Industrial sources of benzene exposure?

actions (e.g., exposure assessments, epidemiologic investigations, exposure registries, surveillance) that should be pursued in communities. Even with the efforts to measure exposure, the important question about latent, adverse health effects remains unanswered. As public health professionals in environmental health committed to protect the health of communities living near hazardous waste sites, we therefore strongly disagree with the authors' statement that "No further assessment of the health risks is needed." Legator and Strawn also make two other points to which we wish to respond. They state "If information on each site were available in sufficient detail, populations from exposed communities could be aggregated or combined. Unfortunately, the data that would help determine the multiple sites for which similar effects could be anticipated does not yet exist." The ATSDR agrees with the approach of combining populations from sites with reasonably common characteristics; this is the exact approach taken in our National Exposure Registry program (8). As an example, the ATSDR Subregistry of Persons Exposed to Trichloroethylene consists of a registry of about 5000 persons in 13 communities. Chemical-specific exposure subregistries provide ATSDR with health information on persons with common chemical exposures and also provide a means for communicating health information back to the registrants. In addition, more recently, ATSDR has developed the database necessary to combine site-specific information. The database is called HazDat. It contains all the environmental contamination, toxicology, and human health effects data from about 1300 Superfund sites. Recently, in conjunction with four state health departments, we conducted a study of lead exposure and toxicity in four different populations that were identified through use of HazDat. We anticipate releasing HazDat to the public later this year. Ascertaining the dangers to public health of hazardous waste sites, together with implementing public health actions to protect against the effects of hazardous substances, is a challenging responsibility. The ATSDR's public policies and public health practices must be based on sound scientific principles and data. This must involve the communities affected by releases from waste sites and other sources of hazardous substance releases. We believe the statutory mandates in the Compre-hensive Environmental Response, Comp-ensation, and Liability Act that bear on public health are consistent with sound public health practices. The translation of these mandates into actions, to some extent in ways inferred by Legator and Strawn, is ATSDR's challenge. We believe we have made progress, but much awaits.

actions (e.g., exposure assessments, epidemiologic investigations, exposure registries, surveillance) that should be pursued in communities. Even with the efforts to measure exposure, the important question about latent, adverse health effects remains unanswered. As public health professionals in environmental health committed to protect the health of communities living near hazardous waste sites, we therefore strongly disagree with the authors' statement that "No further assessment of the health risks is needed." Legator and Strawn also make two other points to which we wish to respond. They state "If information on each site were available in sufficient detail, populations from exposed communities could be aggregated or combined. Unfortunately, the data that would help determine the multiple sites for which similar effects could be anticipated does not yet exist." The ATSDR agrees with the approach of combining populations from sites with reasonably common characteristics; this is the exact approach taken in our National Exposure Registry program (8). As an example, the ATSDR Subregistry of Persons Exposed to Trichloroethylene consists of a registry of about 5000 persons in 13 communities. Chemical-specific exposure subregistries provide ATSDR with health information on persons with common chemical exposures and also provide a means for communicating health information back to the registrants.
In addition, more recently, ATSDR has developed the database necessary to combine site-specific information. The database is called HazDat. It contains all the environmental contamination, toxicology, and human health effects data from about 1300 Superfund sites. Recently, in conjunction with four state health departments, we conducted a study of lead exposure and toxicity in four different populations that were identified through use of HazDat. We anticipate releasing HazDat to the public later this year.
Ascertaining the dangers to public health of hazardous waste sites, together with implementing public health actions to protect against the effects of hazardous substances, is a challenging responsibility. The ATSDR's public policies and public health practices must be based on sound scientific principles and data. This must involve the communities affected by releases from waste sites and other sources of hazardous substance releases. We believe the statutory mandates in the Compre-hensive Environmental Response, Comp-ensation, and Liability Act that bear on public health are consis-tent with sound public health practices. The translation of these mandates into actions, to some extent in ways inferred by Legator and Strawn, is ATSDR's challenge. We believe we have made progress, but much awaits. In volume 82 of EHP, Wallace presented some of the results of the EPA's Total Exposure Assessment Methodology (TEAM) study in an attempt to identify the major sources of exposure to benzene (1). He contended that the results showed ". . . that personal activities or sources in the home far outweigh the contribution of outdoor air to human exposure to benzene" (1: 166). Two tables of statistical data were presented to demonstrate this point.
We have previously commented on the severe problems affecting the benzene data for New Jersey (2) and on the confounding effects of the apparent inversion that occurred concerning the data for Los Angeles, California ( 1: Tables 1 and 2; LAl). We believe that further comment is necessary regarding the North Carolina and Baltimore, Maryland, data, which are reported in the Wallace paper.
