Document mmbqnyM8VOa8v3dDxVR0Nz5yk

Development of an Air Quality Standard for Lead from Community Studies i i Ronald D. Sn*e 1 % I I i i j Volume 16, Number 5 Pages 241-246, May 1982 | Copyright 1982 by the American Chemical Society andreprinted by permission of the copyright owner 1 TFH 0470033 Development of an Air Quality Standard for Lead from Community Studies Ronald D. Snee Engineering Department, E. I. du Pont de Nemours & Co., Wilmington, Delaware 19698 A methodology for the development of an air quality standard (AQS) for lead is presented. It is shown that the results of community studies can be used to calculate the air lead level at which the cumulative frequency distri bution of blood lead values of a population will meet a biological guideline for blood lead. The procedure is il lustrated by using the data on the relationship between blood lead and air lead obtained in two well-known epi demiological studies, A variety of analyses, involving different blood lead-air lead models, blood lead frequency distributions, and data bases, are .included to study the sensitivity of the methodology to variations in assumptions and calculation procedures. The proposed methodology is general and can be used in the development of any AQS for which the appropriate biological guideline is expressed in the form of a cumulative frequency distribution.1 1. Introduction In 1974 Zielhuis (1) proposed a biological guideline for blood lead. This guideline provides for the protection of public health by specifying a desirable distribution of blood lead values. It is therefore proposed that the air quality Standard (AQS) for lead be determined by calculating the air lead level at which Zielhuis* guideline trill be met. The methodology for making these calculations and developing the AQS is described. It is shown that blood lead values from several populations can be combined to give an ac curate estimate of the within-population distribution of blood lead values. The data collected by Azaret ah (2,3) and Tepper and Levin (4) are used to illustrate the pro cedure and to develop an AQS for lead. These results are also compared with the standard adopted by the United States Environmental Protection Agency (5). The ap proach is general and can he used in the development of an AQS for any pollutant for which a biological guideline similar to that of Zielhuis is appropriate. Any group of subjects exposed to a given air lead level will have a distribution of true blood level values because of biological differences between the subjects and differ ences in lead exposures from sources other than air lead. The distribution will be further broadened due to blood lead measurement variation. As the air lead level of the group is increased, the corresponding blood lead distri bution willshift upward. This suggests that an air quality Standard for lead could be determined hy finding the air lead level at which the upper portion of the predicted blood lead distribution will be equal to that of Zielhuis' biological guideline distribution. A schematic of this procedure is shown in Figure 1. In order to make the calculations described above, it is necessary to have a mode] for the relationship between blood lead and air lead and data on the within-population variation in blood lead values. The best available data on the blood lead-air lead relationship for adults is that published by Azar et aL (2,5). In this study air lead and blood lead measurements were obtained on 150 subjects over a 2- to 4-week period. The air lead was continuously measured 24 h/day during the test period by using per sonal monitors. Two to eight blood samples were taken from each subject and duplicate lead determinations were made on each sample. This is the only study of nonoccupationally exposed subjects whose air lead exposure was monitored continuously by personal air monitors. The Seven Cities survey (4) contains a large amount of data on vdthin-population variation in blood lead values. The blood lead levels of 2015 subjects from 12 populations (80-219 subjects/population) were collected in this study. This study is important because a large sample size is required to get a precise estimate of the within-population distribution of blood lead values. The use of the data from the Azar study and Seven Cities survey in the development of an AQS for lead is described in the following paragraphs. 