Tobacco Documents Online

SDecial ReQorti Passive Smoking: How Great A Hazard? By Gary L. Huber, MD, Robert E. Brockie, MD, and Vijay K. Mahajan, MD . Reports from medical journals, the popular media, and federal regulatory agencies about the adverse health effects of passive smoking have convinced many jurisdictions to ban smok- ing in public places. What is often missing from such discussions is the scientific basis for the health-related claims. The following article examines the scientific data concerning the ascertainable risk from inhalation of enoiron- mental tobacco smoke. One of its authors, Dr. Gary Huber, spoke at a recent CR symposium on "Science and Regulation" (see article on page 35).-Ed. © bout 50 million or so Americans are active smokers, consuming well over 500 billion tobacco cigarettes each year. The "secondhand" smoke-usually called "environ- mental tobacco smoke," or more simply "ETS"-that is generated is released into their surroundings, where it potentially is inhaled passively and retained by nonsmokers. Or is it? Literally thousands of ETS-related state- ments now have appeared in the lay press or in the scientific literature. Many of these have been published, and accepted as fact, without adequate critical questioning. Based on the belief that these publications are accurate, numerous public policies, regulations, and laws have been implemented to segregate or restrict active smokers, on the assertion that ETS is a health hazard to those who do not smoke. What quantity of smoke really is released into the environment of the nonsmoker? What is the chemical and physical quality, or nature, of ETS remnants in our environment? Is there a health risk to the nonsmoker? In concentra- Drs. Huber, Brockie, and Mahajan are with, respect- ively, the University of Texas Health Science Center, the Presbyterian Hospital of Dallas, and St. Vincent's Hospital-Medical College of Ohio. 0 10 Consumers' Research tions as low as one part in a billion or even in a trillion parts of clean air, some of the highly- diluted constituents in ETS are irritating to the membranes of the eyes and nose of the non- smoker. Cigarette smoking is offensive to many nonsmokers and some of these highly-diluted constituents can trigger adverse emotional responses, but do these levels of exposure really represent a legitimate health hazard? "Cigarette smoking is offensive to many nonsmokers and some of these highly-diluted con- stituents can trigger adverse emotional responses, but do these levels of exposure really represent a legitimate health hazard?" Clear answers to these questions are difficult to find. The generation, interpretation, and use of scientific and medical information about ETS has been influenced, and probably distort- ed, by a "social movement" to shift the empha- sis on the adverse health effects of smoking in the active smoker to an implied health risk for the nonsmoker. The focus of this movement, initiated by Sir George Godber of the World Health Organization 15 years ago, was and is to emphasize that active cigarette smokers injure those around them, including their families and, especially, any infants that might be exposed involuntarily to ETS. By fostering the perception that secondhand smoke is unhealthy for nonsmokers, active smoking has become an undesirable and an antisocial behavior. The cigarette smoker has become ever more segregated and isolated. This ETS social movement has been successful in

• reducing tobacco cigarette consumption, per- haps more than other measures, including mandatory health warnings, advertising bans on radio and television, and innumerable other efforts instituted by public health and medical professional organizations- But, has the ETS social movement been based on scientific truth and on reproducible data and sound scientific principles? At times, not surprisingly, the ETS social movement and scientific objectivity have been in conflict. To start with, much of the research on ETS has been shoddy and poorly conceived. Editorial boards of scientific journals have selectively accepted or excluded contributions not always on the basis of inherent scientific merit but, in part, because of these social pres- sures and that, in turn, has affected and biased the data that are available for further analyses by professional organizations and governmen- tal agencies.