Tobacco Documents Online

Comments by Dr. Guy B. Oldaker III on CHAPTER 7 EXPOSURE ASSESSMENT IN PASSIVE SMOKING f -a General Comments This chapter presents a model for estimating exposures to respirable suspended particles associated with ETS. This model has as its major assumption that ETS in rooms being modeled is in an equilibrium sti,te. This assumption of equilibrium has not been demonstrated with experimental measurements nor is it consistent with common experience. Thus, the model aisumes that smoke is evenly distributed, ventilation likewise is even, there are no drafts, and there are no temperature gradients from floor to ceiling. Anyoie who has observed a smoke plume rising from a cigarette knows these assumptions are not generally true in real-world settings. Despite the author's assertions that the model has been validated, this is not the case for the application of the model to field measurements. The literature shows only that the model can be fit to data obt<<ined in nonrepresentative experimental settings where unrealistically high levels of ETS RSP are produced with ventilation conditions maintained to ensure thorough mixing and therefore the necessary equilibrium condition. Thus, when equilibrium conditions are forced, the model applies, as it should based upon elementary physical principles. The author seemingly overlooks the basic scientific paradigm for validating a model. According to this paradigm, the model and its assumptions are defined for the experimental setting to be modeled. The model is then used to predict the results of experimental measurements. Then, the experimental measurements are performed after which these are compared to the predictions of the model. The validity of the model is established based upon quantitative, measures of agreement for the experimental settinas of interest. Gr) Ch . ~ I L W

The author's model has not been through this paradigm. Instead, the author approaches the validation process in reverse so that the model is, in effect, used to postdict rather than to predict results from measurements performed in the field. Because the input parameters of the model contain so many assumptions and admit such wide ranges of reasonable values (NAS, 1986), the model can be easily (and "reasonably") adjusted to produce results that agree with any set of environmental measurements. A model that predicts everything, predicts nothing. L L U L- The author provides several examples of experimental results for which he.can adjust the model to produce similar results. A subAantial body of research, including more reliable RSP data, exists providing results which are inconsistent with the some of the fundamental parameters assumed for the model. The author fails to recognize this research. To those who would use this chapter, Mr. Repace is potentially doing a great disservice, because he has neglected to address adequately the assumptions of the model. By its very nature, modeling is the cost effective alternative to measurement; thus, the many of the potential users of this compendium would be expected to perform modeling in lieu of making measilrements themselves. Such users will probably not recognize the assumptions a,:tending use of this model and will instead calculate numbers and assume that Mr. Repace has done the thinking for them. In his exposition, Mr. Repace has not adequately developed and presented his assumptions nor has he presented the logical pathways connecting his selection of values for input parameters, their calculation, and their interpretation. Moreover, Mr. Repace has presented a model which represents the ideal, rather than the real. Some information presented by the author of Chapter 7 is inconsistent with information presented by the authors of Chapters 5 and 6. This reviewer recommends that Mr. Repace consult with these authors to ensure that the Compendium is internally consistent. Some of Mr. Repace's literature citations are not sufficiently specific, thus, making it impractical for the reader to return to the original source to obtain suppofting or supplementary information. The author should at a minimum provide page numbers in connection with: "NRC, 1986"; "IARC, 1987"; and "SG, 1986." The author also should consider citing the IARC publication in greater detail. Mr. 2 L

Repace is the author of the chapter in-the IARC publication that contains most if not all of the material being cited in Chapter 7 of the Compendium. Although brevity and modesty are important considerations he're, it would be far better to identify the author, since some readers may be apt to assume incorrectly that the literature citation constitutes approval of the model by that organization rather than simply publication. Indeed, the IARC (1987, page ii) makes it clear that they do not necessarily support the model: "The authors alone are responsible for the views expressed in the signed articles in this publication." This reviewer strongly recommends that Chapter 7 be omitted from the Compendium because the model, which is the basis for the Chapter, has neither been validated nor.is it consistent with results from aasessments of ETS exposure done in real-world settings. Specific comments for each of the sections of Chapter 7 are provided below. Specific Comments Introduction L u The first sentence of the Introduction accurately presents the current situation regarding•assessment of exposure to ETS: There are currently no direct measures of the dose of ETS absorbed in a population under study; however, expo:;ures to ETS can be assessed by personal air contaminant monitoring, modeling of concentrations (based upon air sampling, time activity patterns, and questionnaires), or biological markers. (NRC, 1986) This sentence, however, does.not include the use of modeling by itself. Mr. Repace states that the concentration [of ETS] is directly proportional to the product of the number of smokers, smoking rate, and emissions per tobacco product, etc. This statement represents the theoretical rather than the real 3

