumption rates of the other church communities were significantly [12.8 g/day as gauged by their nonoverlapping CIs.

Floundera a

Tuna

9.9

Individual Seafood Consumption Rates Means and 95 % CIs of individual seafood consumption rates in reference church-1 (Unity Fellowship Church), reference church-2 (Gloucester Point Baptist Church), the

8.0

123

Arch Environ Contam Toxicol

Fig. 2 Mercury ingestion rates of major reported items (lg/person/ day). The distance between each concentric ring of market fish represents 0.4 lg/person/day, and the distance between each concentric ring of local fish represents 0.1 lg/person/day. Capital letters

represent names of local fish, and lowercase letters represent names of market fish. Data from the two reference churches were combined as one reference group

Consumption Rate and Hair Mercury Level

concentration will increase 0.05 lg/g with an increase in seafood consumption of 1 g/day.

A simple linear regression model was developed in SAS for mean individual seafood consumption rates and mean hair mercury concentrations of the four studied communities (Fig. 4). A strong positive relationship was identified: [Hair total mercury] = 0.05 [individual seafood consumption rates]—1.29. The regression r2 was 0.98, and the 95 % CI of the slope was 0.03–0.07. Dietary mercury ingestion through seafood was concluded to be positively related to mercury exposures in the studied church communities of coastal Virginia. In addition, the slope indicated that a studied committee member’s hair mercury

123

Daily Methylmercury Intake Rate (DMIR) Arithmetic means and 95 % CIs of DMIR generated by the Monte Carlo simulation in the reference, Chinese, and Vietnamese church populations were 0.046 lg/kg BW-day (95 % CI = 0.045–0.048), 0.092 lg Hg/kg BW-day (95 % CI = 0.088–00.095), and 0.119 lg Hg/kg BW-day (95 % CI = 0.116–0.122), respectively. Data from the two reference churches were combined to generate a reference group. The mean DMIR in the pooled reference church

Arch Environ Contam Toxicol Table 3 Hair total mercury concentrations (lg/g) from the current study and the NHANES study (McDowell et al. 2004)

Concentrations

NHANES

Reference church-1

Reference church-2

Chinese church

Vietnamese church

No.

1726

15

47

45

33

Geometric mean

0.20

0.28

0.21

0.52

1.46

(95 % CI)

(0.16–0.24)

(0.14–0.56)

(0.16–0.28)

(0.41–0.65)

(1.21–1.76)

Arithmetic mean

0.47

0.43

0.32

0.69

1.68

(95 % CI)

(0.35–0.58)

(0.05–0.81)

(0.22–0.42)

(0.51–0.86)

(1.31–2.04)

10th

0.04

0.11

0.08

0.22

0.69

5th

0.09

0.16

0.11

0.30

1.15

50th

0.19

0.18

0.20

0.57

1.49

75th

0.42

0.65

0.40

0.82

1.85

90th 95th

1.11 1.73

1.58 1.58

0.87 1.00

1.44 1.75

2.75 3.76

Percentile

Fig. 3 Cumulative frequencies of hair total mercury concentrations of respondents in the reference, Chinese, and Vietnamese churches and the NHANES studies. Data from the two reference churches were combined as one reference group

community was statistically lower than that of the Chinese church communities (P \ 0.01), which itself was statistically lower than the value of the Vietnamese church communities (P \ 0.01). Annual-Average Daily Methylmercury Intake Rate (ADMIR) Arithmetic means and 95 % CIs of the A-DMIR generated by the Monte Carlo simulation in the pooled reference, Chinese, and Vietnamese churches were 0.04014 lg Hg/kg BW-day (95 % CI = 0.04008–0.04021), 0.06825 lg Hg/kg BW-day (95 % CI = 0.06817–0.06833), and 0.07737 lg Hg/kg BW-day (95 % CI = 0.07724–0.07750), respectively. The A-DMIR of Vietnamese church communities was statistically greater than that of Chinese church communities (P \ 0.01), which itself was statistically greater than the value of the reference church populations (P \ 0.01). The

Fig. 4 Relationship between individual seafood consumption rates and hair total mercury concentrations of the studied communities. A Reference church-1. B Reference church-2. C Chinese church. D Vietnamese church. The area of solid lines, the shadowed area, and the area of dashed lines indicate the fit of these data, the 95 % CIs, and the 95 % prediction limits, respectively

means and upper 95 % CIs for all studied communities were lower than the USEPA RfD of 0.1 lg/kg BW-day. Annually, there was no chance that members of the studied communities would take in more mercury than the USEPA RfD (Fig. 5).