-In addition, "negative" studies, even if valid, usually are not published, espe- cially if they involve tobacco smoke, and thus they do not become part of the whole body of literature ultimately available for analysis. Negative results on ETS and health can be found in the scientific literature, but only with great difficulty in that they are mentioned in passing as a secondary variable in a "positive" study reporting some other finding unrelated to ETS. To evaluate critically any potential adverse health effects of ETS, it must first be appreciat- ed that not all tobacco smoke is the same, and thus the risk for exposure to the different kinds of tobacco smoke must be considered indepen- dently.1 What Is ETS? The three most important forms of tobacco smoke are depicted in Figure 1. Mainstream smoke is the tobacco smoke that is drawn through the butt end of a cigarette during active smoking; this is the tobacco smoke that the active smoker inhales into his or her lungs. The distribution of mainstream smoke is sum- marized in Table 1 (page 12). Sidestream smoke is the tobacco smoke that is released in the sur- rounding environment of the burning cigarette from its smoldering tip between active puffs. Many publications have treated sidestream smoke and ETS as if they were one and the same, but sidestream smoke and ETS are clear- ly not the same thing. Sidestream smoke and ETS have different physical properties and they rA burning cigarette has been described as 'a miniature chemicalfaclory ' producing numerous new components from its raw materials. When a cigarette is smoked, the burning cone has a temperature of about 860 to 900°C during active puffing, and smolders at 500 to 800`C between pulfs. When tobacco burns at these temperatures. the products of pyrolyzation are all vapors. As the vapors cool in passage away from the burmng cane. they cpndense into minute liquid droplets, initially about two ten-millionths af a meter in size. Generally, then, all forms ot smoke are mucroaerosols of very small liquid droplets of particulate matter suspended in their surrounding vapars or gases. Thus, all smoke has a-particulate phase" and a"gas phase." Figure 1: Particulate Phase and Gas Phase of Tobacco Smoke* 0 I 0 o e o o e e e e e o 0 0 o a o 0 e oo e e eae a e o 0 0 ooooeaooeeeooosoaeoae aaeeeeeooeooeceeeoo eooaoooeooeoeeooooe oeoo oeoo 0 0 0 0 0 0 eo eaoeeooooooooeoooaaoe 000000 00 o e e o eoeeooooooooeoeeooaoe eeeoooo eee e o 0 00oooooeeooeeoeoooeae ooaueooeeeeaooooaaeoe oeooooo00 oe e 0 Mainstream Smoke Sidestream Smoke Environmental Tobacco Smoke (ETS) - Schematic representation of the particulate phase and the gas phase of tobacco smoke. Environmental tobacco smoke Is nut smoke in the conventional sense, but rather a very limded number ol pi9hly-difuted remnants or residual constltusnfs at mainstream smoke and skfestrwn smoke. July 1991 11

Table 1: Distribution of Mainstream Smoke Total Mainstream Smoke ' Wet Total Particulate Manor Nicotine :~ Water 500* 22 1.3 3.7 "Tar" . 17 Aerosol Gas Phase Water 478 Air Components 50 Carbon Monoxide 350 Carbon Dioxide 50 Other Components 8 •AII data expressed in milligrams fura 500 ms deliver clgarette, as dear- mined by redenl Trade Commisslan criteria. SCtIFlCE: Adapted trum Nubor,19E9. have different chemical properties. Environ- mental tobacco smoke is usually defined as a combination of highly diluted sidestream smoke plus a smaller amount of that residual main- stream smoke that is exhaled and not retained by the active smoker. What really is ETS? In comparisonto mainstream smoke and side- stream smoke, ETS is so highly diluted that it is not even appropriate to call it smoke, in the conventional sense. Indeed, the term "environ- mental tobacco smoke" is a misnomer. Why is ETS a misnomer? Several reports on smoking and health from the Surgeon General's Office, a National Research Council review of ETS in 1986, the more recent Environmental Protection Agency's risk assess- ment of ETS, and several review articles all have provided a long list of chemical con- stituents derived from analyses of mainstream smoke and sidestream smoke, with the implica- tion that because they are demonstrable in mainstream smoke and sidestream smoke these same constituents must, by inference, also be present in ETS. No one really knows if they are present or not. In