Article pubs.acs.org/est
Urinary Cadmium in the 1999−2008 U.S. National Health and Nutrition Examination Survey (NHANES) Anne M. Riederer,*,†,⊥ Anna Belova,‡,⊥ Barbara J. George,§,⊥ and Paul T. Anastas∥,⊥ †
U.S. Environmental Protection Agency, 1300 Pennsylvania Avenue NW, Washington, D.C. 20004, United States Abt Associates, 4550 Montgomery Avenue, Bethesda, Maryland 20814, United States § Office of Research and Development, U.S. Environmental Protection Agency, 109 T.W. Alexander Drive, Research Triangle Park, North Carolina 27711, United States ∥ Center for Green Chemistry and Green Engineering, Yale University, 225 Prospect Street, New Haven, Connecticut 06520, United States ‡
S Supporting Information *
ABSTRACT: Chronic low-level cadmium (Cd) exposure is linked to kidney and cardiovascular disease, fractures, and cancer. Diet and smoking are primary sources of exposure in the general population. We analyzed urinary Cd in NHANES 1999−2008 to determine whether levels declined significantly over the decade for U.S. children, teens, and adults (nonsmokers and smokers) and, if so, factors influencing the decline(s). For each subpopulation, we modeled log urinary Cd using variable-threshold censored multiple regression. Models included individuallevel covariates (age, gender, BMI, income, race/ethnicity/country of origin, education, survey period), smoking, housing (home age, water source, filter use), and diet (supplement use; 24-h calorie, fat, protein, micronutrient, and Cd-containing food intakes), creatinine, and survey year variables. Geometric mean urinary Cd (ng/mL) declined 20−25% in these subpopulations, and the regressions showed statistically significant declines in later years for teens and adults. While certain covariates were significantly associated with Cd by subpopulation (creatinine; age; BMI; race/ethnicity/origin; education; smokers in the home; serum cotinine; 24-h fat, Mg, Fe intakes; use of dietary supplements), they did not help explain the declines. Instead, unidentified time-related factors appeared responsible. Despite the declines, millions of Americans remain potentially at risk of adverse outcomes associated with low-level Cd exposure.
■
and ash, sewage sludge, or phosphate fertilizer applications.14 Tobacco leaves and certain foods, (including leafy green vegetables, organ meats, and shellfish) accumulate Cd in uncontaminated and contaminated areas.14,17 While average dietary intake estimates for U.S. children and adults range from 12 to 29 μg/d, gastrointestinal (GI) absorption is limited: approximately 2−5% in adult nonsmokers and 6−9% in those with low iron (Fe) stores.14,18 Smokers absorb approximately 1−3 μg per day, similar to the amount absorbed from diet.14 Although similar estimates are not available for the amount of Cd absorbed from environmental tobacco smoke (ETS), ETS exposure is associated with elevated Cd body burdens in children and other nonsmokers.14 In the United States, drinking water is considered less influential than diet as Cd is regulated in public water systems and bottled water.14 Cd can also be detected in groundwater near hazardous waste sites.14 At low exposure levels, Cd accumulates in liver, kidney, bone and other tissues. Excretion is slow, with estimated half-lives of 4−38 years in humans.14 Urinary Cd levels are considered markers of total body burden
INTRODUCTION Cadmium (Cd) is a ubiquitous metal with no known nutritive function. Chronic, low-level exposure is linked to cardiovascular and kidney disease, decreased bone mineral density/increased fractures, and cancer (lung, pancreatic, breast, bladder).1−8 Lowlevel exposure in U.S. children (6−15 years old) is also linked to learning disabilities and attention deficit hyperactivity disorder.9 The International Agency for Research on Cancer (IARC) and the U.S. Environmental Protection Agency (EPA) classify Cd as a human carcinogen and a probable human carcinogen, respectively.10,11 Oxidative stress and interference with cellular signaling are thought to underlie Cd’s toxicity and carcinogenicity at low exposures.12,13 Cd is naturally present in soils, sediments, seawater, plants, and animals at low to undetectable background levels. Anthropogenic Cd is extracted during zinc (Zn), lead (Pb), and copper (Cu) ore processing and used to manufacture NiCd batteries, pigments/ coatings, photovoltaic devices, and other products.14 U.S. Cd production decreased from 1190 to 745 tons during 1999−2008, spiking to over 1480 tons in 2004−2005.15,16 Diet and tobacco smoke are considered the primary sources of exposure in the general population. Soil, water, air, and fish/ shellfish may be additional sources, particularly near contaminated sites, fossil fuel combustion sources, waste incinerators/landfills, © 2012 American Chemical Society