Discussion Seafood-Consumption Patterns Seafood consumption patterns were distinct for the studied communities (Fig. 1). For instance, different individual

123

Arch Environ Contam Toxicol

consumption rates were identified for the studied communities. People from reference church-2, the Chinese, and the Vietnamese communities had daily individual consumption rates statistically [2.8 g/person/day (90 % CI = 2.0–13.6), the value reported for the United States population for both sexes and all ages (USEPA 2002). The small sample size (n = 15) from reference church-1 resulted in wide 95 % CIs, which included the value of 12.8 g/person/day, so the difference in consumption rates between the two reference churches could have been influenced by their different sample sizes. Meanwhile, large variations in meal size and consumption frequency were also reported by different communities. The Chinese community reported a larger meal size (arithmetic mean 248.6 g/meal; 95 % CI = 165.8–331.4) than the reference churches (arithmetic mean 156.2 g/meal; 95 % CI = 127.4–175.0) (P \ 0.05), and the Vietnamese community reported an even greater consumption frequency (arithmetic mean 11.3 meals/month; 95 % CI = 8.5–14.2) compared with the reference church (arithmetic mean 4.2 meals/month; 95 % CI 3.3–5.1) (P \ 0.01). In addition, the reference and Vietnamese communities tended to select their seafood items from both the local water and the market, but the Chinese community showed more of a preference for market fish (Table 2, Fig. 1). Reasons for this difference might be related to the fishing activities in a community. Only 6 % of the respondents in the Chinese community reported that they often fish in the summer, but 29 and 43 % of the respondents, respectively, in the reference and the Vietnamese communities reported fishing in the spring, summer, and fall. A high proportion of recreational fishers in the Vietnamese community was one important contributor to their high consumption rates of diverse local species.

Fig. 5 Cumulative frequency of A-DMIR (lg/kg BW-day) among different communities. Data from the two reference churches were combined as one reference group

123

Another important component of seafood consumption pattern is the choice of different finfish parts. People with different ethnic backgrounds do not eat fillets only (Sechena et al. 2003). Fish head, roe, and stomach were consumed by people from the studied Chinese and Vietnamese communities; skin, tail, bones, and eyes were also consumed by the Chinese community. In contrast, no respondents in the reference communities reported any consumption of the finfish parts other than the fillet. Because mercury concentrations vary in different parts of finfish (Newman et al. 2011b), such differences could further contribute to differences in seafood consumption risks between some ethnic communities and the general population. Although this part was not included in the current study due to the lack of mercury information in different organs of many finfish species, it may be valuable to relative studies in the future, especially for the ethnic groups with a high contaminant risk. Few studies on mercury exposure in this area or for these ethnic groups have been performed before. Holloman and Newman (2010) surveyed one African-American community from the Southeast community of Newport News (Virginia, USA) for their mercury exposure through fish consumption. A high consumption rate, much greater than the rates reported by this study, was estimated as 147.8 g/person/day due to the frequent consumption of whiting, shrimp, and canned tuna. In addition, the frequently eaten items were different except for the shrimp. Another study of seafood consumption on 260 Vietnamese refugees in northern Florida reported a high consumption frequency of 32 meals/month (Crane and Green 1980). However, their results are not directly comparable with ours because the Vietnamese population sampled in the current study are not recent refugees.

