Document yrEDdRzbXwpKEdjxQnz8LQ00D

"ASSESSMENT OF BLOOD LEAD LEVELS IN THE USA FROM NHANES ! I DATA", J. M. Pierrard, C. G. Pfeifer, and R. D. Snee E, I. du Pont de Nemours & Co., Inc. Internationa! Conference on Heavy Metals in the Environment Heidelberg, Germany, 6-9 September 1983 TEH 0532522 ASSESSMENT OF BLOOD LEAD LEVELS IN THE U.S,A. FROM KHANESII DATA John M Pierrard*, Charles C Pfeifer* and Ronald D Suae* ABSTRACT Changing demographics of the subject groups at the 64 sites sampled dur ing NHANESII accounts for over half the apparent 5.61 ug/dl blood lead decrease from 1976 to 1960. Site-specific gasoline lead exposure accounts for 0.51 ug/dl of the blood lead decrease* In agreement with the change expected due to the decrease of 0.32 ug/s>3 in average air lead and the accepted range 1 2 for blood lead response to air lead change. INTRODUCTION The U.S. Government's second National Health and Nutritional Examination Survey (NHANESII) was a cross-sectional study designed to assess national health status. Venous blood lead levels (BPb} were measured for 9*936 of the 20*333 examinees. Mobile caravan teams visited 64 sites in the continguous U.S.A. and Hawaii between 1976 and 1980, A preliminary analysis reported s 36.72 reduction In BPb over the four-year study period end noted that the decrease In BPb reflects the decrease in national use of lead In gasoline production (ref 1). Because the NHANESII sample was not selected to Insure valid within-study time analyses* several categories of demographic variables were evaluated in the present study to examine their effect on the reported BPb decline. Three alternative measures of exposure to lead from gasoline were tested for their utility as Indicators of BPb response. Finally* time was Intro duced to represent ell other time-related BPb effects that may be present. ADJUSTMENT FOR DEMOGRAPHICS The effect of changing site demographics on the BPb time trend Was Inves tigated by classifying each examinee in terms of 6 personal variables (P) describing race, sex* age and family income* 3 residence variables (R) describing degree of urbanisation, end 6 caravan Itinerary variables (I) describing season of the year and region of the U.S.A. These classifica tion variables assume s value of 1 If the descriptor applies to the exam inee* 0 otherwise. Means were computed for each site based on examinee records with a venous BPb, end the 64 site means further analysed. Three BPb adjustments were evaluated corresponding to different selection sets of candidate explanatory terms for variation among site mean BPb'e. The first set included the 6 P variables and their 15 two-way Interactions; the second set added the 3 R variables to the first set; and, the third set added the 6 I variables to the second set. A weighted least squares stepwise regression procedure was applied to each full set of selection candidates. Veighting was by number of BPb values at each site. *E. 1. du Pont de Nemours & Co.* Tnc., Wilmington* DE* USA* Petroleum Laboratory* + Engineering Department TEH 0532523 DUP050033792 Terns yete retained! Sf their significance with BPb satisfied the p aC 0.15 criterion. Adjusted site seen BPb's ware obtained by adding the weighted site mean BPb to the residuals from the final regression nodel. Each adjusted set of values, along with the unadjusted BPb's, then was regressed on site mean examination date. Values of BPb were predicted at the first and last site mean examination date in NRANE5XX to evaluate the influence of dif ferential site demographics. The results of the tine analyses on the unadjusted and adjusted site mean BPb's ere given in Table 1. They illustrate the confounding inherent among BPb, site demographics and time during the study period. Each group of demographic variables was able to explain some of the time trend with terms from the third selection set reducing the unadjusted decrease by more than half. Table 1 Time Trend end Four Tear Change of Blood heed Unadjusted end Adjusted for Demographies Unadjusted Adjusted for P for P end R for F, R end I Time T*rendStd. ug/dl/yr Error -1.43 0.16 -1.09 -0.74 -0.57 0.18 0.17 0.14 Remaining Four Tear Decrease BPb. ug/dl X 5.61 32,7 4.29 2.92 2.26 26.0 16.4 14.5 ADJUSTMENT FOR GASOLINE LEAD AND DEMOGRAPHICS Gasoline lead (GPb) is the best documented of the identified sources of lead which include food, water, paint and dust. National GPb use has been claimed to explain the decrease in unadjusted BPb (ref 2), and So was eva luated as one GPb exposure variable. National GPb use {tonnes) was com puted from data by state on lead content and consumption of gasoline. To reflect site to site