world situation. The author should revise the paragraph to make this distinction. The author states that it has been shown that those nonsmokers who report exposure to ETS at home tend to have higher non-domes-tic exposures as well. The author should identify how the non-domestic exposures were assessed. Mr. Repace states that a majority of studies has used RSP as the indicator of ETS exposure. This is not supported by the record of the scientific literature. In fact, RSP, because of its lack of specificity foi•.ETS has been used in the minority of studies. The author fails to recognize tliose surveys which have not used RSP as the indicator nor does he make it clear to the reader why the majority of studies did not use RSP as the inJicator: RSP will always overestimate ETS exposure unless some means is takel to apportion for sources of RSP other than ETS. Other indicators include ultraviolet particulate matter (UVPM, an upper estimate of the contribution of ETS to RSP (e.g., Carson, 1988), fluorescent particulate matter (FPM, a complement to UVPM (Ogden, TCRC 1989), and solanesol ((Ogden, Environ. Sci. Technol. 1989), an indicator specific to ETS RSP). Results from surveys including these indicators show that ETS RSP constitutes on average approximately 50 9'e of the indoor RSP (Oldaker, 1987), significantly and substantially less than the 85 % assumed by the author throughout the text. In addition, the author defines the size range of RSP to be < 2.5 gm; this size range is inconsistent with the size range used by the methods of the "majority of studies" referred to by the author: 3.5 um. By this definition, the experimental method used by the author and Dr. Lowrey (Repace and Lowrey, 1980) is inappropriate for determining RSP. Although Mr. Repace recognizes that the currently accepted cutpoint for defining RSP is 2.5 µm, he fails to inform the reader that the method used Mr. Lowrey and him were performed with a device having a cutpoint of 3.5 µm. The accuracy of 'this device is critically important because it provided all the supporting data from the field for the model. Additionally, Repace and Lowrey (1980) failed to address quality assurance activities taken when measurements of RSP were made. Ingebrethsen and coworkers (1988) have demonstrated that quality assurance activities are a necessary condition for obtaining reliable data froi devices such as the type 05 used by Repace and Lowrey. The author should addre::s the effect of the 3.5 ,im ~ ~ N ~ 4

iji C ( cutpoint on the model and the quality assurance measures taken in connection with the original measurements. At he bottom of page 79, Mr. Repace presents an ideal approach to assessing exposure which involves quantifying "the several thousand compounds in tobacco smoke"; he adds that the enormity of this task has led to simpler approaches. These statements misrepresent the science. An estimated 100 compounds have been identified in the ETS matrix. Because most of these compounds exist at but trace levels they are unsuited for use as indicators, thus simplifying the task of exposure assessment. The author should revise these sentences to present accurately the current science relative to assessing exposure to ETS. A. Sources of ETS t In presenting examples supporting the model, the author employs an argument that is logically incomplete and seemingly inconsistent. Mr. Repace presents results from measurements of nicotine or cotinine in body f'iuids, and by manipulating assumptions he is able to show that the model can "predict" RSP concentrations that are comparable to those from measurements of nicotine and cotinine in body fluids. If this relationship between RSP and nicotine and cotinine in body fluids is valid, then a similar relationship must exist between RSP and airborne ETS nicotine, thus implying that nicotine should be an ETS indicator on equal footing with RSP. If the example including nicotine and cotinine measurements is to be used, the author should address the relationship between RSP and airborne nicotine. Readers of Chapter 7 cannot use the model intelligently because the text does not provide enough detail about the parameters appearing in the model and the assumptions attending the parameters. For example, Mr. Repace states as fact that the average U.S. smoker smokes 32 cigarettes per day at a rate of 2 cigarettes per hour. This statement appears to be only the result of simple arithmetic. This smoking rate is one critical inpLt parameter to the model. In view of its importance it is absolutely essential that the*reader know: (a) the origin of data used for calculation, (b) the deriaation of the smoking rate, and (c) assumptions made and the likely effect on results. It is this reviewer's understanding that these values are obtained through elementary arithmetic operations drawing from the reported number of smokers in the U.S. in 1986, the number of cigarettes sold that year, and the number of hours 5 dD ~ N G't N ~