Arch Environ Contam Toxicol

Mercury Intake Rates Seafood consumption patterns, including consumed seafood items, consumption frequency, meal size, and choices of finfish parts, all contribute importantly to dietary mercury intake. The Vietnamese church community had greater DMIR and A-DMIR than the Chinese church community (P \ 0.01), which in turn, were also greater than those of the reference church communities (P \ 0.01) (Fig. 5). This sequence corresponds to the results of individual seafood consumption rates (Fig. 4). We believe that different choices of seafood items were the primary reason for different mercury intake and exposures among the two ethnic communities. Although the Chinese community reported a large meal size and the Vietnamese community reported a high consumption frequency, their individual consumption rates were not statistically different from each other. However, they had very distinct choices of seafood species (Fig. 1). Figure 2 was that the Chinese church community received most of their mercury exposure from commercial market fish (e.g., salmon), but the Vietnamese community were exposed from both market and local fish (e.g., striped bass and croaker). Striped bass is a top predator in Chesapeake Bay and its estuaries and accumulates more mercury than other fish species (Xu et al. 2013). Plus this fish is usually sold on Sundays in the Vietnamese church by the recreational fishers from that community. Hence, the large amount of striped bass consumption may be the dominant reason of high dietary mercury intake rate in the studied Vietnamese church community. Mercury Exposure Total mercury concentrations in blood and hair are common indicators of methylmercury exposure through fish consumption for people who are not occupationally exposed to inorganic mercury (Carrington and Bolger 2002; Iwasaki et al. 2003). Hair total mercury in this study was used to assess the validity of estimates derived from food frequency questionnaire and provide a more accurate estimate of actual mercury exposure through fish consumption (Mina et al. 2007). According to the World Health Organization (WHO), health effects were not apparent in adults with hair total mercury as high as 50 lg/g (Tsubaki 1968), and apparent fetal effects would be unlikely if maternal hair total mercury was\10 lg/g for pregnant women (Grandjean et al. 1997). The published exposure threshold of the WHO equals 14 lg/ g of mercury concentration in the hair. All observed hair total mercury concentrations in this study were lower than the WHO threshold, although hair total mercury from the Chinese and Vietnamese were statistically greater than that of

the United States general concentration of 0.2 lg/g (McDowell et al. 2004). Correspondingly, all calculated methylmercury intake rates from Monte Carlo simulation were lower than the EPA oral RfD of 0.1 lg/kg BW-day (USEPA 2001a, b). Hair mercury concentrations for women in the nearby African-American community can be predicted by the linear regression between hair mercury and consumption rates (Fig. 4). With the reported seafood consumption rate of 147.8 g/day (95 % CI = 117.6–185.8) (Holloman and Newman 2010), their hair mercury concentrations would be 6.1 lg/g (95 % CI = 4.6–8.0), which is still lower than the WHO threshold. Hair analysis is a reliable and convenient way to determine personal mercury exposure. The clear advantages are that mercury in the hair is not remobilized once deposited, and sampling is simple. The Lower Chesapeake Bay Seafood Consumption Survey was designed on an annual-average basis. Assuming a hair growth rate of 1 cm/ month, a 12-cm segment of hair measured from the scalp would correspond to that deposited in one year. Only 40 % of the respondents’ hair samples were of sufficient length to obtain 12 cm. Consequently, the proximal 2 cm of each respondent’s hair from the scalp was analyzed and was assumed to reflect mercury exposure during the year. Because hair samples were taken during the season of greatest fish consumption (spring and summer), the estimated annual-average exposure reflected by hair results would likely be greater than the actual annual exposure. In addition, the influence of hair treatment on hair mercury levels was not included in this study. The NHANES study (McDowell et al. 2004) indicated that there was no difference in hair total mercury concentrations between the treated hair and untreated hair groups, but Dakeishi et al. (2005) found out that the process of artificial hair-waving could decrease hair total mercury concentrations by approximately 30 %. Survey Biases 1. 2.

3.

Recall bias is a characteristic of surveys asking about a long-term consumption habits (Sechena et al. 2003). Participation rates of the four churches were \50 %, and small sample size could have biased the results to an undetermined degree. For this reason, the relationships between demographic questions and mercury exposure were not explored in more depth. In addition, participation rates differed among communities. It was difficult to randomly sample the targeted populations and select completely unbiased samples. Willingness to engage in the survey might have been lower for people who do not usually eat fish or for people who eat large amounts of fish and would be unwilling to change their diets. Only one respondent in

123

Arch Environ Contam Toxicol

4.

