Report on the Environment

Printer-Friendly PDF

Technical Documentation

Identification

1. Indicator Title

Blood Mercury Level

2. ROE Question(s) This Indicator Helps to Answer

What are the trends in human exposure to environmental contaminants?

3. Indicator Abstract

This indicator describes the presence of mercury in the blood of segments of the U.S. population from 1999 to 2018. Mercury can cause developmental and neurological problems, especially in children. This indicator shows how exposure to this environmental contaminant has changed over time for the general population and varied across subgroups, with particular attention on children and women of child-bearing age.

4. Most Recent Update
07/2023

Top of Page

Data Sources

5. Data Sources

The data used for this indicator represent blood samples collected and analyzed for mercury from a subset of study participants during the continuous (1999–2018) National Health and Nutrition Examination Survey (NHANES). The Centers for Disease Control and Prevention’s (CDC’s) National Center for Environmental Health conducts the laboratory analyses.

6. Data Availability

For Exhibits 1 through 3, the underlying laboratory data are available online in SAS® transport file format, along with the questionnaires and related documentation, at https://wwwn.cdc.gov/nchs/nhanes/Default.aspx (CDC, 2022a). For Exhibits 4 through 6, data were obtained from CDC’s National Report on Human Exposure to Environmental Chemicals: Analysis of whole blood, serum, and urine samples, NHANES 1999-2018, which was updated in March 2022, and presents the results of the ongoing NHANES (CDC, 2022b).

For the most part, individual-level data are available, but data access limitations exist for some variables due to confidentiality issues. Disclosure risks and issues pertaining to confidentiality protection prevent CDC’s National Center for Health Statistics (NCHS) from releasing all of the NHANES demographics variables publicly. Additional information may be publicly accessible through the NCHS Research Data Center (RDC). The RDC website has information about special request data files. Refer to the NHANES 2017-2018 Data Documentation, Codebook, and Frequencies Demographic Variables and Sample Weights (DEMO_J) for additional information on the most recent survey period included in this indicator: https://wwwn.cdc.gov/Nchs/Nhanes/2017-2018/DEMO_J.htm. 

Top of Page

Methodology

7. Data Collection

Survey Design

NHANES is a series of surveys conducted by NCHS that is designed to collect data on the health and nutritional status of the civilian, non–institutionalized U.S. population. Blood samples were collected from a random sample of NHANES participants and are intended to provide estimates for the U.S. population. Participants are selected through a complex statistical process using the most current Census information, meaning that NHANES first divides the U.S. into communities and neighborhoods, and neighborhoods and housing units are selected at random (CDC, 2021). This indicator is based on a national probability-based sampling design and is deemed of sufficient quality for generalization to the nation.

Survey design documentation can be found at:

The survey represents a random sample of participants and is not designed to select or exclude participants based on their potential for low or high exposure to a chemical, nor does the design permit examination of exposure levels by locality, state, or region; seasons of the year; proximity to sources of exposure; or use of particular products (CDC, 2022c). However, NHANES is designed to provide statistically representative national averages. NHANES data provide a snapshot of the health and nutrition of the U.S. population. As a result, survey participants are from a broad range of age groups and racial/ethnic backgrounds, and each participant represents up to 65,000 other U.S. residents (CDC, 2021).

Beginning in 1999, NHANES became a continuous and annual survey. The sampling plan for each year follows a complex, stratified, multistage, probability-cluster design to select a representative sample of the civilian, non-institutionalized U.S. population. Every year, approximately 5,000 individuals of all ages are interviewed in their homes and complete the health examination component of the survey (CDC, 2021). Data are released in 2-year cycles. The 1999-2000, 2001-2002, 2003-2004, 2005-2006, 2007-2008, 2009-2010, 2011-2012, 2013-2014, 2015-2016, and 2017-2018 surveys were used for this analysis.

To produce more reliable statistics (i.e., to enable the detection of meaningful differences between populations), the sampling methodology is purposely designed to oversample selected subgroups. For example, 1999-2006 surveys included over-samples of low-income persons, adolescents (12 to 19 years), older Americans (60 years or older), African Americans, Mexican Americans, and pregnant women (CDC, 2012). Oversampling schemes can change, however, across survey periods. Beginning in 2007, oversampling of pregnant women and adolescents was discontinued to enable oversampling of the entire Hispanic population instead of just the Mexican American population (CDC, 2013b). The primary change in the NHANES 2011-2014 design was adding the oversampling of Asians (CDC, 2014). In the 2015-2016 and 2017-2018 surveys, oversampled subgroups included Hispanics, non-Hispanic blacks, non-Hispanic non-black Asians, non-Hispanic whites, and other non-Hispanics (other than black, Asian, or white) at or below 185 percent of the Department of Health and Human Services’ poverty guidelines, as well as non-Hispanic whites and other non-Hispanics aged 80 years and older (CDC, 2020).