Received: Revised: Accepted: Published: 1137
September 4, 2012 December 11, 2012 December 19, 2012 December 19, 2012 dx.doi.org/10.1021/es303556n | Environ. Sci. Technol. 2013, 47, 1137−1147
Environmental Science & Technology
Article
income-to-poverty (PIR), race/ethnicity, country of origin, and education (age 20−85 only). We created three PIR categories: “low” (PIR < 1.3; eligible for supplemental nutrition assistance), 1.3 ≤ PIR ≤ 3.5, and “high” (PIR > 3.5, defined high income by CDC).34,35 We considered participants born in Mexico or elsewhere as “born outside U.S.” and created race/ethnicity/ origin categories for the three race/ethnicities represented by NHANES (Mexican American, Non-Hispanic Black, NonHispanic White). We included a category for “Other Hispanic” or “Other Race” but do not report these results. We collapsed the NHANES education categories into categories reflecting highest educational attainment: not high school graduate, high school graduate, and college graduate. We examined the effect of NHANES 6-month period (May− October vs November−April) based on evidence for seasonal differences in Pb absorption, although Cd evidence is limited.36 We also examined the effect of living in a home built before vs after 1978, based on evidence that elevated house dust Cd is associated with elevated urine and blood Cd and our assumption that older homes may accumulate more dust.37 Teens and adults were administered different questionnaires on smoking, and questionnaires changed in later survey years. For adults, we defined smokers in NHANES 1999−2004 as those with serum cotinine >10 ng/mL19 and answering yes to “Do you now smoke cigarettes?” or everyday/some days to “Do you now smoke a pipe/cigar(s)?” In 2005−2008, smokers had cotinine >10 ng/mL, or answered everyday/some days to “Do you now smoke cigarettes?” or 1−5 days to “During the past 5 days...on how many days did you smoke a pipe/cigar?” or ≥1 to “During the past 5 days...how many pipes/cigars did you smoke each day?” In 1999−2004, nonsmokers had cotinine LOD). Cotinine LODs were 0.05 ng/mL in 1999−2000 and 0.015 ng/mL in 2001−2008.
while blood levels generally indicate recent exposures (e.g., weeks−months).14 In addition to age, diet, and smoking, gender (female) is associated with higher Cd body burdens.19−23 This is partly attributed to lower Fe stores in women, leading to increased absorption.16,24−26 When gender, age, smoking, and other covariates were controlled, body mass index (BMI) was inversely associated with urinary Cd in a Flemish study of girls and women and in studies of adults in the 1999−2002 and 1988−2008 U.S. National Health and Nutrition Examination Survey (NHANES).17,27,28 In U.S. studies, race/ethnicity is also associated with Cd body burdens. Tellez-Plaza et al. found higher urinary Cd in NonHispanic Black vs Non-Hispanic White adults in NHANES 1988−1994 but not 2003−2008, while Paschal et al. found the highest levels among Mexican Americans in NHANES 1988− 1994.19,21 For NHANES 1999−2006 data, Mijal and Holzman found the highest blood Cd in Mexican Americans among nonsmoking women (age 20−44).29 In some U.S. studies, being born abroad is associated with higher Cd. A 2004 study of New York City adults found higher mean blood Cd in foreign-born Chinese residents than in U.S.-born smokers, despite similar smoking rates between China- and U.S.-born residents.19 Cd is also associated with education and income in U.S. studies, although evidence for the latter is mixed. Mijal and Holzman and McKelvey et al. reported higher blood Cd in participants with fewer years of education.19,27 McKelvey et al. found the highest blood Cd in the lowest household income group, while a 2004−2005 study of Wisconsin women found the highest urinary Cd in the highest income group.2,19 Given the potential adverse health outcomes associated with low-level Cd exposure, it is useful to conduct periodic evaluations of body burdens and the factors contributing to them. We used multiple regression analyses to investigate factors associated with urinary Cd for children (age 6−11), teens (age 12−19), and adult (age 20−85) nonsmokers and smokers in NHANES 1999− 2008.30 We analyzed child, teen, and adult subpopulations separately based on knowledge that their diet, body mass, and other exposure and toxicokinetic factors differ. Our objectives were to test whether mean urinary Cd levels declined significantly over the decade in these subpopulations and, if so, to determine major factors associated with the decline(s).
