Tobacco Documents Online

ENDOENVIRONMENT, INC. AN ENVIRONMENTAL CONSULTING FIRM Commentary: "Environmental Tobacco Smoke: A Guide to Workplace Smoking Policies." EPA Public Review Draft Donald Rosebrook, Ph.D. EndoEnvironment, Inc. 38352 Henry Road Prairieville, LA 70769 My qualifications for commentary on this draft include a Ph.D. in Analytical Chemistry from Kansas State University in 1964 and the ensuing 26 years of experience in analytical and environmental chem- istry. I am the author of over 40 papers and presentations and an author or coauthor of a large number of EPA reports on environmental issues. I am currently a private consultant on environmental issues. Six months ago, I began a review and analysis of the issue of envi- ronmental exposures to benzene. As part of that review, I have extensively examined the EPA TEAM Study, the results of which are cited in your Review Draft (p.13). Specifically, my comments will be restricted to the following passage in the Review Draft, i.e.: Benzene, a hazardous air pollutant which is regulated in outdoor air by EPA because it causes leukemia in humans at occupa- tional levels, has heen found at average levels 504f• higher in homes with smokers than in homes without. People exposed to ETS at work over 50% of the Lime, have shown significantly higher breath concentrations of benzene than those exposed less often. For smok- ers, cigarettes are the greatest source of benzene exposure in the environment. For passive smokers, ETS is a significant source of benzene exposure. (p.13) While the authors of the Review Draft appear to be satisfied with an uncritical acceptance of results reported from a single study (reference 39, Wallace, et al.), careful analysis of the study itself reveals that the above passage is not supported by the data. Several observations can be made about the reference, Wallace, et al., and about the TEAM Study upon which the reference is based: 1 38352 HENRY ROAD. PRAIRIEVILLE. LA 70769 . (504) 673-4049

(1) While benzene has been identified under laboratory conditions in both mainstream smoke and sidestream smoke, its status as a constit- uent in ETS is unknown. However, two studies that examined total volatile organic components in offices reported no essential differ- ences in ambient air concentrations of volatiles in areas where smoking was permitted compared with nonsmoking areas.l (2) The claim that "People exposed to ETS at work over 50% of the time, have shown significantly higher breath concentrations of benzene than those exposed less often" is paraphrased from Wallace, et al (1987), and is based on results from 15 subjects claiming expo- sure over 50% of the time, using self-estimated and self-reported exposure levels. Samples, from a single trip to New Jersey, were taken with a van-mounted spirometer into which ben7ene is known to have permeatcd, and during which benzenc was collected on Tcnax cartridges admittedly contaminated with high and variable levels of benzene. The claim cannot be justified based on the data used in the cited reference; and certainly cannot be extrapolated to the general population as the Review Draft implies. (3) The TEAM study concluded that "Benzene levels in the homes con- taining2smokers were 30-50% higher than in nonsmoking house- holds." Even if the benzene levels presented in the report were accurately measured, this hardly constitutes the factor of two re- ported in the Review Draft. Additionally, that reference reports that the only chemical found at levels significantly higher (in breath and personal air) for smokers is styrene, and w}~en all sources are considered, the elevated levels are not significant j. (4) The reference paper cited in the Review Draft also contains the following remark: "Overnight indoor air exposures in homes with smokers were compared to concentrations in'homes with no smokers... the spring and early summer visits to Los Angeles and Antioch /Pittsburgh, California and the summer and winter visits to New Jersey showed no difference in concentrations."4 The authors of the cited reference also indicate that the summer and winter sampling trips to New Jersey did not yield any valid data for benzene because of contamination of the field blanks. The results of the measurement of benzene in breath in the Fall of 1981 in New Jersey (NJ1) were biased by the permeation of exhaust fumes into the van-mounted spirogneter. There were other problems with the spirometer during NJ1, I and a ge9 erally high and extremely varia- ble level of benzene contamination. Field blanks contained 97 t 64 ng benzene per cartridge and laboratory blanks contained 41 ± 26 ng benzene per cartridge. Such contamination introduces great uncer- tainty into the reported benzene levels. Cr7 ~ ~ ~ ~ 2 G~

