Document mBDL2qZvE8oer8rbazkzRnXjd

<mm> ESt&BUSHED BOS E. I. d u Po n t d e Ne mo u r s & Co mp a n y INCORPORATED Wil min g t o n , De l a w a r e 19898 ENGINEERING DEPARTMENT LOUVIERS BUILDING August 19, 1985 BCC: D.R. Diggs-C&P-B16262 P. Meitner-Legal-D70 82 G,,,,,M<&Lerw-ESD W.H. Walsh-ESD IC13 AH&IIE Dr ,, David Weil Project Manager Air Quality Criteria for Lead Environmental Criteria and Assessment Office MD-52 U. S. Environmental Protection Agency Research Triangle Park, NC 27711 Dear David: Attached is a copy of the paper, "Variation in the Relationship between Blood Lead and Air Lead," by R. D. Snee, S. P. Bailey, W. H. Fellner, and C. G. Pfeifer, published in Atmospheric Environment, Val 19, pp 1017-20 (1935), which is herewith submitted to the Lead Criteria Document docket, ECAO-CD-81-2. As in the past, we wish to continue to contribute to establish ment of a sound technical basis for the Criteria Document. Accordingly, we wish to participate in any workshops or other meetings to resolve the issues on the blood lead-blood pressure relationship which were raised by Professor Royall and others at the May CASAC meeting. Very truly yours, ENGINEERING SERVICE DIVISION Air, Hazards and Health Engineering (jW CLtAAMoi [dl \) 3. M. Pierrard Senior Consultant JMP:dll JMP1:68 Atch cos Mr. Robert Flook U. S, Environmental Protection Agency Dr. Lester D. Grant U. S. Environmental Protection Agency N31014 TEH 0412673 A$sasx+*rsr i : Voi. jc So. 6. pp. 3017-5020. ISfcS * s.-.ui; - too re-punos rn^f Lii SHORT COMMUNICATION ''-ri&A v ar iat io n in t h e r el at io n s hip bet w een bl o o d l ead an d AIR LEAD ' . AN-'rVi-V-'CA;..' . (Firmi-eceuredTS November 3983 end received for p-jolicaiion S November 1984} Abstract--Estimates of the slope of the relationship between blood lead tpf di": ,i anc air lead iy-c rr."; i obtained in ten independent studies are compared. No sienincarti differences ves delected among the estimates which ranged from 0.6 to 3.i jig dT' per and represented Etit adult, female adult and child populations. A single best estimate of 1.2 pgdT; per jig ns~:" (0.2.95 per cent confidence limits) was obtained. It is concluded than if the blood iead-ai? lead slope depends on venous physical and biological factors, then studies providing more precise estimates are required to detect these differences. - 5 "` ' JCey vori index: Air lead exposure, blood lead, uncertainty. Iri ' "s u mma r y ^. .. weighted average slope varies from 0.9 04 to 1.4 0.6 depending oirwhich slopes are included in the analysis. The ,;>r.-Chamberlain < i 9S3) provided a comprehensive review of estimate of the blood lead-air lead siope derived from all the ^'stuoies nivestigating tberelaDanship bet ween-blobd lead and', .data*-in Table 1<.is 1.2.0.1 This combined; estimate is barn JeadkHe concluded that estimates'of the'blood lead-air indicated by the horizontal dashed reference lint in Fig. J. - -lec-ri slope, denoted by.-te. varied from OS to 3.0/red)'1 per vf'^gmj^-depencing on the,study. He noted tfcartbe statistical uncertainty (95'percent confidence limits) in these estimates twas^seiy.ldss than 0.7 and that * ... r' - v T..;: .vV. _ TESTING BETWEEN-STUDY VARIATION OF SLOPES *' 'Y r ..NariciiS extrianarions aiay.be souaht for the divergence A critical step in the interpretation ofa set ofblood iead-air I-'.. ......lead-slopes' fromcdincrimf^rudies is die determination of. & of;.;...... :?"