variation, s GPb density (tonnes/mi*) for each site was computed by multiplying the relevant state GP use by the ratio of population in the site counties to state population, end dividing by land area of the site counties. The basis for this definition is the correla tion between population and gasoline use {ref 3). Subsite GPb density was calculated similarly but based on each BPb examinee's residence char acteristics. Applicable data permitted calculation of all 3 GPb exposures for 35 of the original 64 sites. Kean BPb's from the 55 sites were adjusted as before for 5 selection sets, each containing all of the P,R and 1 demographic terms and one of the GPb exposure variables or its logarithm. The adjusted BPb's then vere regres sed on site mean examination date. These results, along with the signifi cance levels of the GPb exposure terms, are given in Table 2. In all cases inclusion of a GPb term further reduced the BPb decrease from the 2.26 ug/dl unaccounted for by demographics. The remaining trend was con sistent for all but national GPb use. TEH 0532524 DUP050033793 Tabic 2 Time Trend and Four Tear Change of Blood Lead Adjusted for Demographics and Gasoline Lead Exposure GPb Exposure Variable Slgnlf* ' Level P Rational 4. o.oooi Site Density 0.0001 Subsite Density 0.0311 Log (Site Density) 0.0001 Log (Subsite Density) 0.1076 Time Trend Std. ug/dl/yr Error -0.14 *0.44 *0.47 *0.52 -0.39 0.12 0.13 0.12 0.13 0.13 Remaining Four Tear Decrease BPb. ux/dl Z 0.57 1.75 1.85 2.05 1.55 3.8 11.3 11.9 13.1 10.1 TIME-RELATED BLOOD LEAD EFFECTS In the absence of credible and suitable data on lead sources besides gaso line, a time variable, reflecting site mean examination date, vas adopted as a surrogate for other time-related BPb effects. A stepwise regression procedure was used on each of 5 selection sets again containing the full Set of demographic terms, one of the GPb exposure variables end the sur rogate, time. First, the model was forced to include both time and tbe GPb variable used; in companion analyses the selection of ell terms was allowed to proceed solely on the basis of the Inclusion criterion. As shown in Table 3, under both forced end unforced conditions the time vari able was significant for all cases, but only Site GPb Density retained significance. When time is taken into account as e Surrogate for un quantified time-related BPb effects, national GPb use is no longer signi ficant. These results Suggest other time-related BPb effects are present in NHANESII. Table 3 Significance Levels of Gasoline Lead Exposure end Time Terms in Models Including Demographics GPb Exposure Variable Forced GPb Slgnlf. Level Time Slgnlf. Level Unforced ' GPb Slgnlf. Level Time Slgnlf. Level HatIona1 Site Density Subsite Density Log (Site Density) Log (Subsite Density) 0.8850 0.0022 0.6276 0.0019 0.7520 0.0001 0.0001 0.0001 0.0001 0.0001 HS 0.0022 HS 0.0019 NS 0.0001 0.0001 0.0001 0.0001 0.0001 HS * Hot Selected, p>0.15 TEH 0532525 DUP050033794 CONCLUSIONS This analysis has shown that many factors related to personal and resi dence characteristics and sampling Itinerary significantly affect NHANES11 BPb values, and can account for 3.35 ug/dl of the apparent decrease be tween 197$ and 1980. Of the GPb exposure variables only Site CPb Density and its logarithm had a significant effect on BPb in.the presence of the tiae surrogate. The portion of the four-year decrease due to Site GPb Density was determined by the analysis to be 0.5 ug/dl,' This is consis tent with the observed national average air lead decrease of 0.3 ug/3 over the sane period (ref 4), end the well-documented air lead/blood laad relationship (ref 5,6). Figure 1 shows the relative contributions to the four-year blood lead decrease. The unexplained portion of the SPb decrease, 1,75 ug/ dl, is reflected by the tine variable end nay be due to improved nampie handling over the course Of the NHANESII study, effectiveness of numerous government programs to reduce lead intake through food, paint and water (ref 7), or other unquanti fied causes. , REFERENCES 1 Centers for Disease Control, Morbidicy and Morality Weekly Rep 30, 132 (1982) .2 New Scientist, 94, 570 (1982) .3 J M Pierrard et al. Vehicle Emissions Controls and Ambient Air Quality, SAE Australasia, Jubilee Tear Conference, Melbourne (1977) 4. RPA Office of Air Quality Standards, National Trend in the Maximum Quarterly Average Lead Levels, 1970-1979 5. R D Snce, Int Arch Occup Environ Health 48, 219 (1981) . W Sinn, Int Arch Occup Environ Health 47, 93 (1980) and 48, (1981) 7. NatIonal Academy of Sciences, Lead in the Human Environment Washington, D.C. 1980) p. 477 TEH 0532526 DUP050033795