The vglume of air sampled during NJ1 var~ed betwecn f& 10 and 30 liters. Converting these blanks to ug/m in the sampled air dis- closes that the field blanks may represent he equivalent of 1.10 - 16.08 ug/m3 of benzene in the sampled air.~ Wallace et a1.10 claim that daytime personal air exposures were 18 and 1I ug/m3 respec- tivcly for smokers and nonsmokcrs. With the indicated level of ben- zene contamination of the field blanks, it is not realistic to claim that the benzene in the personal air exposures of smokers and non- smokers is significantly different. Benzene measurements during the February, 1984 Los Angeles (LA1) sampling are also suspect because of undesirably high recovery of benzene from field controls (high bias) and field blanks (contamina- tion). The benzene contamination in two batches of Tenax used f r personal air sampling during LA 1 was 22/28 -ng with 48/69 'bRSD.~I Information on the absolute range of contamination and on the range of volumes sampled could not be found in the report; however, approximate values of benzene vapor equivalency can be calculated based on certain assumptions.12 The appr3 ximated maximum ben- zene vapor equivalency would 3.15 ug/m and the approximated minimum would be 0.347 ug/m3. In addition, recoveries of benzenc-d6 from control cartridges showed that benzene recoveries would be biased high by as much as 30% (Pellizzari 1987a). Under these circumstances, it seems probable that two relatively major corrections would have to be made for the personal air sample results. The benzene background and recovery studies from the Tenax batch used for breath sampling exhibited similar, though slightly lesser difficulties. Benzene measurements made during LAI are atypical because an inversion caused excessively high levels of benzene to accumulate in the environment. These levels of outdoor benzene were substa tially higher (4 - 8x) than found during the other California trips.l~ The outdoor benzene levels apparently had a direct effect on the indoor levels of benzene such that LA 1 indoor benzene concentrations were 2x the other California-trips.14 Yet, the breath levels were• essen- tially the same in all studies.15 Since breath levels remained con- stant, this would seem to belie any connection between indoor levels and breath levels for benzene. (5) The Review Draft states that "For smokers, cigarettes are the great- est source of benzene posure in the environment." On the con- trary, the TEAM report~~ indicates that being a smoker is not sig- nificantly correlated to benzene in the persortal air. The total number of smokers in that part of ihe TEAM studies purported to produce valid results for benzene17 was 209. Consid- ering the myriad of problems with the data for NJ1 studies as dis- cussed above, those data should not be considered as part of the data base for benzene in personal air and breath. Removal of the 150 3

NJl smokers leaves 59 individuals to represent the population of smokers in the US. In addition, the 11 individuals from LA2 consti- tute duplicates of LAI rather than additional subjects, thus leaving a final total of 48 individuals to represent the smoking population of the US. Even without the technical problems found in LA2, it is an egregious assumption that 48 smokers can accurately represent all US smokers. Without the flawed results from benzene measurements in NJ1, the results from the California study of less than 200 individuals can hardly be construed to be representative of all of the various strata in the US. The number of subjects in any one stratum of the resid- ual portion of the TEAM study, as designed, is not sufficiently large to represent a cross section of the population of the US and thus can be represented only as a very carefully selected group of indi- viduals whose environmental and respiratory gas contaminant con- ccntration was measured using contaminated sampling media. (6) The results presented in Wallace et al.(1987) comparing benzene in the breath.of smokers and nonsmokers are unweighted geometric means. The use of unweighted statistics in the presentation of the comparison of smokers and nonsmokers requires more explanation, because the samplin$8 of the various populations was weighted for a variety of factors. The use of unweighted statistics will not accurately represent the sampled population. The TEAM study reported that eb or eci is a quantitative measure of the strength of the association between any variable and exposure to it (where b or ci are ratios of the geometric means of concentra- tions found for exposed persons to those for unexposed persons). Given the skewed distributions of the results of the studies, the ratios of the geometric means can hardly be considered to represent the ratios of the "average levels" as purported in the Review Draft. (7) The information regarding the level of benzene exposure in homes with and without smoke16 must be viewed in the context of other results of the same stud-y: that those who had painting as a• hobby would be exposed to only 60% of the daytime styrene of those who did not; that having a gardener in the home would reduce average exposure to daytime m-xylene by 30%, and overnight benzene by 40%; that being a garage worker would reduce average daytime exposure to chloroform and to meta- & para- dichlorobenzene by a factor of 3; and, that having a monitor located outside the residence would reduce daytime exposure to carbon tetrachloride by 33%. Regression analysis as performed in the TEAM study and reported in Wallace (1987) will often produce anomalous, even ridiculous, answers. Some of these interesting relationships in the present case have relatively large sample sizes, e.g., those who travel in a car 4