fact, most are not so present or, if they are, they are present only in very dilute concentrations that are well below the level of detection by conventional technologies available today. Only 14 of the 50 biologically active "proba- ble constituents" of ETS listed by the Surgeon General, for instance, actually have been mea- sured or demonstrated at any level in ETS. The others are there essentially by inference, not by actual detection or measurement. Thus, there are 36 constituents in these lists that are in- ferred to be present in ETS, but their presence has not been confirmed by actual detection or measurement. In this sense, then, ETS is really not smoke in the conventional sense of its defi- nition, but rather consists of only a limited number of "remnants" or residual constituents present in highly dilute concentrations. Because the levels of ETS cannot be quanti- fied accurately as such in the environment, some investigators have attempted to measure one or more constituent parts of ETS as a "sub- stitute marker" for ETS as a whole. The most frequently employed such "marker" has been nicotine or its first metabolically stable break- down product, cotinine. Nicotine was consid- ered an "ideal marker" because it is more or less unique to tobacco, although small amounts can be found in some tomatoes and in other food sources. In the mainstream tobacco smoke that is inhaled by the active smoker, nicotine starts out almost exclusively in the tiny liquid droplets of the particulate phase of the smoke. Because the smoke particles of ETS become so quickly and so highly diluted, however, nicotine very rapidly vaporizes from the liquid suspend- ed particulates and enters the surrounding gas. In technical terms, the process by which nico- tine leaves the suspended aerosol particle to enter the surrounding gas phase is called "denudation." As a vapor or gas, nicotine reacts with or adsorbs onto almost everything in the environ- ment with which it comes into contact. Thus, nicotine is not a representative or even a good surrogate marker for the particulate phase, or even the gas-vapor phase, of ETS. In fact, there are no reliable or established markers for ETS. The remnant or residual constituents of ETS each have their own ahemical and physical behavior characteristics in the environment and none is present in a concentration in our environment that reaches an established threshold for toxicity.z Measuring Health Risks Because the level of exposure to ETS or the dose of ETS retained cannot be quantified under every-day, real-life conditions, the health effects following exposure to residual con- 2A Ihreshold fimil value (usually expressed as milligrams of a substance per cubic meter of air ar as parts ol a substance present per million oarts of res- pirable clean air) is the recommended concentration of a substance as the maximal level that should not be exceeded to prevent occupational disease through exposure in the workplace. Threshold limrt values have not been established tnr our general, every-day environment outside ol industrial expo. sure. Threshold limit values are determined by toxicologists. epidemiologists, and hygienists through their interpretation of literature, and usually are sanc tioned by the American Conference of Governmental Industrial Hygienists No constituent of ETS has been measured in our every-day environment at levels that exceed the threshold limifvalues permitted in the norkplace. 12 Consumers' Research

stituents of ETS have been impossible to evalu- ate directly. In broad terma, two different approaches have been employed in an attempt to assess indirectly the health risks for expo- sure of the nonsmoker to the environmental remnants of ETS. The first of these involves a theoretical concept that is called "linear risk extrapolation." Linear risk extrapolation has been employed extensively in attempts to deter- mine the risk for lung cancer in nonsmokers exposed to ETS? This concept of linear risk assumes that if there is a definable health risk for the active smoker, then there also must be a projected lower health risk for the nonsmoker exposed to ETS. This is represented schematically in Figure 2. The risk has been presumed to be lin- ear from the active smoker to the nonsmoker exposed to ETS, based proportionately on the relative exposure levels and retained doses