5.

the Gloucester Point Baptist Church reported not being a consumer of seafood. The survey method of written questionnaires might create some misunderstandings of the questions, particularly after translation to the primary community language. Imprecision of answers was checked with a qualitycontrol procedure. Questions about important information (e.g., seafood items, consumption frequency, and meal size) were asked in two ways, and the answers were checked for the relative percent difference. The average relative percent difference of meal size and consumption frequency was 22 %. The memory imprecision would either decrease or increase the calculated mercury exposure. However, the investigators believe that the inclusion of specific validations of hair total mercury concentration provide a reasonable certainty as to the accuracy of the estimated consumption data produced in this study.

Focus on Health Advisory The USEPA recommends that women of child-bearing age and young children eat B12 oz (2 average meals) a week of a variety of fish and shellfish that are low in mercury (USEPA 2004). It was suggested that general fish consumption among pregnant women decreased after this national mercury advisory was issued (Oken et al. 2003). Women were forgoing any potential benefits of fish consumption due to the ill-defined potential of harm. Some respondents of our survey considered mercury as a pervasive contaminant, so they adopted very low frequency of fish consumption or stopped eating fish altogether. This common misconception of mercury leads to low consumption of fish that are rich in omega-3 polyunsaturated fatty acids and beneficial to cardiovascular disease. According to Rheinberger and Hammitt’s study (2012), the relevant potential harm to nontarget populations who misunderstand the advisory is large compared with the benefit. In view of these facts, it is helpful to supplement the general fish consumption advisory with specific exposure estimates for specific communities and regions as performed here. Although mercury exposure levels calculated by Monte Carlo simulation and suggested by hair mercury concentrations of the two Asian communities were not high enough to trigger concern about possible health effects, it is possible that other contaminants might cause a health concern with the frequent seafood consumption patterns. For mercury, its concentration in commonly eaten fish from the region was relatively low (Xu et al. 2013), and members of these local communities with high seafood consumption rates remained safe from mercury exposure. Despite this conclusion of low

123

risk, the relatively high seafood consumption rates of the studied ethnic communities, combined with the vague anxiety in women of childbearing age about fish consumption, required clarification beyond the general consumption advice of federal agencies.

Conclusion The two questions can now be answered with the study results. Relative to the first question, dietary customs of recent Asian immigrants to coastal Virginia influenced mercury exposure. Fish consumption patterns, including consumed species, meal size, consumption frequency, parts of finfish consumed, and individual seafood consumption rates, differed among the studied communities, which resulted in different levels of mercury ingestion. The Monte Carlo–derived annual-average daily methylmercury intake rates indicated a greater mercury exposure of the Vietnamese community compared with the Chinese community as well as a greater exposure of the Chinese community compared with the reference churches communities. Hair mercury concentrations correspond to the exposure levels calculated by Monte Carlo simulation because dietary mercury ingestion was positively related to hair mercury concentrations. Relative to the second question, mercury exposures of the studied Asian communities were greater than those of the reference communities, but they were not enough to cause a health concern. According to the results of Monte Carlo simulation and hair mercury concentrations, their daily methylmercury intake rates were lower than the USEPA RfD of 0.1 lg/kg BW-day, and their hair mercury concentrations were lower than the published WHO exposure threshold of 14 lg/g. Even so, their relatively greater seafood and mercury ingestion rates still suggest the need for additional focus related to contaminants other than mercury within the studied region and other regions in the world. Acknowledgments The minister Muliang Gong of the Peninsula Chinese Baptist Church, Joseph Phien Nguyen of Our Lady of Vietnam Chapel, Ginny Roll of the Unity Fellowship Church, and Bud Goude of the Gloucester Point Baptist Church communicated the goal of this study to their church communities and encouraged people to take the survey. Solomon Chak, Yuan Dong, Vi Nguyen, Harry Wang, and Jude Eastman assisted by communicating with people, sending out and taking back questionnaires, and taking hair samples. M. C. Newman was the A. Marshall Acuff Jr. Professor of Marine Science at the College of William & Mary’s Virginia Institute of Marine Science during the tenure of this study. Funding for this study was provided by the Virginia Sea Grant. This paper is Contribution No. 3422 of the Virginia Institute of Marine Science, College of William & Mary. Conflict of interest of interest.