<* p-esEfieewa by both physical anc biolcgicdl.TacipiL.'Tiiesel'. whe:her.rthererarefla^r3^Sst!cal!y' sigriincant differences - among "the esnmaisdjsialusc;-Oite useful ^approach' to the- TAT r:TKjudt:nonlinearity;ofid6se response, eSetts ofparade ' T.T;.s2i. firman variability-in breathing patterns and venti- -tenc "rates. the varying extent to which the.tneasure- solution of this prcblem cs'.the.maximum likelihood pro-' csdure described in this-section, which requires from each i its standard error (St. John ;mer.tsof?bA correspond to the subject's real exposure..- fed'the.co.-ttribud6iitof:andl]ary- variables bon-diared VitivPbA-. - h-.:r- "e "tU3"*n-' where the tth estimated slope is expressed aa the sum of independently distributed Some of these factors contribute t o the statistical un normal variables denoting the true slop; and estimation error, certainty ir. tne estimates. The question remains as to whether we iel e, be tbe ith estimated slope, c its standard error, and thesiudy-to-srudy variation in the estimates of a are due to c\ the beiween-stuoy variance component It follow? that a. A differences in other factors: he. variation above and beyond the vector of e.'s. is distributed normal with unknown mean that contributed by the witfcin-stucy uncertainty in tbe vector c 1 and variance-cova-anee matrix V = tr| 1 different estimate? of a. A. statistical evaluation of this issue is -- diagfrfl. where ! is r vector of ores ar.c 1 the Identity the subject of this article. matrix. V is a diagonal matrix whose tth ci&cor.ri ricmer.t is A: least 10 studies of the relationship between blood lead c\-i-c? and the are assumed Imowr.. Toe procedure arte'air lead have appeared in the literature (Table 21 The presfoes estimates of r\ and the overall average true slope. =. results of. ttt-eie studies arc commonly reported as a slope 1: c'f = 0 then : is the t-tngle slept cemtr.or. tc t.-.t population .(change-in .blood lead per unit change in air lead!and its groups considered. - associated uncertainty. These results are summarized graphi The likelihood. L. is proponiona! to cally by the. approximate 95 percent ccnnoence intervals -shown in F.-c. 1. V>'i:h one exception, a range from 1 to (Vj -* ~ exp[ -- 1 ."le -- a! V v*: i- -- ell]. 2 eg cl's per eg m"f is consistent w;td these studies. It is of Tot likelihood rauo unctr the r.tl' hypothesis, r; = i? interest to determine whether there ate any statistical:) cl cn b; 1 = i_.. L-_. where L: is t:t; itVtiihr-vd ortamec signiSum: differences (i.e. variation:among (s) trteestimaiec using the maximum iikeiiinoc-d esttmatt fo; c and ri ur.c' 2;, is slopes in Table 1 and (:i| the estimated slopes wham male the likelihood ustne the maximum likelihood estimate for e ' adult female aculi and child populations. when tr: - 0. Tne test starisbe. -2 log/- is approximately d Ifthere is no berween-study variation in the slopes, then all cismibmtd as chi square whit one degree of truedcm under the Slopes are estimates of a common slope and it is appropriate null hypothesis. Large values of -- 2 log ;. suggest that the null -''to'.compute a weighted average slope to estimate the single hypothesis, cr; -0, is not true, indicating that the sieges are , "trite" slope of the blood lead-air lead relationship. not all estimating a single true' blood lead-air lead slope. '..A"Using she methodology described ir. the next section, it is Table 2 prorides a summary of the analyses performed concluded that we cannot reject the hypothesis of no Significartt bstw-een-siucy variation is mainly due to the differences among the slopes of the d Io d c lead-air lead Goldsmith adult male slope of - 0.1. which is out ofline with - '-relationship reported in Table hand further, that there is mo the other estimates (3.7 standard errors from the overall between-study variation among the slopes within male adult, estimatel and implausible. No significant between-study femaie adult and child populations. In Table 2 we see that the variation among the slopes of maie adults (analysis 1.4). 