2. Biological Guideline for Lead After an extensive review of the literature, Zielhuis (1) concluded that the following blood lead distribution was an acceptable biological guideline for the protection of public health. percentile blood lead, fig/dL 50 20 90 30 98 35 This guideline has been widely accepted and has been adopted by the European Economic Commission (6). Hus distribution will boused to determine an AQS for lead by first calculating blood lead levels associated with percen tiles of interest and then calculating air lead levels that would raise these blood lead levels to those of the Zielhuis distribution. We concentrated on matching the 90th and 98th percentiles, since individuals at the upper end of the distribution are at greater risk. Results for the 50th percentile are included, however, for comparative purposes. It is important to recognize that, in terms of observed and theoretical distributions of blood lead data, the 50th percentile of the Zielhuis distribution is not consistent with the 90 and 98 percentiles. If we assume that blood lead levels are lognormally distributed, then the 90th percentile of 30 and the 98th percentile of 35 /qj/dL are consistent DUP050082854 TEH 0 4 7 0 0 3 4 AIR LEAD Flour* 1, Schematic of calculation procedure for determining the air lead level at Which a percentile of an observed distribution wll be >nsisterrtwNht)ie Zieli& Unear blood teacHair lead model and 98th percentile are shewn In the figure. with a 50th percentile of 23.2 *tg/dL. In the unlikely event that blood lead levels are normally distributed, the 90th and 98th percentiles of the Zielhuis distribution are con sistent with a 60th percentile of 21.7. It is concluded that the 50th percentile of the Zielhuis distribution is low by approximately 3 fig/dL. 3. Relationship between Blood head and Air head To determine the ah lead level at which the blood lead distribution of a normal population was consistent with the Zielhuis distribution, it was necessary to blow the relationship between air lead and blood lead. Two dif ferent relationships were used, the first model was the `lead exposure" model (7,8), which assumes an exponential relationship between blood lead and air lead. The second model was a linear relationship between blood lead and air lead. Both ofthese mathematical forms have been used widely in correlating air lead and blood lead data. The "lead exposure" model used in this analysis was blood Pb = 12.1(air Pb + B)-269 (1) This model was developed from the Azar data (7). It relates blood lead to total lead exposure by dividing lead exposure Into twoparts--air lead exposure wad background lead exposure (be., food, water, etc.). In eq 1, background lead is represented by the B coefficient, which was found to be 3.28 for the Azar data (7,8). Additional details on the development of this model are given in ref 7. The second model used (eq 2) is a straight fine blood blood Pb = air Pb + B0 (2) lead-air lead relationship with a slope of 1 (ie., an increase of 1 jug of Pb/m8 of air results in an increase of 1 fig of Pb/dL of blood). The blood lead-air lead slope of 1 was found in a combined analysis of eight epidemiological studies of the relationship between bipod lead and air lead (8) and in the studies conducted by Chamberlain et aL (9). 242 Environ. Sd. TechnoL, Vol. 16, No. 6,1982 The slope of 1 is also supported by the lead exposure model (eq 1) for the Azar data and by the analysis of the Azar data conducted by Hammond et al. (10), both of which indicate that the blood lead-air lead slope for the Azar data varies from approximately 1.2 to 0.6 for ambient air lead levels :10 Mg/m3. Slopes huger than 1 have been reported (11,12); however, these appear to be out of line with re spect to the other studies and axe based on olininal rather than epidemiological studies. The methodology described here is general and can accommodate any assumed slope or blood lead-air lead model. 