The authors declare that they have no conflict

Arch Environ Contam Toxicol

References Adams DH (2004) Total mercury levels in tunas from offshore waters of the Florida Atlantic coast. Mar Pollut Bull 49:659–663 Altman DG (2000) Confidence intervals in practice. In: Altman D, Machin D, Bryant TN, Gardner MJ (eds) Statistics with confidence: confidence intervals and statistical guidelines, 2nd edn. British Medical Journal Books, London, pp 6–14 Andersen JL, Depledge MH (1997) A survey of total and methylmercury in edible fish and invertebrates from Azorean Waters. Mar Environ Res 44:331–350 Burger J, Dixon C, Boring CS (2003) Effect of deep-frying fish on risk from mercury. J Toxicol Environ Health 66:817–828 Carrington C, Bolger M (2002) An exposure assessment for methylmercury from seafood from consumers in the United States. Risk Anal 22:701–712 Compeau G, Bartha R (1985) Sulfate-reducing bacteria: principal methylators of mercury in anoxic estuarine sediment. Appl Environ Microbiol 50:498–502 Crane N, Green N (1980) Food habits and food preferences of Vietnamese refugees living in northern Florida. J Am Diet Assoc 76:591–593 Cumming G (2012) Understanding the new statistics: effect sizes, confidence intervals, and meta-analysis. Routledge Taylor & Francis Group, New York Cumming G, Finch S (2005) Inference by eye. Am Psychol 60:170–180 Dakeishi M, Nakai K, Sakamoto M et al (2005) Effects of hair treatment on hair mercury—The best biomarker of methylmercury exposure? Environ Health Prev Med 10:208–212 Davidson P, Myers G, Cox C et al (1995) Longitudinal neurodevelopmental study of Seychellois children following in utero exposure to methylmercury from maternal fish ingestion: outcomes at 19 and 29 months. Neurotoxicology 16:677–688 Davidson PW, Myers GJ, Cox C et al (1998) Effects of prenatal and postnatal methylmercury exposure from fish consumption on neurodevelopment: outcomes at 66 months of age in the Seychelles child development study. JAMA 280:701–707 De Backer G, Ambrosioni E, Borch-Johnsen K et al (2003) European guidelines on cardiovascular disease prevention in clinical practice. Third joint task force of European and other societies on cardiovascular disease prevention in clinical practice. Eur Heart J 24:1601–1610 Ginsberg GL, Toal BF (2009) Quantitative approach for incorporating methylmercury risks and omega-3 fatty acid benefits in developing species-specific fish consumption advice. Environ Health Perspect 117:267–275 Gobeil C, Clermont Y, Paquette G (1997) Concentration en mercure, plomb et cadmium chez diverses especes de poisson de fond, poisons pelagiques et de crustaces de l’estuaireet de golfe du Saint-Laurent et du fjord du Saguenay. Rapp Stat Can Sci Halieut Aquat 1011 Grandjean P, Weihe P, White RF et al (1997) Cognitive deficit in 7-year-old children with prenatal exposure to methylmercury. Neurotoxicol Teratol 19:417–428 Guallar E, Sanz-Gallardo MI, van’t Veer P et al (2002) Mercury, fish oils, and the risk of myocardial infarction. N Engl J Med 347:1747–1754 Harding G, Dalziel J, Vass P (2005) Prevalence and bioaccumulation of methylmercury in the food web of the Bay of Fundy, Gulf of Maine. Environment Canada-Atlantic Region, Occasional Report No. 23. Dartmouth, NS, and Sackville, NB, pp 76–77 Helland IB, Smith L, Saarem K, Saugstad OD, Drevon CA (2003) Maternal supplementation with very long-chain n-3 fatty acids during pregnancy and lactation augments children’s IQ at 4 years of age. Pediatrics 111:39–44