of smoke; it thus requires some measurement of tobacco smoke exposure for both groups. This is fairly easy to achieve in the active smoker, in part because mainstream smoke has been so well-characterized and it is delivered directly from the butt-end of the cigarette into the smoker. Such is obviously not the case, howev- er for the nonsmoker exposed to ETS. Most projections of linear risk for ETS-expo- sure have been based on the use of nicotine as a representative marker of exposure. A few pro- jections have been based an carbon monoxide levels or amounts of respirable suspended par- ticulates in the environment, but these approaches are fraught with even greater error. Since nicotine initially is in the particulate phase of the mainstream smoke inhaled by the active smoker and it is present primarily as a highly diluted gas-phase remnant or residual vapor-phase con- stituent in the nonsmoker's environment, the concept of a linear health risk from the active smoker to the nonsmok- er is based on rather shaky scientific-reasoning. That is to say, it is not valid to estimate a health risk for exposure to the particulate phase in the active smoker and then compare it with the health risk for exposures to the gas phase in the ETS- exposed nonsmoker. Simply stated, "like" is not being com- pared to "like." Mainstream smoke and the residual constituents of ETS represent very dif- ferent exposure conditions. Whether present in mainstream smoke or in ETS, particulate phase and gas phase constituents have very different biological properties, as well as different physi- cal and chemical characteristics, and any asso- ciated health risks are also very different. The concept of linear risk extrapolation for ETS is based on a theory that when applied to ETS incorporates unsound assumptions that are not valid. There is no way, as yet, to evaluate or compare the levels of exposure in active smok- ers and nonsmokers exposed to ETS. The second approach used to evaluate health risks for nonsmokers exposed to ETS has employed epidemiologic studies. Epidemiology is a branch of medical science that studies the distribution of disease in human populations and the factors determining that distribution, chiefly by the use of statistics. The chief func- 3The concept is based on a theoretical extrapolation of the risk for lung cancer in the active smoker to the risk for lung cancer in the passive smoker on the basis ot a 'representative marker" for both smoke exposures. This'9inear risk extrapolation' trom one to the other is a model that is hasen on mathematical theory and on several assumptions. The theory assumes that the risk applies to all exposure levels, even if they are very low. Some advocates of the model even assume a"one molecule, one hi1" mechanism, where exposures so low that they cannot be detected or measured can still cause disease if only a sin- gle molecule reaches a vulnerable body tissue. The linear risk theory also assumes that the risk for accumulative exposure remains constant and, thus, that the exposed individual has no capacity to adapt or develop tolerance mechanisms for the exposure. Since active smokers readily and rapidly devel- op tolerance through a variety of defense mechanisms. it seems illogical to assume those repeatedly exposed to ETS would not do the same, The linear risk model assumes that the risk tor exposure to ETS is independent of any confounding factors. Finally, for this theory to be valid, it must be assumed that the risk is linear for duration of exposure and that it is linear for concen- lratlon of exposure. None of these assumptions holds true on scientific testing for comparative projections ot mainstream smoke to ETS. Figure 2: Linear Risk Extrapolation* 5.0 _ Z 0 40 ~ No Threshold ~ One Molecule Theory s3.0 d w a 2.0 m 0.0 0 2"0 4"0 8.0 8,0 10 Relative Environmental Exposure Level "The concept of linear risk extrapolation. In this theory, Me heatth response (expressed as a rela- tive risk) is dlrectly or lineady related to the relative environmental exposure level. This theory sug- gests that there Is no 'safe" threshold below which there is no response, and that exposure to as little as one molecule of the envlronmenul substance can uuse an adverse response. July 1991 13

