NHANES-UCMR Data Merge: Some Initial Observations

Supplemental Information:

Estimating Perchlorate Exposure from Food and Tap Water Based on U.S. Biomonitoring and Occurrence Data

David Huber, MSUS EPA, Office of Ground Water and Drinking Water

Benjamin Blount, PhD Centers For Disease Control and Prevention

David Mage, PhD Danya International, Inc.

Frank Letkiewicz, BS The Cadmus Group

Amit Kumar, MBBS, MPHThe Cadmus Group

Ruth Allen, PhD, MPHUS EPA Office of Prevention, Pesticides and Toxic Substances

The findings and conclusions in this report are those of the authors and do not necessarily represent the views of the Environmental Protection Agency or the Centers for Disease Control and Prevention
Methods for creatinine-based perchlorate dose estimates

The amount of perchlorate in a 24-hour urine sample would approximate the amount of perchlorate ingested over a 24-hour time frame since virtually all perchlorate is excreted via this pathway . However, since a 24-hour urine sample is not collected in NHANES, an alternate means was used to estimate the daily perchlorate dose from the single urinary concentration values available from NHANES.

Assuming an omnivorous diet and normal renal function, we estimated 24-hour creatinine excretion Urinary perchlorate in a single urination event may be expressed as µg of perchlorate excreted per µg of creatinine excreted. The ratio of perchlorate urinary concentration to the creatinine urinary concentration is equivalent to the ratio of the daily perchlorate intake to the daily creatinine excretion amount:

The NHANES data provide the urinary concentrations of perchlorate and creatinine for subjects, and the daily creatinine excretion rates for those individuals are then calculated from the relationships described subsequently here. From those, the daily perchlorate intake can be calculated using the ratios show above. Because of the varied and episodic nature of perchlorate exposure and the relatively short physiological half life of perchlorate (8 hrs), spot urine perchlorate levels are likely to vary throughout the day. Thus dose estimates based on spot urine samples may be more skewed than actual 24-hr perchlorate doses. This daily perchlorate intakecan be expressed on a mass basis (µg/day) or on a per body weight basis (µg/kg/day), the latter being the most useful for direct comparison to the EPA reference dose (RfD).

Thecreatinine-based approach for estimating intake of various substances from urinary concentration measurements was originally developed by Cockroft and Gault [1], and modified by Mage et al. [2] and was utilized by Blount et al. [3] to estimate perchlorate dose for adults using NHANES data. We used a method further modifiedby Mage et al. [4] and subsequent refinements to estimate perchlorate exposure doses in the U.S. populationfrom the NHANES data set. Creatinine correction equations published for all ages were further adjusted to account the lean body mass and pregnancy.These further adjustments are required because adipose tissue does not produce creatinine, and thus obesity leads to an over-estimation of 24-hr creatinine excretion and perchlorate dose. By adjusting the data to account for varying body composition we reduce error from this source.

Comparative analyses were conducted for subgroups (by age, sex, race/ethnicity, fasting status or other factors) as well as for the overall categories. Using data from the publicly available NHANES file for the 2,820 subsamples with perchlorate and creatinine urinary levels, together with the expanded set of creatinine excretions equations, we estimated perchlorate doses across ages 6 and older. The number of subjects in the tables includes only those participants who had adequate information on the variables of interest.

The following are Cockroft-Gault equations for creatinine adjustment and refinements:

Creatinine Excretion = 10-6 * k * (140-Age) * Weight1.5 * Height0.5

where the Adult Creatinine Excretion rate is in g/day, age is in years, weight is in kg, height is in cm and k = 1.93 for males and 1.64 for females. The perchlorate daily intake is then calculated for each individual from:

Perchlorate Intakeg/kg-day) =[Urinary Perchlorate g/L)/Urinary Creatinine g/L)]*

[Creatinine Excretion (g/day)/Weight(kg)]

Summary of Mage et al. [2, 4] adjustments used in this investigation:

Creatinine Corrections for Adiposity

Correction for creatinine excretion in the obese may be made using estimates of their lean body mass (LBM). The LBM may be computed from weight and weight while submerged in water, or by bioelectric impedance analysis [5]. It may also be estimated by use of charts and tables that predict the ideal weight-for-height [6]. Eddy (cited in James [7])proposed equations 1 and 2 for predicting LBM (total weight – weight of body fat) from measurements of U.K. adult male and female weight and body-mass-index, respectively:

Male LBM (kg) = Wt (1.101 – 0.01281BMI)(1)

Female LBM (kg) = Wt (1.07 – 0.0148 BMI)(2)

Equations 1 and 2 can be used to correct for LBM in NHANES subjects given their age, gender, height and weight. From Kuczmarski et al. [8] the median U.S. BMI for males and females are 23.0 and 21.7, respectively at their full-growth at age 20 years. Note that equations 1 and 2 are consistent in that they predict the same LBM for male and female at a BMI of 15.6 whichis the median value for 8 years of age at which the muscular development of children of both sexes are similar.

As a first approximation we propose to reference the subject’s LBM to that of the median BMI subject for the same Wt. That is, we will assume that the Cockcroft and Gault creatinine equations are valid for the median BMI subject and that the LBM can be scaled in reference to the median individual:

Let an adult male have a Wt of 100 kg and a BMI = 30. The reference is for a male of 100 kg and BMI of 23.

Reference LBM = 100 [1.101 – 0.01281 (23)] = 80.6 kg LBM(3)

The subject LBM = 100 [1.101 – 0.01281(30)] = 71.7 kg LBM(4)

Therefore the subjects creatinine production would be (71.7/80.6) = 0.89 or 89% of the creatinine predicted for the median BMI adult subject of identical weight.

Thus, we would reduce the creatinine predicted in our previous work by 11% for such an individual.

Note that for thinner male individuals than the median (BMI < 23) the multiplier would be greater than 1.

Repeating the same for the adult females at 100 kg with BMI = 30 referred to a median BMI = 21.7.

Reference LBM = 100 [1.07 – 0.0148 (21.7)] = 74.9 kg LBM(5)

The subject LBM = 100 [1.07 – 0.0148(30)] = 62.6 kg LBM(6)

Therefore, the creatinine production is reduced by a factor of (62.6/74.9) = 0.836 or 83.6%

In summary, our creatinine correction equations would have to be multiplied by the following terms for adult males and females to correct for their adiposity.

Male: (1.101 – 0.01281 BMI) / 0.806 = 1.366 – 0.0159 BMI(7)

Female: (1.07 – 0.0148 BMI) / 0.749 = 1.429 – 0.0198 BMI (8)

If we assume that the same general relationship exists during the period when full-growth is not yet attained (< 20 years) we can make a similar correction for each age up to 19 using the median BMI as a function of age and gender shown in Exhibit 1 (from [8]). For example, the correction at age 6 years for males and females are as follows:

Male 6 yr = (1.101 – 0.01281BMI)/ (1.101 – 0.01281[15.4] )

Correction for Male 6 yr = 1.22 – 0.0142 BMI

Female 6 yr = (1.07 – 0.0148 BMI) / (1.07 – 0.0148[15.2] )

Correction for Female 6 yr = 1.27 -0.0175 BMI

Summary of Assumptions and Caveats

The following assumptions are either implicit or explicit in our approach to creatinine prediction as a function of age, gender, height, weight and derived BMI [Wt kg/ (Ht m)2]:

1. The CDC 2000 BMI median data for all U.S. children ages 6 – 20 are similar to the median data for the German children reported by Remer et al. [9], and the U.S. Children reported by Viteri and Alvarado [10]..

2. The average creatinine values (mg/day) reported as a function of height for children apply to the children of median BMI for the given height.

3. The Eddy equations 1 and 2, cited by James [7] for U.K. adult males and females apply to the Canadian adult males and females studied by Cockcroft and Gault [1] of similar BMI as their equations did not extend into the obese range (BMI > 30).

4. The Eddy equations 1 and 2 for male and female adults can be extrapolated into the age range for children for relating LBM to the LBM at the median LBM. We recognize that the absolute value of child LBM as a function of BMI is probably inaccurate, but that the ratio of LBM for a given BMI at age and height to the LBM for the median BMI at age and height may be a better representative of the actual ratio.