Holloman EL, Newman MC (2010) A community-based assessment of seafood consumption along the lower James River, Virginia, USA: potential sources of dietary mercury exposure. Environ Res 110:213–219 Holloman EL, Newman MC (2012) Expanding perceptions of subsistence fish consumption: evidence of high commercial fish consumption and dietary mercury exposure in an urban coastal community. Sci Total Environ 416:111–120 Iwasaki Y, Sakamoto M, Nakai K et al (2003) Estimation of daily mercury intake from seafood in Japanese women: Akita crosssectional study. Tohoku J Exp Med 200:67–73 Kirkley J (1997) Virginia’s commercial fishing industry: Its economic performance and contributions. Special Report in Applied Marine Science and Ocean Engineering Kjellstrom T, Kennedy P, Wallis S, Mantell C (1986) Physical and mental development of children with prenatal exposure to mercury from fish. Stage 1: Preliminary test at age 4. National Swedish Environmental Protection Board No 3080, Solna, Sweden Kjellstrom T, Kennedy P, Wallis S (1989) Physical and mental development of children with prenatal exposure to mercury from fish. Stage 2: Interviews and psychological tests at age 6. National Swedish Environmental Board Report No 3642, Solna, Sweden Lamborg CH, Balcom PH, Fitzgerald WF et al (2002) Modern and historic atmospheric mercury fluxes in both hemispheres: global and regional mercury cycling implications. Global Biogeochem Cycles 16:1104 Legrand M, Arp P, Ritchie C, Chan HM (2005) Mercury exposure in two coastal communities of the Bay of Fundy, Canada. Environ Res 98:14–21 Manohar DM, Krishnan KA, Anirudhan TS (2002) Removal of mercury (II) from aqueous solutions and chlor-alkali industry wastewater using 2-mercaptobenzimidazole-clay. Water Res 36:1609–1619 Mason RP, Fitzgerald WF, Morel FMM (1994) The biogeochemical cycling of elemental mercury: anthropogenic influences. Geochim Cosmochim Acta 58:3191–3198 McDowell MA, Dillon CF, Osterloh J et al (2004) Hair mercury levels in US children and women of childbearing age: reference range data from NHANES 1999-2000. Environ Health Perspect 112:1165–1171 Mina K, Fritschi L, Knuiman M (2007) A valid semiquantitative food frequency questionnaire to measure fish consumption. Eur J Clin Nutr 61:1023–1031 Morgan JN, Berry MR, Graves RL (1997) Effects of commonly used cooking practices on total mercury concentration in fish and their impact on exposure assessments. J Expo Anal Environ Epidemiol 7:119–133 Myers GJ, Davidson PW, Cox C et al (1995a) Neurodevelopmetnal outcomes of Seychellois children sixty-six months after in utero exposure to methylmercury from a maternal fish diet: pilot study. Neurotoxicology 16:639–652 Myers GJ, Marsh DO, Cox C et al (1995b) A pilot neurodevelopmental study of Seychellois children following in utero exposure to methylmercury from a maternal fish diet. Neurotoxicology 16:629–638 Myers GJ, Marsh DO, Davidson PW et al (1995c) Main neurodevelopmental study of Seychellois children following in utero exposure to methylmercury from a maternal fish diet: outcome at six months. Neurotoxicology 16:653–664 Myers GJ, Davidson PW, Shamlaye CF et al (1997) Effects of prenatal methylmercury exposure from a high fish diet on developmental milestones in the Seychelles Child Development Study. Neurotoxicology 18:819–830