0 "Of the 30 ETS-lung cancer stud- ies, 6 reported a statistically significant association... and 24 of those studies reported no statistically significant effect." tion of epidemiology is the identification of pop- ulations at high risk for a given disease, so that the cause may be identified and preventative measures implemented. Epidemiologic studies are most effective when they can assess a well-defined risk. Because ETS-exposure levels cannot be mea- sured or in any other way quantified directly, even by representative markers, epidemiolo- gists have had to use indirect estimates, or sur- rogates, of ETS exposure. For nonsmoking adults, the number of active smokers that are present in the household has been used as a surrogate for ETS exposure. Usually the active smoking household member has been the non- smoker's spouse. With a few limited exceptions, disease rates in nonsmokers exposed to a spouse who smokes have been the basis for all epidemiologic assessments. Almost all of these studies have evaluated nonsmoking females married to a husband who smokes. For children, the surrogate for ETS exposure has been the number of parents in the household who smoke. Estimates of ETS expo- sure based on spousal or parental surrogates have been derived by various questionnaires; no study employs any direct quantification of ETS or of ETS remnant constituents in the actual environment of the nonsmoker. Questionnaires of smoking habits are notori- ously limited and often inaccurate, in part because of the "social taboo" that smoking has become and, in part, for other reasons related to the ETS social movement. Nevertheless, data from questionnaires about smoking behavior in spouses or in parents are the only estimates of ETS exposure available. Rates for three dis- eases in nonsmokers exposed (via surrogates) to ETS have been assessed: lung cancer, coro- nary heart disease, and respiratory illness in infants and small children. Only lung cancer will be discussed in this article. ETS and Lung Cancer What is the state of evidence on ETS and lung cancer? Almost all of the epidemiologic studies that are available to answer that ques- 14 Consumers' Research tion are based on the concept of some measure- ment of relative risk. None of the studies actu- ally has measured exposure to ETS or to any of its residual constituents directly. Relative risk is a relationship of the rate of the development of a disease (such as lung cancer) within a group of individuals exposed to some variable in the population studied (such as ETS) divided by the rate of the same disease in those not exposed to this variable. Relative risk is most frequently expressed as a"risk ratio," which is a calculated comparison of the rate of the disease studied in the exposed population divided by the rate of that disease in some control population not exposed to the variable studied. The terms "risk ratio" and "relative risk" are often used synonymously. Thus, the relative risk in all epidemiologic E'1'S studies on lung cancer is expressed as the rate of lung cancer in the ETS-exposed group (indi- viduals married to a household smoker) divided by the rate of lung cancer where there was no ETS exposure (no household smokers). If the disease rates were exactly the same in these two groups, the risk ratio would be 1.0. There have been 30 epidemiologic studies on spousal smoking and lung cancer published in the scientific literature. Twenty-seven of these epidemiological studies were case control stud- ies, where the effect of exposure to spousal smoking was evaluated retrospectively on data that had already been available for review. The "cases" in these case-control studies were non- smoking individuals with lung cancer married to smokers. The rate of lung cancer in these "cases" was compared, by the derived risk ratio, to the rate of lung cancer in "control" or nonsmoking individuals who were married to nonsmokers. Three of the studies followed cohort popula- tions of individuals exposed to spousal smoking prospectively over the course of time. A "cohort" is any designated group of people. A "cohort study" identifies a group of people that will be exposed to a risk and a group that will not be exposed to that risk, and then follows these groups over time to compare the rate of disease development as a function of exposure or no exposure. The first studies were published in 1982 and the last studies were published in 1990. The studies originate broadly from different parts of the world and, for the most part, involve evalu- ations of lung cancer in nonsmoking females married to a smoking male partner; eight of the studies have limited data on nonsmoking males married to smoking females. Some of the stud- 2074144182