CDC presents the measurements produced by NHANES used for Exhibits 1 through 3 of this indicator in the NHANES laboratory files available online in SAS® transport file format (CDC, 2022a). CDC presents the measurements produced by NHANES used for Exhibits 4 through 6 in its “National Report on Human Exposure to Environmental Chemicals: Analysis of Whole Blood, Serum, and Urine Samples, NHANES 1999-2018” (CDC, 2022b).

Specimen Collection and Analysis

In NHANES 1999-2000 and 2001-2002, blood mercury levels were only measured in subsamples of children aged 1-5 years and females aged 16-49 years. Beginning in the 2003-2004 survey, the analysis of blood mercury was expanded to include survey participants aged 1 year or older.

The measurement of total blood mercury includes both inorganic and organic forms. In the general population, the total blood mercury concentration is due mostly to the dietary intake of organic forms, particularly methylmercury (CDC, 2017).

When the continuous NHANES began in 1999, total blood mercury was measured by flow injection cold vapor atomic absorption analysis with on-line microwave digestion, based on the method by T. Guo and J. Bassner. Decomposition of organic mercury compounds in blood occurs mainly while the sample (mixed with bromate-bromide reagent and hydrochloric acid) flows through the digestion coil in the microwave. Further decomposition of organic mercury is achieved by on-line addition of potassium permanganate. The total (organic + inorganic) mercuric mercury released is reduced to mercury vapor by sodium tetrahydroborate. The mercury vapor is measured by the spectrometer at 253.7 nanometers (nm). Inorganic mercury in whole blood is measured by using stannous chloride as a reductant without employing a microwave digestion system. Mercury vapor (reduced from inorganic mercury compounds) is measured via the same quartz cell at 253.7 nm. The difference in the total reduced mercury (by sodium tetrahydroborate) and inorganic reduced mercury (by stannous chloride) represents organic mercury in whole blood. The unit of analysis used for this indicator was micrograms per liter (µg/L) (CDC, 2005). The NHANES methods for total blood mercury data collection have changed slightly over time. For more information, see the overview of lab procedures for all NHANES surveys considered in this indicator:

Detailed specimen collection and processing instructions are discussed in the NHANES Laboratory Procedure Manual (LPM) for each survey period (e.g., CDC, 2005, 2013a). Each chapter in the LPM specifies the procedures to be used for collecting, labeling, processing, preserving, and transporting specimens for each method used in the survey.

Documentation

Details of the design and content of each survey and the public use data files are available at https://wwwn.cdc.gov/nchs/nhanes/Default.aspx.

Information on the program and collection procedures is available at https://www.cdc.gov/nchs/data/series/sr_01/sr01_056.pdf (PDF) (37 pp, 452K). 

NHANES analytic guidelines are outlined in https://www.cdc.gov/nchs/data/series/sr_02/sr02_161.pdf (PDF) (24 pp, 735K) for 1999-2010, in https://www.cdc.gov/nchs/data/nhanes/analytic_guidelines_11_12.pdf (PDF) (13 pp, 106K) for 2011-2014, and in https://www.cdc.gov/nchs/data/series/sr_02/sr02-184-508.pdf (PDF) (35 pp, 1.7MB) for 2015-2018.

Survey-specific documentation can be found at:

8. Indicator Derivation

To create Exhibits 1 through 3 of this indicator, NHANES public use data files were downloaded from the CDC’s NCHS website (https://wwwn.cdc.gov/nchs/nhanes/Default.aspx) (CDC, 2022a). Compilation and statistical queries of NHANES data were done in strict accordance with the data and analytic guidance provided by CDC (https://wwwn.cdc.gov/nchs/nhanes/analyticguidelines.aspx).