available in an average day in'which the cigarettes can be smoked, specifically, 16 hours. If this understanding is correct, the z;uthor is assuming that a cigarette sold is a cigarette smoked; common experienc:e tells us that the former will be greater than the latter. The effect of sales practices employed by the industry on these values is unknown. Additionally, the author is assuming that smokers and, more importantly, nonsmokers will report their status accurately. Again, common experience tells us that the number of actual smokers will be greater than the number reported. Based upon these considerations, one can assume that smoking rates assumed for the model might be biased high. However, without additional information; this reviewer and Mr. Repace can only speculate on the quality of the assumed estimates of smoking rates. Furthermore, Mr. Repace assumes that the average smoker consumes cigarettes at a constant'rate be it on the basis of times figured yearly, daily, or hourly, and that this constant rate of consumption is scaled proportionally to a sixteen-hour day. Information on the temporal variability of smoking rates in the U.S. population is not provided; the reviewer is unaware of such information. f L u Next Mr. Repace uses the observation that the average person spends 90 9'e of their time indoors and by extension the average sm3ker smokes 90 % of their cigarettes indoors. No information is available reyarding where people smoke be it indoors or out. Again, common experience suggests that. because of social concerns, smoking frequency will tend to be shiftec to outdoors. Assuming a homogeneous smoking population is inaccurate. Demographics tell us that smoking i s more preval ent among l ower and l ower middl e i ncome groups than among the upper middle and upper income groups. These groups where smoking is expected to be more prevalent are those also where we can expect a greater probability of occupations th'at entail outdoor work. In contrac'iction to this exposition, the author fails to adjust the smoking parameter for the fraction of time the population spends indoors. From these assumptions Mr. Repace derives a parameter that predicts that in any indoor setting with more than nine occupants and where smoking is permitted, 0.111 cigarettes should be observed smoking per person at any instant when observations are made. The author fails to provide any visual observations made to confirm this assumption in spite of the fact that this parameter is critical to the model and is easily measured. Much additional effort needs to be expended to assess the distribution of the values of this parameter for the microenvironments of interest. 6 L

Mr. Repace also notes that the percentage of the smoking population .has decreased over the last 30 years and then speculates that the increase in smoking rate may tend to offset the trend to lowering nonsmoker exposure to ETS. Mr. Repace has no data to address this issue. In the absence of such data, Mr. Repace should inform the reader that the smoking rate used by the model must be calculated from data for the year of interest. B. Indoor air transport of ETS L. ~. The author states that equilibrium models are best suited to evaluating and predicting ETS concentrations in field studies, particularly when average concentrations over a period of days or longer are of interest. This statement implies that equilibrium models are not particularly suited for applications with time frames less than "days." The author must provide the reader with quantitative criteria describing the time frames for which the model is suited. Scientifically, the implication that the model is unsuited for short time frames implies that the model is either invalid or suffers from great imprecision. Mr. Repace should address these issues. The use of an equilibrium model for evaluation purposes assumes that the model*.has been validated to the extent that it has been demonstrated to show agreement with results from field studies: the model employed by Mr. Repace has yet to be evaluated in this context. Indeed, common sense says this is so, for otherwise there would be no justification for the ongoing field studies being done to assess exposure to ETS. Mr. Repace states that Leaderer (1984) has given 3 detailed review of [the equilibrium] model. Leaderer did not. The NRC (1986) is the best reference and should be used instead. The author states that the most extensive use of the mass-balance equation for assessing RSP levels due to ETS in occupied spaces ias been due to Repace and Lowrey. To this reviewer's knowledge, only Repace arid Lowrey have applied this model outside of chamber settings. The author should make this clear. The au-thor notes that the model assumes equilibrium conditions and that errors are introduced when any of the generation or removal terms are intermittent. The generation of ETS is fundamentally an intermittent process. Such 7