ies are quite small, listing fewer than 20 sub- jects; others are based on larger populations, with four studies reporting between 129 and 189 cancer cases. Of the 30 studies, six reported a statistically significant association (identified by a positive relative risk ratio in the spousally- exposed to the non-exposed population) and 24 of the studies reported no statistically signifi- cant effect. The average esti- mated relative risk ratio for each study and each sex is list- ed in Table 2, as are the confi- dence intervals reported by the authors or, where not reported, calculated by others in pub- lished review articles * Some of the negative studies- that is, some of the 24 studies that did not show a statistically significant association between the development of lung cancer and exposure to spousal smok- ing-contained data that sug- gested to the authors or to other reviewers a "positive trend." In most of science, "trends" do not count; data stand as either sta- tistically significant or not sta- tistically significant, with sig- nificance determined by specif- ic accepted rules of biostatis- tics. New rules should not be "made to fit" an otherwise unproved hypotheses, just because the subject is tobacco and the observed results do not support the hypothesis investi- gated. ETS Risk Weak A relative risk is called strong or it is called weak, depending on the degree of association, or the magnitude of the risk ratio. A strong relative risk would be reflected by a risk ratio o£ 5 to 20 or greater. Weak relative risks, by conventional defini- tion, have risk ratios in the range of 1 to 3 or so. Within 4A confidence interval is a range of values that has a specdiad probability of including the true value (as opposed to the estimated average value) within that ranqe. In the data presented in Table 2, the confidence intervals are set such that there is a 95% probability that the true value will tall within the range ot values listed. the 30 epidemiologic studies on ETS and lung cancer, there are 37 different total reported sets of risk ratios for male or female nonsmok- ers. None of the studies reports a strong rela- tive risk. Nine of the studies report risk ratios of less than 1.0. Thus, the results from all epidemio- (See SMOKE, page 33.) Table 2: Studies of ETS and Lung Cancer in Nonsmokers 95% Study Sex Number af Cases Relative Risk• Confidence Interval Case Control Studies Chan and Fung,1982 F 34 0.75 (0.43. 1.30) Trichopoulos at a1.,1983 F 38 2.13•' (1.18, 3.83) Correa et a1.,1983 F 14 2.07 (0.81, 5.26) M 2 1.97 (0.38, 10.29) Kabat and Wynder,1984 F 13 0.79 (0.25, 2.45) M 5 1.00 (0.20, 5.07) Bufiler et al., 1984 F 33 0.80 (0.34, 1.81) M 5 0.51 (0.15, 1.74) Garfinkel eta1.,1985 F 92 1.12 (0.94, 1.60) Wu et al., 1985 F 29 1.20 (0.50, 3.30) Akiba et at.,1986 F 73 1.52 (1.00, 2.5) M 3 2.10 (0.5, 5.6) Lee et a1.,1986 F 22 1.03 (0.37, 2.71) M 8 1.31 (0.38, 4.59) Brownson et a1.,1987 F 19 1.68 (0.39, 2.97) Gao et a1.,1987 F 189 1.19 (0.6, 1.4) Humble et al., 1987 F 14 1.78 (0.6, 5.4) Koo et al., 1987 F 51 1.55 (0.87, 3.09) Lam et a1.,1987 F 115 1.65" (1.16, 2.35) Pershagen et al.,1987 F 33 1.20 (0.70, 2.10) Geng et a).,1988 F 34 2.16" (1.03, 4.53) Inoue and Hirayama,1988 F 18 2.55 (0.91, 7.10) Katada et a1.,1988 F 17 - (NS;p=0.23) Lam and Cheng,1988 F 37 2.01•• (1.12, 1.83) Shimizu at al., 1988 F 90 1.10 N/A He,1990 F 45 0.74 (0.32, 1.68) Janerich at a1.,1990 F 129 0.93 (0.55, 1.57) Kabat, 1990 M 13 1.20 (0.54, 2.68) F 35 0.90 (0.46, 1.76) Kalandidi et a).,1990 F 91 2.11 (1.09, 4.08) Sobue et a1.,1990 F 64 0,94 (0.62, 1.40) Svensson, 1990 F 17 1.20 (0.40, 2.90) Wu-Williams et a1.,1990 F 205 0.7 (0.6, 0.9) Cohort Studies Garfinke1,1981 F 88 1.17 (0.85, 1.89) (0.77, 1.61) Gillis at al., 1984 F 6 1.00 (0.59, 17.85) M 4 3.25 Hirayama, 1984h F 163 1.45 (1.04 2.02) 1984a 7 2.28" (1.19 4.22) 'Weak relative nsks have tlsk ratios at between 1 and 3, or sa. My risk raeo below i represents a neea- tive rtWnonship. Note that none of the studfes show a atrnnp relative risk " StatisGczlly slqnifkant at the 5% Ievei. July 1991 15

---