■
METHODS NHANES Urinary Cd Data. Detailed survey documentation is available at the NHANES Web site.28 Briefly, urinary Cd was measured for a randomly selected subsample of participants age ≥6 using inductively coupled plasma-mass spectrometry.31−33 The limit of detection (LOD) was 0.06 ng/mL in 1999−2002 and for some 2003−2004 samples, and 0.04 ng/mL in 2005− 2008 and some 2003−2004 samples. CDC applied an adjustment factor to the 1999−2002 data to account for interference from molybdenum (Mo) oxide formed in the instrument from Mo normally present in urine. We used CDC’s adjustment equations to calculate alternative LODs for adjusted observations (Supporting Information (SI)). In 2003−2004, CDC upgraded the analytical procedure eliminating the need for adjustment.29,30 Covariate Selection. We derived relevant covariates from NHANES questionnaire, examination, and laboratory files (Table 1). Occasionally, CDC releases updated/corrected data; covariate data we used were current as of July 2012. We treated “refused” and “don’t know” responses as missing. Demographic covariates we examined included age, gender, ratio of family 1138
dx.doi.org/10.1021/es303556n | Environ. Sci. Technol. 2013, 47, 1137−1147
current, former, nonsmoker
highest education level completed (not high school graduate, high school graduate, college graduate)
smoking statusg,k
demographic questionnairea
a NHANES 1999−2008 Demographic data and documentation available http://www.cdc.gov/nchs/nhanes/nhanes2007-2008/ DEMO_E.htm, http://www.cdc.gov/nchs/nhanes/nhanes2005-2006/ DEMO_D.htm, http://www.cdc.gov/nchs/nhanes/nhanes2003-2004/DEMO_C.htm, http://www.cdc.gov/ nchs/nhanes/nhanes2001-2002/DEMO_B.htm, http://www.cdc.gov/nchs/nhanes/ nhanes1999-2000/DEMO.htm [all accessed 28 Feb 2011]. bBody measures data and documentation available http://www.cdc.gov/ nchs/nhanes/nhanes2007-2008/BMX_E.htm, http://www.cdc. gov/nchs/data/nhanes/nhanes_05_06/bmx_d.pdf, http://www.cdc.gov/ nchs/nhanes/nhanes2003-2004/BMX_C.htm, http://www.cdc.gov/nchs/nhanes/nhanes2001-2002/BMX_B.htm, http://www. cdc.gov/ nchs/nhanes/nhanes1999-2000/BMX.htm [all accessed 28 Feb 2011]. cHousing data and documentation available http://www.cdc.gov/nchs/nhanes/nhanes2007-2008/HOQ_E.htm, http:// www.cdc.gov/nchs/data/nhanes/nhanes_05_06/hoq_d.pdf, http://www.cdc.gov/nchs/data/nhanes/nhanes_03_04/hoq_c.pdf, http://www.cdc.gov/nchs/data/nhanes/nhanes_01_02/ hoq_b_doc.pdf, http://www.cdc.gov/nchs/nhanes/nhanes1999-2000/HOQ.htm [all accessed 7 Mar 2011]. dDietary recall - total nutrients files available http://www.cdc.gov/nchs/nhanes/nhanes2007-2008/DR1TOT_ E.htm, http://www.cdc.gov/nchs/data/nhanes/ nhanes_05_06/ dr1tot_d.pdf, http://www.cdc.gov/nchs/nhanes/nhanes2003-2004/DR1TOT_C.htm, http://www.cdc.gov/nchs/nhanes/ nhanes20012002/DRXTOT_B.htm, http://www.cdc.gov/nchs/nhanes/nhanes1999-2000/DRXTOT.htm [all accessed 9 Mar 2011]. eDietary recall - individual foods