4. Blood head Distribution So that the blood lead distribution for the Azar date could be determined, a blood lead value was calculated for each subject corresponding to an air lead level of zero. Equation 1 indicates that the ratio of the blood lead Bt at air lead Ai to the blood lead B2 at a second air lead A2 is JBt T A1 + 3.28102669 W2 ~ |_ A-i + 3.28 (3) Hence, the lead exposure model (eq 1) predicts that the blood lead of a subject at an air lead of zero will be where Bo is the calculated blood lead at an air lead of zero, A is the air lead exposure of the subject, and BA is the blood lead of the subject. In the case of the linear model, the blood lead at zero air lead is given by eq 5, Bo w Ba -- A (6) where BQ, BA, and A are defined as in eq 4. After tee bipod lead values were obtained for the zero air lead level, the next step was to determine tee distri bution of blood lead levels calculated from eq 4 and 5. The blood lead levels corresponding to selected percentiles were obtained from the best fitting distribution of tee Pearson, system (13, 14) because the frequently used lognormal distributionwas found to give an inadequate fit in several instances. The Pearson system encompasses a wide variety of frequency distribution forms and does not require an as- distribution to which the data conform. Many of tee distributions that are frequently used in the analysis of air quality data (i.e., normal, lognormal, gamma, beta, and Weibull) are special cases of tee Pearson system. This family of models selects that distribution whose first four moments (be., average, standard deviation, skewness, kurtods) are the same as those of the observed distribution. The lognormal distribution will be fit when it is the ap propriate distribution; hence, the fitof the Pearson system will be equal to or better than that of the lognormal dis tribution. It is shpwn in section 8 that the Azar data and some of the data in the Seven Cities survey are not lognormally distributed. It is well-known that tee lognormal distri bution is adequate for estimating geometric mean rela tionships among air pollution variables such as that de scribed by eq 1. We will see later that, in tee case of the 50th percentile, there is no practical difference between the cur lead levels developed from the lognormal and Pearson distributions. In the case of tee Azar Study and Seven Cities survey, however, the lognormal distribution TEH 0470035 DUP050082855 .Table I. Air Lead Levels Associated with Zielhuis Biological Guideline for Lead .Based ps the Analysis of Azar Data ` blood percen Pb, airPb, tile Mg/dL Mg/m* lead exposure model, Pearson system lead exposure model, normal distribution6 linear model, Pearson system linear model, normal distribution6 60 90 97,6 SO 90 97.6 98', 60 90 97.6 SO 90 97.6 98 15.8 22.4 26.9 16.7 22.6 27.1 27.8 16.7 23.2 28.5 16.6 23.5 28.6 29.4 4.2 6.6 5.5 4.8 6.2 5.8 4.5 4.3 6.8 6.6 4.4 6.6 6.4 5.6 At air Pb =0. 6 The normal distribution gave an ade quate description of the Variation in (blood Pb)M. does not provide an adequate description of the upper percentiles of the blood lead distribution. It is very im portant to get accurate estimates of these values because the Zielhuis distribution and the associated air qualify standard are baaed on the upper percentiles. The Pearson system is able to better address this accuracy because of its rich family of models and its use of four moments to select the best fitting distribution. Results for the log normal distribution have been included, however, for comparative purposes. An alternative procedure to the Pearson system is to transform blood lead data to another scale such that the distribution of the transformed values is closely matched by a normal distribution. As noted earlier, the