Male correction = [Weight/(14 + 1.433 (Age – 3) + 0.22 (Age – 3) 2 - 0.00533 (Age – 3) 3)] 0.5 (13)

Female correction = [Weight/(14 + 0.40 (Age – 3) + 0.52 (Age – 3) 2 – 0.024 (Age – 3) 3)] 0.5 (14)

The equations to be used are the fundamental creatinine excretion equations that predict mg/day as a function of age, gender, height and weight. They incorporate 3 adjustments:

1) Black race: This increases LBM and creatinine production for the same age, gender, height and weight as non-black individuals. These equations are in Exhibit 2.

2) Body surface area: This increases heat loss requiring more metabolic energy to maintain body temperature. These equations 13 and 14 are footnoted to Exhibit 2

3) Obesity (Adipose tissue): This decreases LBM and creatinine production for the same age, gender, race, height and weight. The equations require either Table lookup to Exhibit 1 for the median BMI at the age and gender of the children in NHANES (6–18) or a cubic equation fit to those data.

Example of male adult equations for mg creatinine per day:

0.926*1.93 [140 – Age] (Wt 1.5 Ht 0.5 ) (1 + 0.18 B)/1000 From Exhibit 2 showing all correction factors including race and BSA.

0.00179 [140 – Age] (Wt 1.5 Ht 0.5 ) (1 + 0.18 B) (1.366 – 0.0159 BMI) with Equation 7 for obesity (terms multiplied out)

Example of male child equations (age 6, < 168 cm height) for mg creatinine per day.

1.085 Ht [6.265 + 0.0564 (Ht – 168)] From Table 2 (Black race is not a factor under 15 years)

1.085 Ht [6.265 + 0.0564 (Ht – 168)] [Wt/(14 + 1.433 (Age – 3) + 0.22 (Age – 3) 2 - 0.00533 (Age – 3) 3)] 0.5 Correcting for BSA

1.085 Ht [6.265 + 0.0564 (Ht – 168)] [Wt/(14 + 1.433 (Age – 3) + 0.22 (Age – 3) 2 - 0.00533 (Age – 3) 3)] 0.5 (1.22 – 0.0142 BMI) Correcting forobesity

Discussion of Results

We found a decrease of 19% in urinary creatinine concentration in pregnant compared with non-pregnant women in the 2001-2002 NHANES data, consistent with decreased serum creatinine concentrations in pregnancy [11]. Glomerular filtration rate (GFR) and serum creatinine are inversely related. The GFR increases in pregnancy as demands on the kidneys increase along with the blood volume. Creatinine clearance is used to estimate GFR as an indicator of kidney health, and Cockroft-Gault type equations are used to estimate 24 hour creatinine clearance. Inserting a glomerular filtration rate to predict creatinine clearance (even within the range typical of pregnant women) may introduce additional error. Creatinine clearance used to report GFR systematically overestimates it because creatinine is secreted in the proximal tubule, not only the glomerulus [12, 13]. Thus, it might be that using creatinine clearance to estimate GFR, and using GFR to predict the creatinine clearance in a pregnant woman is tenuous. Davison and Noble [14] tabulated creatinine clearance changes throughout the phases of the menstrual cycle. These are similar in magnitude to the changes in the first trimester of pregnancy and changes documented in the third trimester as well [13]. These changes present opportunities for further analysis of the data; we did not attempt to correlate creatinine excretion and time of month or week of pregnancy in this study.

The issue of whether women gain or lose lean body mass during pregnancy is also complex and likely to varyby individual and length of gestation, so attempting to calculate the creatinine excretion without lean body mass and urine volume is problematic. The calculated increase in perchlorate dose logically correlates with overall caloric intake, although the reported servings of higher perchlorate foods (fruits, vegetables and milk) did not correlate with the four individuals with the highest perchlorate values. The adjustment in perchlorate dose taking into account the adiposity correction was more apparent in general among women who apparently had relatively more adipose tissue than men or children as seen in the differences between Exhibits 2and 3 where the calculated dose of perchlorate decreased.