files available http://www.cdc.gov/nchs/nhanes/ nhanes2007-2008/DR1IFF_E.htm, http://www.cdc.gov/nchs/data/nhanes/nhanes_05_06/dr1iff_d.pdf, http://www.cdc.gov/nchs/nhanes/nhanes2003-2004/ DR1IFF_C.htm, http://www.cdc.gov/ nchs/nhanes/nhanes2001-2002/DRXIFF_B.htm, http://www.cdc.gov/nchs/nhanes/nhanes1999-2000/DRXIFF.htm [all accessed 9 Mar 2011]. fDietary supplements data and documentation available http://www.cdc.gov/nchs/ nhanes/nhanes2007-2008/DS1TOT_E.htm, http://www.cdc.gov/nchs/data/nhanes/nhanes_05_06/dsq_d.pdf, http://www.cdc.gov/ nchs/data/nhanes/nhanes_03_04/ dsq_c.pdf, http://www.cdc.gov/nchs/data/nhanes/nhanes_01_02/dsq_b_doc.pdf, http://www.cdc.gov/nchs/nhanes/nhanes1999-2000/DSQFILE2.htm [all accessed 9 Mar 2011]. gSerum cotinine data and documentation available http://www.cdc.gov/nchs/nhanes/nhanes2007-2008/COTNAL_E.htm, http://www.cdc.gov/nchs/nhanes/ nhanes2005-2006/COT_D.htm, http://www.cdc.gov/ nchs/nhanes/nhanes2003-2004/L06COT_C.htm, http://www.cdc.gov/nchs/ nhanes/nhanes2001-2002/L06_B.htm, http://www.cdc.gov/nchs/nhanes/nhanes1999-2000/LAB06.htm [all accessed 9 Mar 2011]. hFamily smoking data and documentation available http://www.cdc.gov/nchs/nhanes/nhanes2007-2008/SMQFAM_E.htm, http:// www.cdc.gov/nchs/data/nhanes/nhanes_05_06/ smqfam_d.pdf, http://www.cdc.gov/nchs/data/nhanes/nhanes_03_04/smqfam_c.pdf, http://www.cdc.gov/nchs/data/nhanes/nhanes_01_02/smqfam_b_doc.pdf, http:// www.cdc.gov/nchs/nhanes/ nhanes1999-2000/ SMQFAM.htm [all accessed 9 Mar 2011]. iUrine metals data and documentation available http:// www.cdc.gov/nchs/nhanes/ nhanes2007-2008/UHM_E.htm, http://www.cdc.gov/ nchs/nhanes/nhanes2005-2006/UHM_D.htm, http://www.cdc.gov/nchs/nhanes/ nhanes2003-2004/L06HM_C.htm, http://www.cdc.gov/nchs/nhanes/nhanes2001-2002/L06HM_B.htm, http:// www.cdc.gov/nchs/ nhanes/nhanes1999-2000/LAB06HM.htm [all accessed 9 Mar 2011]. jBiochemistry data and documentation available http://www.cdc.gov/nchs/nhanes/nhanes2007-2008/ BIOPRO_E.htm, http://www.cdc.gov/nchs/nhanes/nhanes2005-2006/ BIOPRO_D.htm, http://www.cdc.gov/nchs/nhanes/nhanes2003-2004/L40_C.htm, http://www.cdc.gov/nchs/nhanes/ nhanes2001-2002/L40_B.htm, http://www.cdc.gov/nchs/nhanes/nhanes1999-2000/LAB18.htm [all accessed 9 Mar 2011]. kSmoking questionnaire data and documentation available http://www.cdc. gov/nchs/nhanes/ nhanes2007-2008/SMQRTU_E.htm, http://www.cdc.gov/nchs/nhanes/nhanes2007-2008/SMQ_E.htm, http://www.cdc.gov/nchs/data/nhanes/nhanes_05_06/ smqrtu_d.pdf, http://www.cdc.gov/nchs/nhanes/nhanes2005-2006/SMQ_D.htm, http://www.cdc.gov/nchs/data/nhanes/ nhanes_03_04/smqmec_c.pdf, http://www.cdc.gov/nchs/nhanes/nhanes2003-2004/ SMQ_C.htm, http://www.cdc.gov/ nchs/data/nhanes/nhanes_01_02/smqmec_b_doc.pdf, http://www.cdc.gov/nchs/nhanes/nhanes2001-2002/SMQ_B.htm, http://www.cdc.gov/nchs/nhanes/ nhanes1999-2000/SMQMEC.htm, http://www.cdc.gov/nchs/nhanes/nhanes1999-2000/SMQ.htm [all accessed 9 Mar 2011].