Azar blood lead data at aero air lead were not lognormally distributed; however, blood lead raised to the 0.3 power, (blood lead)0'1, was found to be approximately normally distributed. The blood lead values calculated from the Azar data with use of the Pearson system are summarized in Table 1. The corresponding values calculated from a normal distribution approximation to the transformed data ((bloodlead)0-3) are also included in Table I. In each instance, the results obtained from the Pearson system are in close agreement with those obtained from the normal distribution, indi cating the Pearson system gave a good description of the observed data. The 97.6th rather than the 98th percentile, is used in the Pearson system. This slight deviation from the Zielhuis distribution was necessary because published tabulations ofthe Pearson system contain the 97.6th rather than the 98th percentile. After tite blood lead values Corresponding to zero air lead were calculated, it was then possible to compute the air lead at which the blood lead distribution was equivalent to that ofthe Zielhuis distribution at the higher percentiles. In the case of the lead exposure model (eq 1), tine air lead is given by eq 6, where B0 is the blood lead at zero air lead At m 3.28[(Bz/B0)1/- 2> - 1] (6) and Az is the air lead associated with the corresponding blood lead (B^) of the Zielhuis distribution. In the case of the linear air lead-blood lead model, the air lead level that equates the two distributions is given by eq 7, where Az = Bj j - B0 (7) A-b Bz, and Bo have the same definition as in eq 6. 5. Air Quality Value Developed from. Azar Data The air lead at which the Azar blood lead distribution Was equivalent to the Zielhuis blood lead distribution was calculated from eq 6 and 7 (Table I). The availability of two different air lead-blood lead models (linear, lead ex posure), two different blood lead distribution models (Pearson system, normal-power transformation), and three different percentile levels (60,90,97.5) made it possible to analyze the Azar data in a variety of different Ways. This enables a determination of the sensitivity of the re sulting air lead levels to the calculation procedure used. In Table I we see that in all cases the calculated air lead levels are greater than 4 /*g/m3. In the case of the upper (90,97.5) percentiles all the air lead levels are greater than 5 /%/m3 except for the lead exposure model, nonnal-power transformation case. 6. Air Quality Value Developed from Seven Cities Survey Data The calculations! methods used in the analysis of the Azar data were applied to the data obtained in the seven cities survey. This is important because the seven cities survey included blood samples obtained from 2015 subjects in 11 different locations. This large data base enables one to obtain an accurate estimate of the upper percentiles of the distribution of blood lead values. The air lead expo sures were obtained from stationary samplers located in the areas where the subjects lived and are thus less spe cifically related to individual blood lead than was the case in the Azar study. In this analysis, the lead exposure and linear blood lead-air lead models were both used. In view of the close agreement of the results of the Pearson system and the normal-power transformation, only the Pearson system was used to describe the blood lead distribution. The air lead levels and the associated 50th, 90th and 97,5th blood lead percentiles calculated from the Seven Cities survey data are summarized in TabbIL The air lead values used were the annual average air lead results obtained at each site. The air lead levels needed to make the observed blood lead distributions equivalent to the upper end of the Zielhuis distribution were calculated from eq 8, Where Ay is the Az = (Ay + 3.28)[(BZ/B7)l/01668 - 3.28] (8) annual average air lead associated with the observed blood lead By obtained from the Pearson system and Az is the air lead associated with the blood bad Bz of the Zielhuis distribution. In the