l L1 intermittency is fundamental because the productior of ETS derives from twoo discrete parameters: (a) a smoker must exist, and (b) the smoker smokes but one cigarette at a time. Hence, the equilibrium model is fundamentally in error unless the number of smokers and number of cigarettes smoked per unit time is sufficiently large. To this end, Mr. Repace assume:: that nine occupants in a room will produce the condition required for equilibrium, specifically, that at any instant, one cigarette is burning. No error analysis has been presented in connection with the equilibrium model. Logically, t1is condition implies that the model is applicable only to rooms occupied by nine or more people, otherwise, the model will predict erroneously. The author should provide the reader with- estimates of the probable errors which would be introduced if fewer than nine persons were in the room to be modeled. Firthermore, this condition implies that the "derivative equations" presented by the author will predict erroneously. (This reviewer suspects that the utility of the model is severely limited because the number of places occupied by nine or more people is relatively few.) The model presented by the author at equation (2) has no derivation for the constant term, 650. Derivation is important to the user because without it the user is unaware of assumptions being made. Thus, the term includes assumed smoking rates, RSP emission rates, mixing factors, ventilation rates, and sink rates. As noted previously, quality data on smoking rates are unavailable; such rates, however, could be determined experimentally ir each setting of interest. The users of the model could determine this information themselves, rather than assuming that the value given by Mr. Repace was accurate. RSP emission rates have not been established, primarily because the methods necessary for determining this parameter have not been applied consistently. Sink rates for RSP also have not been quantified; the absence of this information explains why investigators are researching this area. Finally, raixing factors are for all practical purposes the greatest, and therefore the most important, unknown. Common experience tells us that smoke is seldom distributed homogeneously within an air space. This lack of homogeneity explains why Leaderer and McCarthy in Chapters 5 and 6, respectively, correctly recognize that ETS concentrations (including RSP) exhibit temporal and spatial va-iability. This spatial variability has bearing on another issue relating to RSP exposure. Put simply, many persons potentially exposed can and do move away from the source of the smoke, an action which directly reflects the importance of the mixing factor in interpreting predicted exposures. This reviewer strongly recommends that the 8

author provide equations which allow users to supply values for as many of the input parameters as possible. The author speculates that gathering data on ... smoking rates or volume can substantially reduce the variability of the estimated RSP levels. This statement must be supported by an error analysis performed on the model. Additionally, Mr. Repace does not provide the reader with the means of applying such data to the equilibrium model. Users of the Compendium can be expected to forgo such measurements for convenience and instead rely on the model as presented in equation (2). Of all the parameters, the mixing factor is probably the only one which, if• refined with experimental measurements, would effect substantial reductions inn the variability of prediction. As those who assess exposure by measurement (as distinguished from modeling) know, actual measurement of exposure is far more easily performed than measurement of the mixing factor. Again, this explains why researchers continue to pursue exposure assessment rather than modeling. The National Research Council (NRC, 1986) correctly recognizes these limitations when they conclude that a better understanding of the variability of the input parameters is needed. Mr. Repace states that the International Agency for Research on Cancer (IARC) has published derivatives of equation (2) which incorporate advances in understanding. Readers may be misled into believing that publication reflects adequacy and acceptance. The author should provide a more complete citation, which would inform the reader that the author of this publication is Mr. Repace, presenting only elaborati.ons on the same theme. The author's statement that the derivatives of equation (2) incorporate advances in understanding is incorrect. In the IARC publication, the author has used only :;imple arithmetic to scale equation (2) for the number of observed smokers; such scaling violates the condition necessary for equilibrium and therefore is scientifically invalid. Mr. Repace attempts to show the "validity" of his model through use of data reported by Spengler and coworkers from the "Six City Study." Because his model includes so many assumptions with undefined ranges of values he is able (as anyone would) to use simple arithmetic to show agreement. He assumes that two "habitual smokers" in a household would double the ISP level from 20 ,%g/m3 to 40 µg/m3. He then assumes that if the 24-hour RSP average concentration is 40 ug/m3 then the 16-hour average would be 60 gg/m3. Wext, he assumes an average air exchange rate for the heating season for typical middle income housing and 9