  • Statistical Software and Procedures for Preparing Data: All summary statistics presented for this indicator were conducted using Stata Release 15 (StataCorp, 2017). For each available survey year, SAS® transport files containing the relevant demographic data and laboratory results were downloaded from the NHANES website, imported into Stata using the “FDA import” command, and merged. Subpopulation groups of interest were identified using relevant demographic variables (age: ridageyr, gender: riagendr, and race/ethnicity: ridreth1). Frequency distributions were generated for the eligible population, as well as individual subpopulations.
  • Summary Statistics: The process for calculating the geometric means is described at https://www.cdc.gov/exposurereport/data_sources_analysis.html, and the process for calculating the percentiles and confidence intervals around these measures is described at https://www.cdc.gov/exposurereport/estimate_percentiles.html. 

    Geometric means are calculated by taking the log of each concentration, calculating the mean of those log values, and taking the antilog of that mean. The geometric mean is influenced less by high values than is the arithmetic mean, providing a better estimate of central tendency for data that are not normally distributed (e.g., a long tail at the upper end of the distribution). This type of distribution is common when measuring environmental chemicals in blood.

    Percentiles (50th, 75th, 90th, and 95th) are reported to provide additional information about the shape of the distribution. Confidence intervals for percentiles were estimated using CDC-recommended methodologies (see https://www.cdc.gov/exposurereport/estimate_percentiles.html). Specifically, the method of Korn and Graubard (1998) was used to compute Clopper-Pearson 95 percent confidence intervals around the percentile estimates.
  • Weighting of Samples: NHANES is a complex sample survey; therefore, statistical analysis of the NHANES data must consider the sample design information, including sample weights and the first stage of the cluster design (the masked variance unit or primary sampling unit [PSU]). Sample weights are used to adjust for the unequal probability of selection into the survey and to adjust for possible bias resulting from non-response. PSUs must be utilized for variance estimation.

    Geometric means and percentile estimates were calculated using weighted data. Because data from the Mobile Examination Center (MEC) were used in the analysis, the Full Sample 2 year MEC Exam Weight (WTMEC2YR) was used to produce appropriate results by survey cycle, with one exception. For the 2013-2014 and 2015-2016 survey cycles, a special sample weight (WTSH2YR) is required to analyze these data properly due to blood mercury being measured in a one-half sample of participants aged 12 years and older. The MEC exam sample weight variable is provided in the Demographic Variables and Sample Weights data file for each survey period. The blood metal subsample weights for 2013-2014 and 2015-2016 are provided in the laboratory data files.
  • Reporting Data Below the Limit of Detection: Concentrations less than the limit of detection (LOD) were assigned a value equal to the LOD divided by the square root of 2 for calculation of geometric means and percentiles. The LOD is the level at which the measurement has a 95 percent probability of being greater than zero. Percentile estimates that are less than the LOD for the chemical analysis are reported as “< LOD” by CDC. If the proportion of results below the LOD was greater than 40 percent, geometric means were not calculated. Data points reported as “< LOD” are not displayed in the indicator exhibits. 

LOD values may change over time as a result of improvements to analytical methods. Consequently, for example, results may be reported as “< LOD” in the 1999-2000 data but be reported as a concentration value above the LOD in later surveys because the analytical method had improved.

No extrapolation methods were used to transform the data. NHANES selects a representative sample of the civilian, non-institutionalized population in the U.S. using a complex, stratified, multistage, probability-cluster design. The continuous NHANES survey is designed to give an annual sample that is nationally representative.

Exhibits 4 through 6 of this indicator were created by pulling data directly from the CDC’s National Report on Human Exposure to Environmental Chemicals: Analysis of whole blood, serum, and urine samples, NHANES 1999-2018, which was updated in March 2022, and presents the results of the ongoing NHANES (CDC, 2022b). The data presented by CDC were derived adhering to the same principles described above.

Of note, for some of the mercury data, the distribution of measured values is positively skewed (nonnormal) such that the majority of values lie below the mean of the distribution, with a long tail at the upper end of the distribution. In addition, reporting limitations of the measurement instrument, particularly in earlier survey periods, can result in large numbers of identical measured values. Combined, these characteristics of the data can result in asymmetric and/or narrow confidence intervals around a percentile estimate. This is more apparent for lower percentiles. For example, in the 2001-2002 survey period, the 50th percentile value is estimated at 0.8 µg/L. The lower and upper 95 percent confidence limits are computed at 0.8 and 1.0 µg/L, respectively. Because the distribution of the reported values is skewed heavily to the right and because so many values were reported at 0.8 µg/L, the point estimate is the same as the lower bound with a 0.2 difference between the lower and upper limit.