case of the linear model, excluding the Ritten- house location, the calculated air lead levels are greater than 5 ag/m3 in 28 out of 33 instances and greater than 4 Mg/m3 in 30 out of 33 instances. Bach of the five in stances where the air lead level was less than 5 Mg/m3 was associated with matching of the 50th percentile. The United States EnvironmentalProtection Agency (15) has pointed put that Rittenhouse is an old section of Phila delphia in which many of the houses contain lead plum bing, which probably contributed to the relatively high blood lead values. The lead exposure model calculations showed results. Excluding Rittenhouse, the air lead values were greater than 5 /<g/m3 in 26 out of 33 instancea.and greater than 4 ag/m3 in 30 out of 33 instances. Three of the seven instances where the air lead level was less than S Mg/m3 were associated with matching of the 50th percentile. We also see in Table II that there is a wide variationin the air lead values calculated for the different locations. This reflects variations in exposures to lead other tfn that Environ. Sd. Technol., Vol. 16, No. 6, 1982 243 TEH 0470036 DUP050082856 Table II. Air Lead Levels Associated with the Zielhuis Biological Guideline for Lead Based on Analysis of.Seven Cities Survey Data airPb6 lead expo- percen- blood sure linear site airPb tile Pb* model model Okeana Ardmore Rittenhouse Pasadena Los Alamos, male Los Alamos, female Washington, D.C. Port Washington Greenwich Village Lombard Bridgeport Houston 0.32 1.15 1.67 3.39 0.17 0.17 1.19 1.13 2.08 1.18 (7 mo) 1.76 0.86 50 90 97.5 50 90 97.6 50 90 97.5 60 90 97.5 60 90 97.6 50 90 97.6 50 90 97.6 50 90 97.5 50 90 97-5 50 90 97.6 50 90 97.5 50 90 97.5 15.9 23.0 27.6 18.8 26.1 29.8 20.4 29.5 35.6 17.5 25.0 29.9 17.0 23.5 28.2 15.2 20.3 23.8 19.2 26.0 29.6 16.2 21.4 25.4 16.4 22.9 27.6^ 14.0S 19.0 22.9 17.2 24.2 29.8 12.7 17.8 20.8 5.2 6.5 6.5 2.3 4.2 4.8 1.3 2.0 1.4 7.7 9.9 8.8 3.1 5.3 4.5 6.4 11.6 11.4 1.9 4,4 6.1 9.1 12.4 11.4 8.0 11.5 10.0 13.7 21.4 18.6 5.6 8.0 5.0 19.4 25.9 26.7 4.4 7.3 7.7 2.4 5.1 6.4 1.3 2.2 1.2 5.9 8.4 8.5 3.2 6.7 7.0 5.0 9.9 11.4 2.0 6.2 6,6 5.9 9.7 10.7 6.7 9.2 9.6 7.2 12.2 13.3 4.6 7.6 7.0 8.1 13.0 15.0 Blood Pb value (pg/dL) estimated by fitting the Pearson System to the observed frequency distribution. 6 Air Pb level (pg/m*) at which the upper portion of the observed blood Pb distribution would be equal to the Zielhuis guideline. in the .air and to variations arising from analytical un certainties in the data. 7. Air Quality Value Developed from the Analysis of the Combined Azar and Seven Cities Survey Studies The combination of the Azar study with its excellent estimate of the blood lead-air lead relationship and the Seyen Cities survey until the large number of subjects provided the beat basis for establishing an air quality value forlead. So that this could be done, the blood lead values for the 2016 subjects in the Seven Cities survey were used to establish a blood lead distribution. This distribution was constructed by subtracting the average blood lead for each site from each of the blood leads collected at that site. Next, these deviations from the site average were combined into a single distribution. The standard deviation of this distribution was the pooled within-gite standard deviation and was a measure of the blood lead variation in the normal population. Since the blood lead standard devia tion typically increases with increasing blood lead level, 244 Environ. Set Technol., Vol. 16, No. 5, 1982 Table III. Air Lead Levels Associated with Zielhuis Biological Guideline for Lead Based on the Analysis of the Combined Azar and Seven Cities Studies dev percen from tile av blood Pb,6 7a*irPb, pg/dL pg/m* lead exposure model, Pearson system lead exposure model. normal distribution linear model, Pearson system linear model. normal distribution 50 90 97.5 50 90 97.5 98 60 90 97.5 60 90 97.5 98 0.0000 0.0738 0.1188 0.0000 0.0766 0.1170 0.1226 0.0000 0.0738 0.1188 0.0000 0,0766 0.1170 0.1226 15.8 21.7* 26.1 16.8 21.9 25.9 26.5 16.8 21.7 26.1 15.8 21.9 25,9 26.5 4.2 7.8 6.6 4.2 7.4 6.8 6,0 4.2 8.3 8.9 4.2 8.1 9.1 8.4 Distribution of (blood Pb l*-15: average = 0, standard deviation^ 0.0697, skewness = -0.02, kurtosis-- 3.81. The normal distribution has skewness = 0 and kurtosis = 8.0. 