asks the reader to trust a personal communication. This air exchange rate derives from a study of "older middle class homes" as distinguished from "typical middle income housing." (The reader must assume that the "Six City Study" included middle income housing.) He notes that these air exchange value were obtained during tests when the occupants were asked to keep windows and doors closed. Besides being nonrepresentative, these conditions would reduce measured ventilation rates thus leading the model to predict higher concentrations of RSP. ( L L Mr. Repace refers to time-budget studies in connection with assuming the number of cigarettes smoked per day in the homes. A citation to these studies should be provided. _ Mr. Repace next shows that by assuming that the air exchange rate is "only a third of a standard deviation from the 14-city mean' the RSP levels can be fit exactly. The author concludes by stating: "This example illustrates the utility of models in estimating nonsmokers' exposures to ETS." No; rather, this example illustrates only that by making a host of assumptions, agreement can be achieved. Mr. Repace does not provide enough information for the reader to understand the steps taken in the arithmetic. Mr. Repace should provide the 'process by which,he arrives at the value for the density of habitual smokers, Dhs, as well as the attendant assumptions. For example, it is unclear how the volumes within the household were obtained and what he means by a "two smoker" home. Mr. Repace states that the utility of equation 4 depends on the assumption of an air exchange •rate. This statement is misleading as well as untrue. The utility of equation 4 depends on a host of assumpticns many of which are buried in the "constant coefficient." In this same vein, Mr. Repace concludes that "this example illustrates the utility of models in estimating nonsmokers' exposure to ETS." Logically, the reader must return to the topic sentence of the paragraph; in so doing the reader is confronted with a circular argument, namely, that the illustrated utility of the model depends on the assumption of an air exchange rate. The circular argument illustrates only one truth: the model predictions can be fit to experimental data provided enough assumptions can be made. 10

Mr. Repace also concludes that his illustration provides results consistent with those shown in figure 5 for the "Six City Study." Without some quantitative information the subject of consistency admits broad interpretation, with one possible interpretation being that there. is no adequate degree of consistency to justify use of the model. One can reasonably assume that the heating season includes the months from October to March. To be valid and therefore useful the model must be able to predict the large variability shown in the results for the cases of one smoker and more than one smoker. It is important to note that the categories for the field results differ from those of the model results thus constraining the ability to compare between the two. Taking March 1978 as an example, field measurements show a mean RSP level of 35 µg/m3; by contrast the model predicts 44 µg/m3, a bias of 26 %. Mr. Repace apparently attaches significance to his prediction of 84 ug/m3 for his :,cenario of three "habitual smokers." How this result is consistent with a measured result for "greater than one smoker" (whether this average smoker is "habitual" is unknown) is unclear. 6 Mr. Repace provides another example to "validate" the model. He starts by noting the "large impact" caused by smoking on the levels of RSP. This "large impact" is 10 ug/m3 for an assumed occupation of the room by one "habitual smoker." If the author is to make such a statement it would be instructive to know what represents a "low impact." Mr. Repace next proceeds to calculate the RSP level for the case of a chain smoker, one, who according to Mr. Repace, is consuming cigarettes at a rate of six cigarettes per hour. The reader can only assume the purpose of this example, since its representat'iveness is unknown and since it delves into the world of the bizarre. Is the reader to assume that this smoking rate continues (as it must if the model is to be appropriate as defined by Mr. Repace)? If this assumption is taken to its illogical end, the reader must conclude that the person is smoking almost five packs of cigarettes per day and gobbling food between puffs. In the same paragraph Mr. Repace notes that ETS can be very persistent, taking, for example, 3 hours for 95 % removal. The reader can only speculate why this factor was riot taken into account when Mr. Repace "modeled" RSP in the forgoing example dealing with homes. Instead he assumes that RSP exists only during the 16 hours he associates with the "habitual smoker" being awake. . Cb Mr. Repace then shares with the reader "several interesting factors" revealed ~ by recent research. He relies on a personal communication to provide this N Ctt 11 W