during the daytime hours are exposed to only 75% of the m,p-dichlo- robenzenes as those who do not; the sample size is 174 individuals. All of the relationships cited earlier were found to be significant by the statistical methods employed. Experience teaches us that if one starts out with enough variables and few enough samples almost any kind of relationship between variables may result. Experience should also teach us to carefully consider all of the raw data in- volved in such analysis. Consideration of all of the facts must result in discounting the value of the smoker/nonsmoker relationship. Conclusion: For these reasons, it is a disservice to both science and the American public to uncritically represent the results of a single report as established fact. The reference to ETS as a significant source of benzene in the environment is completely unwarranted from a scientific point of view and serves only to increase anxie- ty in the general population. ~ ~ ~ ~ ~ 5 6D

NOTES (1)Bayer, C.W. and Black, M.S., 1987, "Thermal Desorption/Gas Chromatographic/Mass Spectrometric Analysis of Volatile Organic Compounds in the Offices of Smokers and Nonsmokers," Biomedical and Environmental Mass Spectrometry 14(8): 363-367. Hollowell, C.D. and Miksch, R.R., 1981, "Sources and Concentration of Organic Com- pounds in Indoor Environments," Bulletin New York Academy Medicine 57(10): 962-977. (2) Wallace (1987), p 104 (3) Wallace, L.A. (1987), tables 39 and 40. (4) Wallace, et al (1987), p 274 (5) Wallace (1987), p 273, footnote to Table 1; and Pellizzari, et al (1987), p 149. Accord- ing to Pellizzari, participants had great difficulty using the spirometcr, did not keep their mouth attached to the mouthpiece, and often could not fill the bag completely. Unfilled bags mean smaller samples and greater impact of the background contamination on the Tenax. (6) Pellizzari, et al (1987), p. 150. No spirometer blanks were taken during the first season and therefore no blank correction could be made. (7) Wallace, L.A. 1987. "Total Exposure Assessment Methodology (TEAM) Study: Summary and Analysis," Vol 1. Washington, DC: United States Environmental Protection Agency. Table 8, p22. (8) Pellizzari, E.D. 1987. "Total Exposure Assessment Methodology (TEAM) Study: Elizabeth & Bayonne, NJ; Devils Lake, ND; & Greensboro, NC," Vol 2, part 1. Washing- ton, DC: United States Environmental Protection Agency. (page 149) (9) 97 - 64 ng x I ug/ 1000 ng -------------------------------- - 1.10 n g/ m 3 30 L x 1 m3/1000 L 97 + 64 ng x 1 ug/1000 ng -------------------------------- - 16.08 n g/ m 3 10 L x 1 m3/1000 L 6

(10) Wallace, L.A., et al. 1987 "Exposures to Benzene and Other Volatile Compounds from Active and Passive Smoking," Arch. Environ. Health 42 (5): pp 272-279. (11) Pellizzari, E.D., 1987a. "Total Exposure Assessment Methodology (TEAM) Study: Selected Communities in Northern and Southern California," Vol. 3. Washington, DC: United States Environmental Protection Agency. Table 52, page 157. (12) The minimum range of benzene equivalency among measured values can be estimat- ed by assuming the range of sample volumes was 15 - 25 L (a large improvement over the first NJ trip), and that the mean value times 1± 96RSD/100 divided by the low (high) sample volume equals the maximum (minimum) benzene equivalency (a conservative assumption). (13) Wallace (1987), p 55. A table presenting an unknown statistic indicates the "weight- ed estimates of ... concentrations" for benzene in outside air were 16 for LAI; 3.6 for LA2; and 1.9 for CC. (These are Tost probably arithmetic or geometric means of concen- trations expressed as ppb or ug/m .) (14) Wallace (1987), p 55, Table 27. Indoor levels for the three California trips were 18, 9.2, and 7.5 ug/m3. (15) Wallace (1987), p 55, Table 27. Breath levels of benzene were 8, 8.8, and 7.0 ug/m3 for the three California trips. These values would be the same within the errors de- scribed for the TEAM study. (16) Wallace (1987), Appendix A, Table A-1. "Smoker in the home" is interpreted to include smokers and those who live with smokers. The correlations with benzene in personal air are purported to be significant for smoker in the home but not for smoker. Again however these tables use the highly suspect benzene data from the Fall of 1981 in New Jersey. (17) Wallace, et al. (1987), Table 1. (18) Wallace (1987), pp 14-16. (19) Wallace (1987), Appendix A, Tables A-1 and A-2. 7