6 At air Pb = 6. c Blood Pb = ((16.77)-" + O.Q738),,*J* = 21,7, it was necessary to transform the blood lead levels before the combined distribution was constructed. The objective was to make a symmetrical distribution. It was found that tiie power transformation (blood lead)019 resulted in a symmetrical distribution; however, the tails of the dis tribution were longer thaw those of the normal distribution (kurtosis ~ 3.81, whereas for a normal distribution kurtosis = 3.0). For this reason, the percentiles were estimated by both the Pearson system and the normal distribution (Table III). In the Azar data, the blood lead equivalent to the 60th percentile point at air lead - 0 pg/m3 (lead exposure model calculation) was 15.77 pg/dL. The blood lead values associated with the upper percentiles of the blood lead distribution were obtained by adding the cor responding values of the "deviation from site average" distribution to the blood lead 50th percentile of the Azar__ data at zero air lead. A sample calculation using the" Pearson system for the 90th percentile is.shown in eq 9 blood Pb " [(15.77-16 + dev from site av)]1/al5 = [(15.77)0-13 + 0.0738] = 21.7 (9) The air lead levels that equate these blood lead values to those of the Zielhuis distribution are summarized in Table m. The calculated air lead level for the 50th percentile is greater than 4.2 pg/ms for all cases. The ten air lead values associated with the 90th and 97.5th percentiles ranged from 6,0 to 8.9 pg/m3. It is thus apparent that an air lead levelof 4 pg/m3 is consistent with all parts of the Zielhuis guideline and that the upper percentiles (90 and 97.5) of the blood lead distribution of a population exposed to an air lead level of 5 pg/m3 would be within the Zielhuis guideline, 8. Log Blood Lead Distribution The lognormal distribution has been widely used in the analysis of blood lead data. For this reason it is appro priate to.consider what air lead levels would be obtained if the lognormal distribution is used to describe the blood lead distribution rather than the more general Pearson system. The resulting air lead levels are summarized in Tables IV and V. It is important to understand that these TEH 0470037 DUP050082857 TEH 0 4 7 0 0 3 8 3I Table IV. Azar Study: Estimation of an Air Quality Standard for Lead Assuming Blood Lead Levels Are Lognormally Distributed lead exposure model linear model lognormal blood percentile Pb AQS6 blood Pb AQS6 50 15.5 5.2 16.3 4.7 90 22.9 6.7 23.8 6.2 97.5 28.2 4.1 30.1 4.9 98 29.0 3.4 31.0 4.0 Blood Pb (jig/dL) at air Pb- 0 (lognormal distribu tion). 6 Air quality standard - air Pb levels (pg/m9) at which the upper portion of the observed blood lead dis tribution would be equal to the Zielhuis biological guide line for lead. results are representative only if the lognormal is the correct distribution to apply. We tested the adequacy of the lognormal distribution by Computing the Skewness and kurtosis statistics for the log blood lead data sets in the Azar study and the Seven Cities survey and comparing these results with the corre sponding parameters of the normal distribution. Using the statistical tests discussed by Pearson and Hartley (13) we found that at least 7 of the 13 observed distributions (Azar'splus 12 sites in the Seven Cities survey) were Sig nificantlydifferent from the lognormal distribution (Table VI). It is important to note that the direction of the skewness of the log blood lead distribution is reflected in how the resulting AQS compares to that computed by using the Pearson system. Negative skewness results in lower values than the Pearson system while positive skewness results in higher values (compare Table I with Table IV and Table II with Table V). Chi the log.scale the lognormal distribution has no skewness; hence, when the skewness of the data is negative, the blood lead levels predicted by the lognormal distribution will be higher than those of the observed data and will result in lower air lead levels. The opposite effect occurs when the skewness is positive. As noted earlier, the Pearson system has the flexibility to describe a wide variety of distributions, including the lognormal. It is concluded that, in the case of the Azar study and the Seven Cities survey, the Pearson system provides a more accurate characterization ofthe blood lead distribution than the lognormal. 