i information. Although it is certainly possible that floor-to-floor communication of RSP can occur, the significance of such communication and its representativeness are unknown. Mr. Repace is specul,iting; the paragraph should either be omitted or the above issue should be addressed with scientific data. The paragraph discussing work by Williams and coworkers should be omitted. Green and coworkers (1985) published a letter (In the same journal that contained the paper by Williams and coworkers) that discredited the method, results, and conclusions reported by Williams and coworkers. Following his citation- of the paper by Williams and coworkers, Mr. Repace presents results from surveys of air exchange rates done in offices. This information does not fit within the paragraph. P: can be assumed that the author might be attempting to make a point about values for air exchange rates in offices; however, the information he provides, specifically, only mean air exchange rates, is insufficient to be useful for one interested in assessing the assumption of his model. The information on air exchange rates might be useful if ranges and standard deviations were provided. Mr. Repace's summary to Section B is unsupported. To his credit he notes that "limited" field tests of the general equilibrium mcdel have been done. This reviewer is concerned that Mr. Repace may be statirn3 that the two examples he provides in the section represent the limited field i:ests. If these indeed are the limited field tests, then Mr. Repace is not applying the scientific paradigm, which would require that the model predictions be performed before the field measurement rather than after as he has done in all his manuscripts where the model is used. On the contrary, the model does not predict RSP levels reasonably well, rather, the model can be manipulated so that its "predictions" agree with results from field studies selected by author. Rather than being "clear that both models and observations ... yield consistent results," it is clear that whether they can or cannot is open to question. Mr. Repace's concluding statement in this paragraph is at odds with the data he uses to "validate" his model. If the statement "RSP levels wnen smoking is allowed will result in substantial increases over RSP levels ir nonsmoking occupancy" is true, how can it be that the Data of Figure 5 show essentially no differences among RSP levels in June 1977 for "outdoor," "indoor," "no smokers," and "indoor one smoker"? The reader deserves to know the meaning of "substantial" in quantitative terms. 12

I C. Measured concentrations of RSP From ETS t L 0 L Figure 8, which shows results from RSP measurements arid predictions and National Ambient Air Quality Standards (NAAQS) is inappropri<<te and should be omitted. Users of the Compendium can be expected to misunderstand and misinterpret this presentation. While there are several reasons why inclusion of NAAQS levels is inappropriate, the major one deals with averaging times. Assuming all other factors the same, the averaging times for the measurements and the standards must be the same for comparisons to be meaningful, as distinguished from appropriate. The U.S. EPA has algorithms which the author should use to adjust for averaging times. Even if all information was Fresented on the same time scale, discussions of exceedances would be meaningless, since from a regulatory standpoint criteria documents, standards, and reference methods do not exist. The data reported by Repace and Lowrey are on the timia scale of minutes. If the standards were adjusted to that same basis, it is reasonable to assume that all these data would be below adjusted "standards." If this presentation is to be included in the Compendium, the standards should be adjusted to time scales equivalent to those of the measured RSP values. Associating the Table with IARC is unnecessary and misleading; only the first published reference need be identified: Repace and Lowrey, 1980. The quality of the data reported by Repace and Lowrey in Figure 8 is unknown; nonetheless, it is known that these results are biased high because no apportionment was made for sources of RSP other than ETS. The degree of Has also is unknown. The representativeness of the sample population is qui!stionable. Results from surveys conducted since 1980 (Repace and Lowrey's experimental approach cannot properly be categorized as a survey) indicate that '2epace and Lowrey's sample population is probably nonrepresentative. Mr. Repace infers from his few measurements of questionable quality and representativeness that "the bulk of the RSP found in buildings where there is smoking is due to ETS." By selective referencing he is able to suggest that 85% of the indoor RSP in those buildings is due to ETS. Since 1980, surveys done in offices show mean RSP levels of approximately 125 µg/m3, a result indicating that Repace and Lowery's data base is probably biased. In addition, methods capable.of apportioning ETS have shown that less than half of the RSP found from surveys can be attributed to ETS (Ogden, TCRC 1989). Again, the data base of Repace and Lowrey is indicated to be biased. 13 L