9. Quality Assurance and Quality Control

NHANES data undergo rigorous quality control and quality assurance procedures, and all protocols meet the 1988 Clinical Laboratory Improvement Act mandates. Detailed quality control and quality assurance instructions are discussed in the NHANES LPM for each survey period (e.g., CDC, 2005, 2013a). Quality assurance activities are conducted before data collection and consist of equipment calibration and training. Quality control activities occur during data collection/processing and consist of automated software edits, data analysis of technician performance, and analytic processing.

In order to ensure the utility of its statistical and analytic information products, NCHS, which oversees NHANES, conducts independent research and consults with experts in areas such as data collection, data analysis, and a variety of other relevant topics and issues. NCHS reviews the quality (including the objectivity, utility, and integrity) of information before it is released and treats information quality as integral to every step of the process. NCHS assures the security of its statistical and analytic information products through the enforcement of rigorous controls that protect against unauthorized access to the data, revision or corruption of the data, and unauthorized use of the data. See NCHS Guidelines for Ensuring the Quality of Information Disseminated to the Public for more details: https://www.cdc.gov/nchs/about/policy/quality.htm.

Top of Page

Analysis

10. Reference Points

This indicator simply provides information that exposure to mercury has occurred. The measurement of mercury or any other environmental chemical in a person’s blood does not by itself mean that the chemical has caused or will cause harmful effects.

For methylmercury, EPA uses a Reference Dose (RfD) of 0.1 microgram/kilogram (µg/kg) body weight/day as an exposure level without recognized adverse effects. The RfD is based on estimated ranges of maternal intakes and blood levels reported in epidemiological studies where neuropsychological effects were seen among offspring. A full description of EPA’s RfD for methylmercury may be found at https://cfpub.epa.gov/ncea/iris2/chemicalLanding.cfm?substance_nmbr=73.

The American Conference of Governmental Industrial Hygienists recommends that the inorganic mercury blood level of workers not exceed 15 µg/L (CDC, 2017). Occupational criteria are provided here for perspective only, and do not imply a safety level for general population exposure.

11. Comparability Over Time and Space

The data set is comparable among locations and over time. As a national survey, NHANES has a set of standardized, nationwide procedures. Changing NHANES to a continuous survey in 1999 reduced the data set’s variability.

12. Sources of Uncertainty

The primary source of uncertainty for this indicator is sampling error. The continuous NHANES is considered nationally representative, but is subject to sampling error because the statistical measurements used to characterize the national population are estimated from only a subset, or sample, of that population. To provide an estimate of that uncertainty, the method of Korn and Graubard (1998) was used to compute Clopper-Pearson 95 percent confidence intervals around the geometric mean and percentile estimates presented for this indicator (see "Indicator Derivation"). The indicator exhibits include error bars that represent the calculated 95 percent confidence intervals.

Because of the relatively small sample size for any 1-year survey period, the results of individual survey participants may be weighted heavily and greatly influence the mean value reported, especially for subgroup analyses. Variance estimates for single-year data can therefore be relatively unstable and associated with large standard errors. To improve the statistical reliability and stability of estimates with larger variances, CDC recommends using two or more 2-year cycles of the continuous NHANES data as was done for this indicator (CDC releases all NHANES survey data in 2-year cycles [e.g., 1999-2000, 2001-2002]). Combining data from 2-year cycles is particularly important when producing estimates for demographic subgroups, as doing so increases the sample size and reduces uncertainty.

A further description of uncertainty and variance estimation measurements associated with the continuous NHANES data sets is available at

https://www.cdc.gov/nchs/data/nhanes/guidelines1.pdf (PDF) (28 pp, 191K),
https://www.cdc.gov/nchs/data/nhanes/nhanes_general_guidelines_june_04.pdf (PDF) (6 pp, 28K),
https://www.cdc.gov/nchs/data/nhanes/nhanes_03_04/nhanes_analytic_guidelines_dec_2005.pdf (PDF) (14 pp, 48K),
https://www.cdc.gov/nchs/data/nhanes/analyticnote_2007-2010.pdf (PDF) (3 pp, 19K), 
https://www.cdc.gov/nchs/data/series/sr_02/sr02_161.pdf (PDF) (24 pp, 237K),
https://www.cdc.gov/nchs/data/nhanes/analytic_guidelines_11_12.pdf (PDF) (13 pp, 108K), and
https://www.cdc.gov/nchs/data/series/sr_02/sr02-184-508.pdf (PDF) (35 pp, 1.7MB).