9. Discussion A methodology for determining an air quality standard for lead from epidemiological studies has been described and illustrated by using the best available data on blood lead-air lead relationships and within-population variation in blood lead values. The Azar study and the Seven Cities survey include a wide Variety of different adult population groups that provide a firm basis for the determination of an air quality standard for lead. The standards developed from these studies using a variety of models and assump tions are summarized as follows: calculated air quality standard, ug/m9 * * study average range Azar 5.6 4.2-6.8 Seven Cities survey 8.0 1.2-25.9 combined Azar and 6.5 4.2-8.9 Seven (Sties survey With a few exceptions in the Seven Cities survey (Rit- Table V. Seven Cities Survey: Estimation of an Air Quality Standard for Lead Assuming Blood Lead Values Are Lognormally Distributed AQS6 lead expo- percen blood sure linear site airPb tile Pb model Okeana Ardmore Rittenhouse Pasadena Los Alamos, male Los Alamos, female Washington, D.C. Port Washington Greenwich Village Lombard Bridgeport Houston 0.32 1.16 1.67 3.39 0.17 0.17 1.19 1,13 2.08 1.18 1.76 0.86 50 90 97.6 98 50 90 97.6 98 60 90 97.6 9$ 50 90 97.5 98 50 90 97.5 98 50 90 97.6 98 50 90 97.5 98 50 90 97.6 98 60 90 97.5 98 50 90 97.6 98 50 90 97.5 98 50 90 97.5 98 15.6 23.6 29.8 30.2 18.0 27.1 33.6 34.6 20.6 29.3 35.4 36.4 17.5 24.9 30.0 30.8 17.2 23.2 27.2 27.8 15.0 20.6 24.3 24.9 19.1 25.7 30,1 30.7 15.4 21.1 26.5 25.5 16.6 22.7 26.8 27.5 14.0 19.1 22.5 23.0 17.6 23.9 28.0 28.7 12.6 17.8 21.5 22.1 6.9 5.6 8.7 3.0 3.3 3.2 1.9 1.3 1.2 2.1 1.5 1.0 7.7 10.1 8.6 7.5 2.8 5.8 5.6 4.9 6.9 10.8 10.3 9.1 2,0 4.7 4.6 4.0 8.6 13.2 12.5 11.2 7.6 12.0 11.3 10.0 13.7 20.9 20.1 18.2 4.9 11.8 8.3 7.3 20.0 26.9 22.4 19.8 4.7 6.7 6.0 5.1 3.2 4.1 2.6 1.6 1.2 2.4 1.3 0.3 5.9 8.6 8.4 7.6 8.0 7.0 8.0 7.4 5.2 9.6 10.9 10.3 2.1 5.5 6.1 6.5 4.6 10.0 11.2 10.6 6.6 9.4 10.3 9.6 7.2 12.1 13.7 13.2 4.2 7.9 8.8 8.1 8.2 13.0 14.3 13.7 0 Blood Pb value (pg/dL) estimated from a lognormal distribution. 6 Air quality standard: air Pb levels (ug/m9) at which the upper portion of the observed blood Pb dis tribution would be equal to the Zielhuis biological guide line for Pb. tenhouse and the 50th percentiles for Ardmore, Los Ala mos (male), Washington, D.C.) all of.these air lead levels are greater than 4 Mg/m3, The air quality standard of 4 Mg/m3 5is considerably higher than the 1.5 Mg/m3 pro mulgated by the U.S. Environmental Protection Agency (5) and the value of 2 Mg/m3 proposed by Yankel et aL (16). In the development of their standard, the EPA used a different and more stringent risk level than the Zielhuis guideline used in this work. They identified children as the high-risk population, assumed blood lead values were Environ. Set Technol., Vol. 16, No. 5, 1982 245 DUP050082858 T&ble VI. Summary Statistics for Log Blood Lead Distributions from Azar and Seven Cities Studies data set log blood skew kurto n Pb, av std dev ness sis Azar date* linear 149 lead exposure 149 linear-power 149 lead exposure 149 Okeana 162 Ardmore 150 Rittenhouse 136 Pasadena 193 Los Alamos, 80 male Los Alamos, female 191 Washington, D.C. 219 Port Washington 198 Greenwich 140 Village Lombard 208 Bridgeport 147 Houston 191 1.186 1.189 2.281 2.283 1.194 1.256 1.312 1.244 1.236 1.176 1.282 1.188 1.219 1.147 1.246 1.099 0.149 -0.35d 0.133 -0.31 0.232 -0.01 0.208 -0.06 0.139 -0.817 0.138 -0.767 0.121 040 0.119 -0.06 0401 0.44d 0.107 -0.34d 0.100 0.106 0.107 0.06 0.20 0.21 0405 -043 0403 0,657 0419 -0.09 3.27 2.90 347 3.00 6.477 3.877 2.84 3.42* 3.05 3.96<i 2.42 3.09 3.28 4,927 3.69* 2.69 For a normal distribution, skewness = 0, kurtosis = 3.0 (skewness < 0 indicates the distribution is skewed to the left). 