Table 8 contains mostly old literature citations. Many more investigations have been conducted since this table was published in 1986. In addition, the reference to the U.S. Department of Transportation study done in aircraft cabins is incomplete since results were provided in that report. The mean level of RSP in cabins before smokers were segregated was only 40 ug/m3. (The reviewer recommends that Mr. Repace address this result with the equilibrium model.) Mr. Repace notes that in a setting such as a work environment, where the average exposure is several hours, ETS would be expected to disseminate throughout the airspace where smoking is occurring. Inasmuch as no data are available to support this speculation, the author should drop thi> statement from the text. Similarly, 'Mr. Repace speculates about exposures in transit and selectively references his paper (Repace, 1988) in connection with ETS in aircraft. Mr. Repace omits all otfier references surrounding his paper (Oldaker, Environ. Sci. Technol. 1987, 1988) and fails to inform the reader i:hat his position is based on an assumed relationship between RSP and nicotine. Specifically, Mr. Repace assumes that a constant ratio exists between the two indicators thereby allowing him to predict RSP from nicotine. Oldaker (Environ. Sci. Technol. 1988) has addressed some of the assumptions contained in this approach and Oldaker and coworkers (Excerpta Medica Interantional Congress Series 1989) have presented results showing that ratios have no predictive value. D. Exposure of Nonsmoking populations to ETS U The first paragraph of this section addresses risk, a subject which is inappropriate for•inclusion in view of the absence of risk assessment results. Mr. Repace identifies the Seventh Day Adventists as a subpopulation where few of its members are nonsmokers. The connection between this population and lung cancer incidence is unclear; the author should clarify the importance of this subgroup within the discussion. The author identifies cotinine as an indicator of exposure to ETS. Although cotinine is related to nicotine exposure because cotinine is a nicotine metabolite, the relation between cotinine levels in body fluids and exposure to ETS, as distinguished from nicotine, is currently unknown. Some investigators assume that the presence of cotinine in body fluids provides a lower estimate of exposure to ETS because of results from experiments in environmental chambers 14 Vv

that show nicotine to decay more rapidly than RSP. Although Eudy and coworkers (1987) have found appreciable background levels of nicotine in the absence of ETS in one setting, systematic investigations of this phenomenon have not been performed. Nonetheless, because nicotine can adsorb as well as desorb from surfaces of rooms, it is reasonable to assume that appreciable background levels of nicotine might exist in typical real-world settings, thus admitting the possibility that nicotine might overestimate exposure to ETS. In addition, it is known that this background can persist in the absence of RSP from ETS (Eudy, 1987), the implication being that cotinine might indicate only that exposure to nicotine has occurred, or, stated another way, tha1: nonsmokers have occupied spaces where smokers have been but where no ETS currently exists. The author notes: "The forgoing illustrates that exposure to ETS is very widespread in the population."•Based upon the uncertainties raised in the above paragraph, it is perhaps more accurate to state that exposure "appears" to be widespread. The author notes: "However, additional data on the Jistribution of smokers in the nonsmokers' environment as well as the distribLtion of ETS levels in the environment are needed in order to characterize the <<ctual ETS exposures of the population." This statement, which represents an a:curate description of the current understanding of ETS exposure, directly speaks to one of the assumptions of the equilibrium model: the distribution of smokers in nonsmokers' environments is unknown (this information is contained in the assumed term Mr. Repace identifies with Dhs). In addition, this statement speaks to the quality of the equilibrium model: if estimates provided by the model provided results of adequate quali'ty, then characterization of actual ETS exposures would be unnecessary. The author follows with the statement that in the absence of such data, population exposures can be estimated by models or by extrapolation from biological markers. Clearly both these approaches are valid, but only within the context of the statement; however, the quality of such estimates remains vague because the issue is not addressed within the text. The reviewer recommends that the author provide some information addressing the quality of such estimates. 15

E. Integrated Exposure Analysis 1J, L L I The author states: "Estimating the magnitude of the passive smoke dose is difficult, and it is of doubtful validity to extrapolate from the uptake of one marker to another." This statement, which correctly describes current understanding, contradicts many of the statements made earlier by the author. In sections above, the author states that estimates of the exposure of the population to ETS can be estimated by extrapolation from biological markers. Both these statements cannot be true. The author states: "under extreme conditions of incloor air pollution, it has been calculated that a nonsmoker would inhale volatile nitrosamines equivalent to 10 nonfilter cigarettes or 35 filter cigarettes." The significance of this statement is unclear and consequently it should either be revised or omitted. The extreme conditions are not described, thus admitting the possibility that they are unrealistic as is the case for most experiments addressing the uptake of ETS components. (ETS levels are typically made unrealistically high to enable'detection of ETS components in body fluids.) In addition, the author's choice of words implies fact, "would," rather than extrapolation and assumption, which would require use of "possibly might." The author's assumption that formation of cotinine from nicotine and clearance from the body does not differ substantially from non:,mokers to smokers need not be an assumption, because research indicates that this is false. The remainder of the paragraph is not supported and therefore should be omitted. Mr. Repace uses an RSP to nicotine ratio of 13:1 to calculate RSP levels which might have accompanied a mean nicotine concentration of 15 gg/m3. Use of ratios for predictive purposes is invalid (Oldaker and coworkers, Excerpta Medica Interantional Congress Series 1989). In addition, a 13:1 ratio is unjustified because the value was computed from results of an experiment done in an environmental chamber where the ETS levels were unrealistically high. Mr. Repace continues to address the ratio approach through mainstream and sidestream smoke. These are not ETS. The quality of sidestream data varies from unknown to poor. The use of ratios derived from these data is unsupported. The paragraph containing these speculations should tie omitted. 16 L