Exhibit 1: Median BMI by Age and Gender for <20 years of age

(prior to full growth).

Exhibit 2: Proposed sets of creatinine excretion equations for healthy, non-obese, Black and White males and females, from birth through adulthood (92 years). #

* Estimated from totals of 28 male and 22 female of age 14 – 18. Height breakout for male ages not available.

** Male creatinine (Daniels and Hejinian, 1929) assumed identical to the female by Viteri and Alvarado (1970).

*** 51 children of unspecified gender. Subtotals estimated by the 5% male excess birth rate.

# See Equations 13 and 14 for BSA corrections when child’s weight (kg) is not the median weight for age.

B = 1 if Black, B= 0 if not Black. Modified to match at 3 and 18 years at median heights and weights for U.S. children.

1

REFERENCES

[1]Cockroft, D.W., and Gault, M.H. Prediction of Creatinine Clearance from Serum Creatinine. Nephron 1976: 16:31-41

[2]Mage, D.T., Allen R. H., Gondy G., Smith W., Barr, D. B., and Needham, L.L. Estimating Pesticide Dose from Urinary Pesticide Concentration Data by Creatinine Correction in the Third National Health and Nutrition Examination Survey (NHANES-III). J Expo Anal Environ Epidemiol 2004;14: 457–465.

[3]Blount, B.C., Valentin-Blasini, L., Osterloh, J.D., Mauldin, J.P., and Pirkle, J.L. 2007. Perchlorate Exposure of the US Population, 2001-2002. J Expo Sci Environ Epidemiol. 2007 Jul 17: 400-7. Epub 2006 Oct 18.

[4]Mage, D. T., Allen, R. H, Kodali,A. Creatinine Corrections for Estimating Children’s and Adult’s Pesticide Intake Doses in Equilibrium with Urinary Pesticide and Creatinine Concentrations. J Expo Anal Environ Epidemiol. 2007 Sep 19; [Epub ahead of print].

[5]Segal, K. R., Van Loan, M. D, Fitzgerald, P. I, Hodgdon, J.A, Van Itallie, T.B. Lean body mass estimation by bioelectrical impedance analysis: a four-site cross-validation study. Am J Clin Nutr 1988;47:7–14.

[6]Stavig, G.R., Leonard A.R., Igra, A., Felten, P1984 Indices of relative body weight and ideal weight charts. J Chronic Dis 37:255–262

[7]James, W. P. T., 1976. DHSS/MRC Report on research on obesity. London, HMSO, p.9.

[8]Kuczmarski et al., CDC Growth Charts, Vital and Health Statistics no. 314, NCHS, 2000

[9]Remer, T., Neubert A., Maser-Gluth, C., Anthorpometry-based reference values for 24 h urinary creatinine excretion during growth and their use in endocrine and nutritional research. Am J Clin Nutr. 2002 Mar: 74 (3): 561-9.

[10]Viteri, F.E., and Alvarado, J., The creatinine height index: its use in the estimation of the degree of protein depletion and repletion in protein calorie malnourished children. Pediatrics. 1970. Nov: 46(5): 696-706

[11]Kristensen, K., Lindström, V., Schmidt, C., Blirup-jensen, S., Grubb, A., Wide-swensson, D.and Strevens, H. 'Temporal changes of the plasma levels of cystatin C, β-trace protein, β2-microglobulin, urate and creatinine during pregnancy indicate continuous alterations in the renal filtration process', Scandinavian Journal of Clinical and Laboratory Investigation, 2007; 1 – 7

[12]Stevens, L.A., Coresh, J., Greene, T., Levey, A. Assessing Kidney Function—Measured and Estimated Glomerular Filtration Rate N. Engl J. Med June 2006;354:23.

[13] Davison, J.M.., et al. 24 Hour Creatinine Clearance During The Third Trimester Of Normal Pregnancy. British J. of Ob and Gyn. Feb 1980, p 106-109

[14]Davison, J. M.., and Noble, M.C.B. Serial Changes In 24 Hour Creatinine Clearance During Normal Menstrual Cycles And The First Trimester Of Pregnancy. British J. of Ob and Gyn Jan 1981; pp 10-17.

1