The North Carolina data presented in Wallace's Table 1 do not have an outdoor counterpart in Table 2. The reason for this is that only six fixed-site outdoor samples were obtained (3). The arithmetic mean benzene concentration of those six samples was about 19 pg/m3 for both day and night, or about twice the level found in the personal air samples (4). However, neither personal exposures nor outdoor levels of benzene in North Carolina should have appeared in the paper because of the extremely high and variable levels of benzene contamination on the Tenax sampling medium. The contamination was 193 ± 216 ng benzene/tube for both personal and outdoor air samples. Regarding this contamination, the EPA report (5) says "The benzene determinations should also be viewed with suspicion. . ." We agree and believe that none of the North Carolina data should be used to draw major conclusions.
The Maryland data shown in Table 1 of Wallace's paper represents only half of the available data from that portion of the study. Wallace reports here data from the segment of the study that was downwind of an industrial district. Another segment of the study, equal in size, from an area upwind of potential industrial sources has apparently not been reported except in the final report prepared for EPA (6). Table 1 compares data from the upwind segment of the study to data from the downwind segment of the study. Outdoor benzene levels are not reported because they were obtained by a different sampling technique.
There is no serious question about the values from the second group of data from Los Angeles (LA2) and from Antioch-Pittsburg, California (AP), but subsequent comments and conclusions regarding benzene exposure or breath differences should be reconsidered based only on results from the remaining total of 30 smokers and 89 nonsmokers. These remaining subjects can hardly be considered to be representative of the U.S. population. The correlations between passive smoking and benzene are very weak. This weakness is further demonstrated in another EPA report (7) that shows when the New Jersey and the California data for matched indoor and outdoor samples are regressed, only the first group of data for Los Angeles (LAI) show a significant correlation with the presence of a smoker in the home, and then only with p = 0.1 (probability that a smoker in the home was a significant variable). A later study conducted in Los Angeles (8), continuing primarily with the same homes at two different times of the year, was unable to show a significant difference between benzene in the air in the homes of smokers and those of nonsmokers. This result held true regardless of the season, the time of day, and the area of the house that was studied. What this later study did show was that the location of the outside samplers was important because there was not a good correlation between fixed or area samples and individual samplers located outside homes. This implies that the location of individual samples outside of homes is critical. We know of no published work in which this variable has been studied. Hence, when examining earlier data, the emphasis should be on matched indoor-outdoor results, and even then one should not be overly confident in the results.
Wallace's Figure 2, which compares West German data to U.S. data, appears to contain an error. Krause et al. (S) give the concentrations of benzene in West German homes as 9.3 and 6.9 KJg/m3 for smoking and nonsmoking, respectively, not 11 and 6.5 pg/mi3 as quoted by Wallace. We are also suspicious of the practice of comparing two different statistics, i.e., U.S. geometric means and West German medians.
We are aware of the breath levels of benzene in self-reported work exposure as discussed by Wallace et al. (10). Those results, obtained by the TEAM study in New Jersey when exhaust fumes infiltrated the van containing the spirometer, indicated that nonsmokers exposed to passive smoke more than 50% of the time at work could probably reduce their exposure by becoming smokers! Neither the experimental conditions during the New Jersey study nor the finding about the equivalent workplace exposure to benzene inspires much faith in the passive smoking conclusions from the TEAM study.
Readers with a need to incorporate the results of the TEAM study into their own findings would be well advised to critically review all of the TEAM study reports to determine when problems detracted from the significance of the study's conclusions and to what extent this occurred. Some important unanswered questions remain regarding the true impact of proximity to industrial sources, the potential for indoor sinks to ballast the effects of outdoor concentrations of benzene, whether smokers and nonsmokers have different lifestyles, and how representative these data are of the subjects' average day. Until these questions can be answered more conclusively, one can put little faith in risk analyses that use TEAM data.
We contend that the problems enumerated above invalidate the benzene exposures and risks shown in Wallace's Table 3. In addition to the problems with the appropriateness of the bases for the numbers in the calculations, examination of the numbers used in the exposure budget and risk analysis reveals some contradictory and unsupportable assumptions. For example, the text appears to say that two-thirds of the population are passive smokers, which we take to be 160 x 106 individuals. The footnotes to Table 3 imply that 80% of the population is exposed to environmental tobacco smoke, which we take to be 190 x 106. Table 3 claims a population at risk of 200 x 1o6. Footnote c of Table 3 and the text imply that the average increase in benzene due to environmental tobacco smoke is 3 pg/m3 for 17 hr spent at home and at work. But the data in the EPA report indicated that there was essentially no difference between the homes of smokers and nonsmokers in the second Los Angeles and Antioch-Pittsburg studies (10), and these are the only data not subject to serious questions. Finally, the variables presented in Wallace's Table 3