13. Sources of Variability

Variation in biomonitoring data may be observed from person to person and over time for the same person. In addition to environmental exposure conditions, the measured blood concentrations may be influenced by age, sex, race/ethnicity, behavior, and elimination kinetics (individual and temporal variation). This indicator presents data across age, sex, and racial/ethnic subgroups to explore potential disparities across one or more of these groups.

14. Statistical/Trend Analysis

The indicator presents blood sampling data across survey periods and subpopulations. No special statistical techniques or analyses (e.g., tests of statistical significance) were used to characterize differences observed over time or across subgroups.

Top of Page

Limitations

15. Data Limitations

Limitations to this indicator include the following:

  1. The relatively small number of samples collected in a 2-year cycle (e.g., 1999-2000 or 2001-2002) may, in some cases, result in measures of central tendency that are unstable from one survey period to the next.
  2. Health-based benchmarks for blood levels of mercury have not been established.

Top of Page

References

CDC (Centers for Disease Control and Prevention). 2022a. National Center for Health Statistics, National Health and Nutrition Examination Survey. Cadmium, lead, mercury, cotinine & nutritional biochemistries (1999-2000, 2001-2002); Cadmium, lead, & total mercury - blood (2003-2004, 2005-2006, 2007-2008, 2009-2010); Cadmium, lead, total mercury, selenium, & manganese - blood (2011-2012); Blood lead, cadmium, total mercury, selenium, and manganese (2013-2014, 2015-2016, 2017-2018). Accessed April 19, 2022. https://wwwn.cdc.gov/nchs/nhanes/default.aspx.

CDC. 2022b. National report on human exposure to environmental chemicals: Analysis of whole blood, serum, and urine samples, NHANES 1999-2018. Updated March 2022. Accessed April 20, 2022. https://www.cdc.gov/exposurereport/data_tables.html.

CDC. 2022c. National report on human exposure to environmental chemicals: Data sources and data analysis. Reviewed March 24, 2022. Accessed August 3, 2022. https://www.cdc.gov/exposurereport/data_sources_analysis.html.

CDC. 2021. National Center for Health Statistics (NCHS), survey participants. Last reviewed July 22, 2021. Accessed August 3, 2022. https://www.cdc.gov/nchs/nhanes/participant.htm.

CDC. 2020. National Health and Nutrition Examination Survey, 2015-2018: Sample design and estimation procedures. National Center for Health Statistics. Vital Health Stat 2(184). https://www.cdc.gov/nchs/data/series/sr_02/sr02-184-508.pdf (PDF) (35 pp, 1.7MB).

CDC. 2017. National Biomonitoring Program, biomonitoring summary for mercury. Last reviewed April 7, 2017. Accessed August 3, 2022. https://www.cdc.gov/biomonitoring/Mercury_BiomonitoringSummary.html.

CDC. 2014. National Health and Nutrition Examination Survey: Sample design, 2011-2014. National Center for Health Statistics. Vital Health Stat 2(162). https://www.cdc.gov/nchs/data/series/sr_02/sr02_162.pdf (PDF) (33 pp, 2.3MB).

CDC. 2013a. National Health and Nutrition Examination Survey (NHANES) MEC laboratory procedures manual. NHANES 2013-2014. https://wwwn.cdc.gov/nchs/data/nhanes/2013-2014/manuals/2013_MEC_Laboratory_Procedures_Manual.pdf (PDF) (683 pp, 13.0MB).

CDC. 2013b. National Health and Nutrition Examination Survey: Sample design, 2007–2010. National Center for Health Statistics. Vital Health Stat 2(160). https://www.cdc.gov/nchs/data/series/sr_02/sr02_160.pdf (PDF) (32 pp, 1.0MB).

CDC. 2012. The National Health and Nutrition Examination Survey: Sample design, 1999-2006. National Center for Health Statistics. Vital Health Stat 2(155). https://www.cdc.gov/nchs/data/series/sr_02/sr02_155.pdf (PDF) (48 pp, 946K).

CDC. 2005. National Health and Nutrition Examination Survey (NHANES) laboratory procedure manual. Total and inorganic mercury in blood, NHANES 1999-2000. https://wwwn.cdc.gov/nchs/data/nhanes/1999-2000/labmethods/lab06_met_blood_mercury.pdf (PDF) (18 pp, 259K).

Korn, E.L., and B.I. Graubard. 1998. Confidence intervals for proportions with small expected number of positive counts estimated from survey data. Survey Methodology 24:193-201.

StataCorp. 2017. Stata Statistical Software: Release 15. College Station, TX: StataCorp LLC.

Top of Page