1017 N 31014.01 TEH 0412674 DUP050453729  w-y Short Communication Table 1. Studies of the relationship between blood lead anc air iead Population Study Ko. of subieets Biooc ?b-Air Pp slope* Estimate Sid. De'. Via:; adults F trr.a,c acuits Clinical adults A3a: j ohnson Nordman Tsuchiya ... .. _ _ ;' Golasmithty Fugas t: 336 537 ' -89.' 352 0.9 0.8 i.: 3.1 -0.1 > n 0..-5 0.50 3.10 0.35 0.-5 Teppcr-bevlr. J ohnson Nordman Goldsmith Dairies (Spring) Dames (fall! 1S35; 56 578 79 61 88 i.i 0.5 (t.6 . 0.7 1.6 3.4 0.90 0 30 0.45 C.35 0.85 0.60 Griffin (37g2m~5) 13 1.7 1.06 Griffin (10.9 pgra-*): . 8' 1.5 078 '{Johnson (Male) -;..:37 c.;; V..-- .3.4 :~ /_ 030 i-VJobnson. (r entale)b'-''.-f5TSfe D "43 cd viU 2 j .-r: .-0:30 ffXanhelt- > '-880 " 17 035 :Go!dsmithT - 486 7.0' 0.65 * Estimatestarid-siandard deviations reported by'Snee (1981%' tTneTanlrel and Goldsmith studiesincJuded'boih male and female children. Siiss8 ........'...............................-a-..y- ;,w.... ,-it US. 7-^ W<r -v; rt >j ms ' I it-*.'..-i. ~ v !! iJi i .1----- i----- ;----- :------1----- 1 - i-- ------ i----- ;----- i----- 1 A- -J ... N. T G .. F-- .. . T-L J N G D- D, ma l e a d u l t s FEMALE ADULTS G3 Z Grc.S CLMVICAl ADULTS Ji,. A Y C CK!LD=.= !V rig.'i:' Plot ofblood lead-air slope estimates and assc>ciaied approximate 95 per cent confidence intervals. S: udies arc identified bv first letter of senior author's last name. The horizonml dashed reference !mc corresponds to I bt combined slope estimate from all studies. !vwV'^'.:; TEH 0412675 DUP050453730 ** , Short i Tabieu. Blood lead--air lead slope:combining estimates from various studies Nc. cf studies Male adW4 Populations in analysis Female aeub.s Clinical acuits. Children Average slope Setweenstudv Estimate l r-ce--rr - Cm-vqua'e l i . ~'..3A -^$2. 6X 5 X -6 V. ' 3 i: x :-a u X *. X AA 7 X < 14 X 5A 13 X 64 .7 . 7A ' . 16 .. 35' X X gtfgps rX-- -cTC^-X X X X V X X X XX x :.i ;o.cO* 3.4 -- 0.6 3.57 N-S 09 0.4 0.00 NS 3.5 0.4 939- i ~ 0.4 :* i.: mCe : o.7 o' 3.4 = 0.4 ;.?s n s i.: = 0.4 30.20* 1.2 0.4 %oy X 3.5 . 0A 0.00 NS X i2 0.3 7.5K- X 351 0.3 2.31 NS :TM12,sLr 12 -03. ......7.88V .. 0' 220 NS '. *3erweer:-study-variance component significanily different from zero (F < 0.05). - .-?.* '-T-'iceGoidsmitb-AdulxssiODeof --0.1 was deleted from all analyses denoted bvan A. This slope is untealisucallv low nor-eonsisteerwirh rheother estimates: " v.r-' '*. ;T"0 standard deviation limits which are approximate 95 per cent confidence limits. These limits aje based on .... -estimated values of the betw-een-srudy and wjthin-siudy variance components. l-.V-'"'-' ,-EChi-square statistic (--0 logv.) for ihe between-srudy variance in-slope estimates. . KS Betw-een-study variance component not statistically significant , a ?t .'.vift,..s.-.:cy--p . v (analysisHkcmidrenfanalysts 6),|oj.ai.-Eroups ' ' iVccshbir.ec (analysis SA.) was detected, norderiihessgnincance. '- : 6.05j was detected in two of the other-four-analyses (3A r litis-analysis has indicsi*3That no significant, differences and SA) where some population groups were combined. Dae f F sS 0.051 were detected among the estimates of the siope of toihe lidr of stronp evidence of within orbetween population the blood lead-air iead relationship from the ten studies ._ t.j. -.V.