age 20−85 only
serum albumin (g/L), serum Fe (ug/dL), serum P (mg/dL)
biochemistry profilej
ages 12−19, 20−85 only
urine Cd (ng/mL), urine creatinine (mg/dL)
urine metalsi
number of smokers in home (0, 1, 2, >3)
serum cotinine (ng/mL)
serum cotinineg
family smoking questionnaire
any dietary supplements taken past month (yes/no)
diet supplements questionnairef
h
total nutrients files:d previous 24-h total intake of: Ca (mg), calories (kcal), fat (g), Fe (mg), Mg (mg), P (mg), protein (g), Se (mg), Zn (mg); individual foods files:e previous 24-h total intake (g) of any foods with highest Cd mean and/or maximum levels in FDA Total Diet Study 1999−2005 (asparagus, beef liver, collards, lettuce, peanuts, potato chips, shrimp, spinach, sunflower seeds)
home built before 1978 (yes/no), tap water source (water company/other), use of water treatment devices (yes/no)
dietary recall interview (24 h)
housing questionnaire
BMI (kg/m2)
c
age (years), born in U.S. (yes/no), gender, NHANES survey year, NHANES survey 6-month period (May−Oct/Nov−Apr), race/ethnicity, ratio of family income to poverty (PIR)
variable(s)
body mass indexb
all age groups
demographic questionnairea
NHANES data set
Table 1. Variables Tested in Multiple Regression Models of Log Urinary Cd
Environmental Science & Technology Article
1139
dx.doi.org/10.1021/es303556n | Environ. Sci. Technol. 2013, 47, 1137−1147
Environmental Science & Technology
Article
Next, we estimated models with the survey year dummy variables and all covariates to assess the magnitude, directionality, and statistical significance of each covariate. The survey year effects captured systematic changes in Cd averages not picked up by other covariates. The urinary Cd GSDs were allowed to differ across survey years. To preserve sample sizes, covariates with >10% missing values (PIR; home built before 1978; cotinine; smoking status; serum albumin, Fe, and P) were modeled as follows: we coded missing values as zero in the original variable and introduced a dummy variable set to one if the original value was missing, zero otherwise. This allowed estimation of the relationship between Cd and the severely missing covariate, controlling for potential effects of missing observations. We evaluated the bias potentially associated with using the missing data indicators by rerunning the regressions with only complete cases. In general, this sensitivity analysis showed the regression results to be robust to our choice of missing data handling method. We also conducted sensitivity analyses using the original model statements and only the 2003−2008 data to see whether CDC’s Mo adjustment of the 1999−2002 data affected results. In general, the 2003−2008 analyses showed similar patterns in the explanatory variables and similar downward trends as the full 1999−2008 analyses. To evaluate whether any covariate contributed to observed Cd trends, we analyzed covariate trends by estimating survey year-specific covariate means and testing for significant trends by subpopulation, focusing on year-to-year differences and comparisons to 1999−2000. We combined the coefficients of the multiple regression models with the survey year-specific covariate means to decompose the overall change in Cd GMs into components driven by changes in the covariates. Specifically, suppose X and Z are two covariates in a model of log urinary Cd (Y):
NHANES includes a dietary recall interview for all food/ beverages consumed 24 h ending midnight before urine collection. Twenty-four-hour intakes do not necessarily reflect longer-term consumption patterns, which may be more predictive of urinary Cd than short-term intakes. However, we felt it was important to examine associations between urinary Cd and the available NHANES data on consumption of foods potentially containing Cd. We downloaded data from the 1999− 2008 Total Diet Study (TDS), a periodic survey of contaminants in the U.S. food supply.17 We designated foods with the highest mean (detects only) and maximum Cd concentrations during 1999−2008 as “Cd-foods” (sunflower seeds, spinach, lettuce, shrimp, beef liver, asparagus, collards, peanuts, potato chips). We created a variable indicating whether a participant reported eating ≥1 of these foods. We used reported creatinine levels on the right-hand side of the regressions to control for urinary dilution. CDC is investigating the use of specific gravity in addition to creatinine to control for urinary dilution;38 specific gravity measurements were not available for NHANES 1999−2008. To examine effects of micronutrient status, we used 24-h intakes of Fe, Zn, and other nutrients (Mg, P, Se), assuming these were reasonable proxies for longer-term status. Serum Fe and P measurements were available for teens and adults, so we included these in extended analyses. Serum albumin measurements were also available for teens and adults, so we included these in the extended analyses to control for serum Cd binding capacity.39 Statistical Analysis. We modeled urinary Cd using a lognormal distribution; the survey year-specific distributions appeared consistent with this assumption, except for the child 2005−2006 and 2007−2008 and teen 2005−2006 distributions, which had heavier right tails than expected. Because the Cd LOD varied, we used a variable-threshold censored regression approach to account for 94%), and lowest for children (66−95%, by survey year). Children’s