6 Blood Pb at air Pb = Q. Results of power trans formation - (blood Pb)0,3. d Distribution not normal (p < 0.05). * Deviation from normality significant (p < 0.10). 7 Distribution not normal (p < 0.01). lognormally distributed with a geometric standard devia tion of 1.3, and calculated the air lead level at which a blood lead 99.5th percentile of 30 pg/dL would be ob tained. In making this calculation, the EPA used a blood lead-air lead slope of 2.0, which they estimated from the data reported by Yankel et ah (16), who studied a single population of children that lived close to a lead smelter. The control population in this study had an average blood lead level of approximately 30 pg/100 mL, which is con siderably higher than that of typical nonoccupationally exposed populations. The authors of the Yankel study have concluded from more recent analysis that the appropriate blood lead-air lead slope for this study is approximately 1.0 (27). Our analyses (8) of the Yankel data and other children studies reported in the literature suggest that the blood lead-air lead slope for children is approximately 1.0 and not sig nificantly different from that of adults. Since the EPA has identified young children as a highrisk papulation, it is appropriate to perform the analysis described in this paper for child populations when the appropriate data became available. It is also recommended that the proposed methodology he used in the development ofan sir quality standard for any pollutant for which the appropriate biological guideline is expressed in the form of a cumulative frequency distribution. Acknowledgments I express my appreciation to D. W. Marquardt for many helpful discussions during the course of this work and to the referees, whose comments helped improve the pres entation of this paper. Literature Cited (1) Zielhuis, R. L. bit. Arch. Arbeitsmed. 1974,32,103-127. (2) Azar, A; Snee, R. D.; Habibi, K. Comm. Eur. Communities, [Rep.] EUR 1973, EUR-5004, 681-694, (3) Azar, A; Snee, R. D.; Habibi, K. Environ. Qual. Saf. 1975, 2, 254-290. (4) Tepper, L, B.; Levin. L. S. Environ, Qual. Saf. .1976, 2, 162-197. (5) US. Environ. Prot. Agency, Fed. Regist. 1978,43, No. 194. (6) Commission of European Communities, Council Directive ofMarch 19,1977, on Biological Screening ofthe Population for Lead, Off. J, Eur. Communities, 1977, 20, No. L106, 10-17, Brussels. (7) Snee, R. D. Report PLMR-10-78; El I. du Pont de Nemours and Co., Petroleum Laboratory: Wilmington, Delaware, unpublished data. (8) Snee, R. D. Int. Arch. Occup. Environ. Health 1981,48, 219-242. (9) Chamberlain, A C.; Heard, M. J,; Scott, A. B. N.; Clough, W. S.; Newton, D.; Wells, A. C. Postgrad, Med. J. 1976,51, 790-794. (10) Hammond, P. B.; O'Flaherty, E. J.; Gartside, P. S. Food Cosmet. Toxicol. 1981,19,631-838. (11) Chamberlain, A C.; Heard, M. J.; Little, P.; Newton, D.; Wells, A C.; Wiffen, R. D. UJC. At. Energy Auth. Rep. AERE-9188,1978. (12) Rabinowitz, M. B.; WetheriH, G. W.; Kopple, J. D. J. CUn. Invest. 1976,68, 260-270. (13) Pearson, E. S.; Hartley, H. O. "Biometrika Tables for Statisticians"; Cambridge University Press: London, 1972; VoL n. (14) Hahn, G. J.; Shapiro, S, S. `Statistical Models in Engineering"; Wiley: New York, 1976. (15) United States Environmental Protection Agency, Office of Research and Development, Washington, D.C., November 1976. (16) Yankel, A J.; von Lindera, L H.; Walter, S. D. J. Air Pollut. Control Assoc, 1977,27,763-767. (17) Walter, S. D.; Yankel, A J.; von Lindem, I. H. Arch, En viron. Health 1980,35, 63-58. Received for review September 18,1980. Revised manuscript received June 29,1981, Accepted December 21,1981, 248 Environ. Set. Techrioi., Vol. 18, No. S, 1982 TEH 0470039 DUP050082859