-'iisroupvhiistioninibe slopes, it is'concluded thatacocntiinsd considered or,from amcnsisiimates within adult rnaie. adult SC- riopeiprorideSTSi best'estimaie ofdhe rehtlidnsFiplbetween: rtfemajeTor cniid -populaac(a%fp!ips. It follows -that, a: this . bloodJead ar.d air lead. V' time, the-overall weighted average of 3.2pcdl~ per ugitri i C.2, 95 per cent eonScence iinhsi provider the best estimate of a common sic-pe. .-.a,*.- COMPUTING THE WEIGHTED AVERAGE SLOPE it is apparent that the uncertainty in many of the estimates \fhvTr ;-V.. ; ' V;-;, ........... ' is large with 95 per cent confidence limits of 0.7 io 1.0 ''LV'TCiw -weighted average `slopes reported in Table' were about the estimate. is concluded that studies which yield crtnp'Jiei by itsrativtty soiling the following set of equations more precise slope rr.ima'es at; needed tc- dtte--.me the e -- U'V~ :xi (I'V: 1) importance cf tie terror; rc-tec by CVmbertritt. This v. necessitate the r-i-.ner rcr.tro; of some variables and the mote '-rf-VaflSt]'1 - tract (V*1)** fc - ir\~* -21) Y = cji -r-diag refg. accurate measurement cf others. Snee < 195I I recommended that studies of the relationship betw-etr. blot c lead and air leac should have the ht '-ir.r key YTHTse'S estimates, which depend on the estimates of between" study variation. 6|. are in good agreement wj;h those derived . by Snee (1951 X which w>ere computed using or.iy the epi- denioiopr studies and where :: was implicitly assumed or = 0. characteristics! air lead exposures be measured by personal monitors: biooc samples be obtained from each subject or. at least two different days: lead itvei in each sampie be measured :r capikate: and. statistics: calculations tv made tc ensure that t.Sficiert fur, sets ar-. tided to estimate the rl--r-e - tab the desired r'irifcr. Ar- a: certainty cf -0.: or let* -rf per 5ioc>d ?b-A.ir Fh slope cent confidence limits; is a reasonable coai. Persons ctsgmnf such programs may wish tc review- the study of .Anar a: of. pobuiation Saee (39S3) Tnis analysis il97r! which had these characteristics. ` Y. '?aA'i503 Female actulss . ... QP0.4 Oiidren ' 13 = 0.3 AE subjects Not reponed ' 3.4 0.6 0.9 0.4 13=0.4 35! 053 Engineering Dencnm^ni E. 1. du Pom Cr Semours one Co. HiImington, Deia^-are 39S9S. i .S.A. Ro n al d D. Sn h e" St e v en ?. Ba il e s 'ATl l ia v . K. Fe l l n er Ch a r l e s G. Pf ej f er ' 'Ai -i jA- .diagnostic check on the validity of the distribution " "assutaprion underlying this analysis was made using a probability plot of the quantities (a, - S', (<r j -- which .showed reasonable agreement with ihe standard normal " New- address: Er.pnstrirtg cepar.mtm. Beaumont works. E. 1. cu Pont de Nemours and Co.. P.O. Box ~yrc x raLmcr.t. TX 77704, U.S.A. i J TEH 0412676 DUP050453731 Shon Corr.irtunica.tion REFERENCES Ahar A_ Snee R.. D. and Kabibi K. t!S'!5l An epidemiological approach to conmuntry air lead exposure using personal air S2crpies. in Em-tronmemal Qwihty end Sedeiv--bead Suppiemen:. Yel. II 'edited by GriSn 7. 3. and Kneisor J. H .. ~r. 1 Asedstrsic Press. Lcncon. ib.arr.beriarr A. C Effect of airborne ieac or; biooc ie^C. .-'.''-.. rrr.t-.; r.rrenn:i-r.: if. 67"--e92. Su John D. S. Bailey S. P. Feline: W. H.. Minor 2. M. ana SnteR.D. (19S1J Time series search for trend :r. iota1 ocone measurements. J. geopnys. Res. KoiCSi. 7299-7F ii. Sr. john D. S., Bailey S. P_reimer W. H_ Minor j. M. and Sn=e R. D. (19S2t Time series analysis of stratospheric ozone. Commun. S:e:is:. The.-*-- Merhc'ds J~. 1292-1".-". Snee R. D (1PM i rvai jaber stccirs of :: c rcia'.o-.s:.:; beta-eer. blood ieac ar.c a:: ieid e: .-;,`sr v:- 22 tin 48. 2; 1--292. -r i Afj'Ss'r!:! TEH 0412677 DUP050453732