GMs were one-fourth to one-third those of adult nonsmokers while teens’ were one-half of GMs for adult nonsmokers. GMs of adult smokers were approximately 40−70% higher than those of nonsmokers. GMt/GMt−1 and GMt/GM0 ratios showed downward trends for all subpopulations, with statistically significant trends in later years for children, teens, and nonsmoking adults.Ninety-fifth percentile ratios (Table S1) also showed downward trends in later years for teens and nonsmoking adults. Downward trends persisted for teens and adults in the multiple regressions (Table 3). Creatinine was significantly associated with Cd in all subpopulations, while age was significant only for adults. BMI was inversely associated with Cd, independent of gender, for children, teens, and adult smokers. The different signs of the age and BMI coefficients suggest effect modification; although we did not test this explicitly, the age−BMI−urinary Cd relationship may merit further exploration. Mexican Americans born outside U.S. had significantly higher Cd than the referent category (Non-Hispanic White, born in U.S.) among children and adults, while Mexican Americans born in U.S. had higher Cd than the reference category among children, teens, and adult nonsmokers. Non-Hispanic Blacks born outside U.S. had higher Cd levels among teens and adult nonsmokers, while those born in U.S. had higher levels among teens compared to the reference category. Adult smokers who did not finish high school had higher Cd than those who did, while nonsmoking college graduates had lower levels than nonsmoking high school graduates. Of the dietary covariates, 24-h calorie intake was inversely associated with Cd in adult nonsmokers, while 24-h fat and Mg intakes were positively associated. Twenty-four-hour Fe was inversely associated with Cd in adult smokers, while dietary supplements were inversely associated in teens. Adult smokers living with other smokers had higher Cd than those living with nonsmokers. Serum cotinine was also associated with higher Cd in adult smokers but not nonsmokers. Among adult nonsmokers, those with undetectable cotinine had lower Cd than those with detectable cotinine. Table S2 shows results of the extended analyses. Serum albumin was inversely associated with Cd in adults, and serum P was inversely associated in adult nonsmokers. Other results were similar to the original with these exceptions: downward Cd trends were no longer significant for adult nonsmokers, while declines relative to 1999−2000 became significant in almost all survey years for adult smokers. Age became significant in the extended teen analysis, as did belonging to the Mexican American, born-outside-U.S. category. Generally, most covariates did not exhibit significant trends relative to 1999−2000 (Table S3); that is, there were no statistically significant differences in covariate survey year averages relative to their 1999−2000 averages (data not shown), with several exceptions. For teens and adults, serum albumin decreased and serum P increased in all survey years relative to 1999−2000, while 24-h Ca and fat intakes increased in later years. For adult smokers, 24-h Mg and P intakes increased in later years, while 24-h protein intake increased in 2005−2006.
For teens and nonsmoking adults, the proportion of cotinine nondetects increased most years. Figure 1 shows the overall % decline in GM Cd from 1999− 2000 to 2007−2008 for each subpopulation as well as the percent attributable to different covariate categories. While overall declines were significant for each subpopulation, no covariate category helped explain them. Instead, other unidentified timerelated factors appeared to explain large fractions of the declines among teens and adults.
■
DISCUSSION Public Health Implications. We found significant declines in urinary Cd among U.S. children, teens, and adults over the past decade. From 1999 to 2008, children’s GM Cd decreased 25% (Table 2) while the 95th percentile (Table S1) decreased 31%. For teens, the GM and 95th percentile decreased 25% and 33%, respectively. For adult nonsmokers, the GM and 95th percentile decreased 20% and 20%, respectively. For smokers, the GM and 95th percentile decreased 23% and 14%, further illustrating the downward trend for U.S. adults since 1988 reported by Tellez-Plaza et al.19 These declines cannot be fully explained by measurement variability, which likely did not exceed 10%the precision target of the CDC laboratory that analyzed the samplesin any given year.45 Moreover, we would expect the direction of any measurement variability (not attributable to instrument bias) to vary randomly, not to result in the steady declines observed in the NHANES data. Although the modest declines are good news, epidemiologic studies controlling for smoking and other risk factors indicate that millions of Americans are still potentially at risk of adverse outcomes associated with low-level Cd exposure. Menke et al. followed NHANES 1988−94 adults until 2000, estimating an all-cause mortality hazard ratio of 1.68 (95%CI 1.09−2.58) in men with the highest tertile urinary Cd (≥ 0.48 μg/g) versus the lowest.6 In NHANES 2007−2008, approximately 36 million men had urinary Cd ≥0.48 μg/g. Everett and Frithsen found an 80% increased risk (odds ratio [OR] 1.80, 95%CI 1.06−3.04) for myocardial infarction among NHANES 1988−1994 women 45−79 years old with urinary Cd ≥0.88 versus