Because the paragraph going from page 89 to paqe 90 relies on so many assumptions, its message is speculation; the paragraph should be omitted. The author notes that these estimates must be interpreted with caution because they may substantially underestimate exposure to other components of ETS. In view of the assumptions made, the estimates should not be -interpreted at all. As for the cautionary note, this is pure speculation and therefore should be omitted-. L :. I The author correctly states that RSP lacks specificity for ETS; however, the extent of this lack of specificity is not given. The author also states that "the prevalence of smokers correlates well with RSP levels in homes and other enclosed areas." Descriptive statistics relating to this correlation should be included. Without such statistics, the reader may misunderstand the author by thinking that correlation implies prediction, which for the data provided by the "Six City Study" is not the case. The author also states that the "Six City Study" demonstrated that ETS exposures in the home ard at work were significant contributors to personal exposure to RSP. This statement should include statistics that quantify the significance of such exposures. Finally, the author states: "In general,'measurements in a large number of locations using measures of smoke generation such as the number of people smoking or the number of cigarettes being smoked have shown a definitive relationship of smoke generation to particulate levels." This statement is vague and therefore misleading. First, the conclusion is not-a generalization, because, as the author as stated several times before, the relationships have been assessed from a limited number of studies, where the author's analysis employed only the summary results from those studies. Thus, while the limited studies to which the author refers included a large number of locations, analysis was performed only on the summary results of the studies. The author chose not to include results of the majority of studies in his analysis. Finally, the author's use of the "phrase definitive relationship" is unclear since it is accompanied by no quantitative information. Although correlation can be interpreted as a "definitive relationship," correlation does not mean that the relationship can be used for prediction. The author also makes the blanket statement that "in U.S. homes, there are few other sources of RSP, and therefore, the relationships of RSP measurements to ETS are quite accurate." This statement is untrue and insupportable. Spengler and coworkers concluded that 50 % of the RSP in the studies of homes could be attributable to ETS. The accuracy of these relationships is, ,from the standpoint of prediction, not supported, again, 17 'L

because while correlation may be "accurate," the strength of correlation is too poor to allow predictions to be made. On page 90, Mr. Repace summarizes efforts in conne,:tion with the model. He states that the model was validated. The paper by Repace and Lowrey shows only that the model could be fit to data from experiments (lone in artificial settings as distinguished from real-world settings. Equilibrium conditions were imposed on these spaces and levels of RSP were excessive. Finally, these "demonstrations" did not address all the input parameters of the model, such as the number of habitual smokers, the smoking rate, the smoke generation rate, etc. To support the validity of the model, the authur cites work by Kuller and coworkers, which is not included with the references. Mr. Repace's discussions relating to probability-weighted exposures to ETS that appear in Table 13 cannot be followed. The reviewer recommends that the two paragraphs be revised to include literature citations supporting the values assumed as well as a development of the arithmetic used to obtain the estimates. In addition, the paragraphs should present the issumptions precisely and describe the extent to which the assumptions would be expected to affect the final results. In summarizing this section, Mr. Repace notes the disadvantages of the model and states that the NRC, Surgeon General, and IARC have utilized this data base ["the RSP-based estimates "] for exposure assessment purposes. Although this reviewer recognizes that the equilibrium model has been described in these documents, he is unaware that any of these organizations has used the model. The author should'cite publications describing use of the model. 18