123

Arch Environ Contam Toxicol Newman MC, Xu X, Condon A, Liang L (2011a) Floodplain methylmercury biomagnification factor higher than that of the contiguous river (South River, Virginia USA). Environ Pollut 159:2840–2844 Newman MC, Xu X, Cotton CF, Tom KR (2011b) High mercury concentrations reflect trophic ecology of three deep-water chondrichthyans. Arch Environ Contam Toxicol 60:618–625 Oken E, Kleinman KP, Berland WE et al (2003) Decline in fish consumption among pregnant women after a national mercury advisory. Obstet Gynecol 102:346–351 Oken E, Wright RO, Kleinman KP et al (2005) Maternal fish consumption, hair mercury and infant cognition in a US cohort. Environ Health Perspect 113:1376–1380 Oken E, Radesky JS, Wright RO et al (2008) Maternal fish intake during pregnancy, blood mercury levels, and child cognition at age 3 years in a US cohort. Am J Epidemiol 167:1171–1181 Priori SG, Aliot E, Blomstrom-Lundqvist C (2003) European Society of Cardiology update of the guidelines on sudden cardiac death of the European Society of Cardiology. Eur Heart J 24:13–15 Rheinberger CM, Hammitt JK (2012) Trade-offs in fish consumption: a public health perspective. Environ Sci Technol 46:12337– 12346 Sechena R, Liao S, Lorenzana R, Nakano C, Polissar N, Fenske R (2003) Asian American and Pacific Islander seafood consumption—A community-based study in King County, Washington. J Expo Anal Environ Epidemiol 13:256–266 Smith SC, Allen J, Blair SN et al (2006) AHA/ACC guidelines for secondary prevention for patients with coronary and other atherosclerotic vascular disease: 2006 update. Circulation 113:2363–2372 Sunderland EM (2007) Mercury exposure from domestic and imported estuarine and marine fish in the US seafood market. Environ Health Perspect 115:235–242 Tsubaki T (1968) Methylmercury poisoning along the Agano river— Researches by Niigata University [in Japanese]. Rinsho Shinkeigaku 8:511–520 United States Environmental Protection Agency (USEPA) (1997) Mercury study report to the Congress. EPA-452/R-97-006. Volume IV: An assessment of exposure to mercury in the United States. Office of Air Quality Planning & Standards and Office of Research and Development, Washington, DC United States Environmental Protection Agency (USEPA) (2001a) Integrated Risk Information System (IRIS): Methylmercury. http://www.epa.gov/iris/subst/0073.htm. Accessed 15 June 2014 United States Environmental Protection Agency (USEPA) (2001b) Water quality criterion for the protection of human health: Methylmercury (EPA-823/R-01-001). Office of Water, Washington DC

123

United States Environmental Protection Agency (USEPA) (2002) Estimated per capita fish consumption in the United States (EPA-821- C- 02-003). http://water.epa.gov/scitech/swguidance/ fishshellfish/outreach/upload/2002_08_28_fish_consumption_ report.pdf. Accessed 16 June 2014 United States Environmental Protection Agency (USEPA) (2003) Mercury in Marine Life Database. Office of Wetlands, Oceans, and Watersheds Oceans and Coastal Protection Division, Washington United States Environmental Protection Agency (USEPA) (2004) National listing of fish advisories (EPA-823-F-04-016). Office of Water, Washington United States Environmental Protection Agency (USEPA) (2009) EPA releases report on fish contamination in US lakes and reservoirs (EPA-823-F-09-008). Office of Water, Washington United States Environmental Protection Agency (USEPA) (2011) Revised technical support document: National-scale assessment of mercury risk to populations with high consumption of selfcaught freshwater fish (EPA-452/R-11-009). In support of the appropriate and necessary finding for coal- and oil-fired electric generating units. Office of Air Quality Planning and Standards Health and Environmental Impacts Division, North Carolina United States Food and Drug Administration (USFDA) (2013) Mercury concentrations in fish: FDA Monitoring Program (1990–2010). Food and Drug Administration, Washington, DC. http://www.fda.gov/Food/FoodborneIllnessContaminants/Metals/ ucm191007.htm. Accessed 16 June 2014 Virginia Department of Health (2013) Consumption advisories and restrictions in effect for Virginia waterways. http://www.vdh. virginia.gov/epidemiology/DEE/PublicHealthToxicology/Advi sories/index.htm. Accessed 16 June 2014 Virginia State Water Control Board (1991) Mercury concentration of yellow goby from Virginia. http://fish-mercury.findthedata.org/l/ 34456/Yellowfin-goby. Accessed 16 June 2014 Wang J, Newman MC (2013) Mercury dietary exposure of three bird species nesting on a contaminated floodplain (South River, VA, USA). Integr Environ Assess Manag 9:285–293 Wang J, Newman MC, Xu X, Condon A, Liang L (2013) Higher and more variable methylmercury biomagnification factors for floodplain than the contiguous river (South River, Virginia USA). Ecotoxicol Environ Saf 92:191–198 West Virginia Department of Health and Human Resources (2014) Fish consumption advisories available for 2014. http://www. wvdhhr.org/fish/Current_Advisories.asp. Accessed 16 June 2014 Xu X, Newman MC, Fabrizio MC, Liang L (2013) An ecologicallyframed mercury survey of finfish of the lower Chesapeake Bay. Arch Environ Contam Toxicol 65:510–520