Tobacco Documents Online

Rirk Analysir, I!ol. 13. No. 4, 1993 An Enforceable Indoor' Air Quality Standard for Environmental Tobacco Smoke in the Workplace' James L. Repace2'• and Alfred H. Lowrey' Reedvrd Novrmber 4, 199z* irvistd March 11, 1993 :A::.i;J F.nvitonmental tobacco smoke (ETS) has recently been determined by U.S. environmental and occupational health authorities to be a human carcinogen. We develop a model which permits using atmospheric nicotine measurements to estimate nonsmokers' ETS lung cancer risks in m- dividual workplaces for the first time. We euimate-that during the 1980s, the U.S. nonsmoking adult population's median nicotine lung exposure (homes and workplaces combined) was 143 micrograms (µg) of nicotine daily, and that most-exposed adult nonsmokers inhaled 1430 µg/day. These exposure estimates are validated by pharmacoldnetic modeling which yields the cotrespond- ing steady-state dose of the nicotine metabolite, cotinine. For U.S. adult nonsmokers of working age, we estimate median coti}tine values of about 1.0 nanogram per milliliter (ng/ml) in plasma, and 6.2 ngbml in urine; for most-exposed nonsmokers, we estimate cotinine concentrations of about 10 ng/ml in plasma and 62nghal in urine. These values arc eonsistent to within 1596 of the eotinine values observed in contemporaneous clinical epidemiological studies. Corresponding median risk from ETS exposure in U.S. nonsmokers during the 1980s is estimated at about two lung cancer deaths (LCDs) per 1000 at risk, and for most-exposed nonsmokers, about two LCDs per 100. Risks abroad appear s_im7ar. Modeling of the lung cancer mortality risk from passive smoking suggests that de nu,umis [i.e., "acceptable'! (10-')], risk occurs at an 8-hr time-weighted-average exposure concentration of 7.5 nanograms of ET'S nicotine per cubic meter of workplace air for a working lifetime of 40 years. This model is based upon a linear acposurt-response relationship validated by physical, clinical, and epidemiological data. From available data, it appears that workplaces without effective smoking policies considerably cuxed'this de mininus risk standard. For a substantial fraction of the 59 million nonsmoking workers in the U.S., current workplace exposure to ETS also appears to pose risks exceeding the de marrifestia risk level above which carcinogens are strictly regulated by the federal government. KEY WORflS: Fnviromaental tobacco smoke; indoor air quality standard; nicotine; cotinine; phumacokinetie modeling; risk assessment. • 1. IIVTRODUCTION In 1991, the National Institute for Occupational Safety and.Health (NIOSH). deciared.environmental to- 'Disclaimer. This work was perfonaed by the authors in their private capacity. No official support or endoisement by the Environmental Protection Agency, the Naval Researeh Labotatory, or any other, fed- eral agency Is intended or should be inferred. 2U.S. Environmeaqt Protection Agency, Washington, D.C. 20460. sNavsi Research Labontory, Washington. D.C. 20375. lb whom ali correspondence should be addressed. 463 bacco smoke (E?S) to be a "potential occupational car- cinogen," legal terminology for a substance capable of causing human cance.r or reducing its latency period c't Based.. upon bioiogical. plausibility and epidemiologicai studies,tl-~ a number of risk assessments have estimated the lung cancer mortality caused by passive smoking among U! S. nonsmokers to be of the order of 5000 deaths N per year.(6) In 1992, thie American Heart Association declared ETS to be a "major preventable cause of car- diovascular 0 diovascular disease and death,"rn and estimated ETS- N related mortality, from heart disease and cancer com _ oz72a33n47Aaoo4asJSOm.oaa 0,1993 Soc;ay ror NiskAnays:a M 0

464 bined, to approach 53,000 annualIy, placing passive smoking as the third leading preventable cause of death{ after active smoking and alcohol.(8) Workplace exposure of nonsmokers to environmen- tal tobacco smoke is widespread, with 28.5 million non- smoking workers in 1988 (36.5% of all nonsmoking workers) being employed in workplaces with few or no restrictions on smoking.0) In 1992, 25% of the adult population were current cigarette smokers. However, among the 100 largest U.S. industrial corporations, only 35% had banned smoking in the workplace, and of these, many restricted the ban solely to corporate headquar- tezs.10t NIOSH recommends that employers ban smoking in the workplace if possible, or "minimize" nonsmok- ing workers' exposures if not. Because of the NIOSH advisory, and a major report issued in 1993 by the U.S. Environmental Protection Agency declaring ETS to be a "known human carcinogen",(2) workplace smoking policies to restrict nonsmokers' expostme to ETS are being increasingly mandated by legislaturWl') considered by regulatory agencies,03j or voluntarily adopted by busi- ness.t'at However, workplace smoking policies shoit of bans may reduce, but do not eliminate, nonsmokers' exposures.0') Accordingly, there is a need for an en- forceable indoor air quality standard for ETS, so that regulators may quantify the risks in any indoor airshed. With such a standard, any residual ETS exposure after imposition of controls can be evaluated as "acceptable" ot1 "unacceptable," relative to established regulatory ai- teria for risk from exposure to environmental carcino-. gens. In 1985, Repace and Low*") proposed a health- based indoor air quality standard based upon respirable suspended particulate (RSP) air pollution from ETS; however, although strongly associated with ETS, RSP is not unique to ETS,W) and hence an ETS-RSP standard would be difficult to enforce. Quantification of ETS ex- posure and risk for regulatory purposes must be predi- cated upon substances uniquely associated with tobacco combustion, such as nicotine in workplace air.JA Newly available data now permit development of a method to use ETS nicotine levels as a quantitative surrogate for the carcinogenic risk of ETS in.individuat workplaces, and for deriving a legally enforceable health-based in- door air quality standard for.ETS, usable by federal and . state regulatory agencies. 2. METHODOLOGY In creating an enforceable indoor air quality stan- dard for the workplace carcinogen ETS, the fundamental Repace and L6wrey ' problem to be solved' is to develop a model correlating levels of substances uniquely associated with ETS ex- posure (nicotine and'its metabolite, cotinine), and highiyt- correlated with number of cigarettes smoked,tls•t't ana!:, hence with the lung cancer risk caused by ETS. To ac- compiish this, we modify our previously developed model relating the lung cancer risk from ETS exposure to the nonsmoking population's exposure - to ETS-associated' respirable suspended particulate (RSP).(6•1`-u•'a-"- 25) By relating atntcupheric nicotine in buildings to RSP from ETS,tI6-2`t a nexus may be established between lung can- cer risk and the ETS constituent, nicotine. In this man- ner, nicotine and cotinine, which are the best available markers for ETS exposure and dose,(2-3-5) therefore also serve as the most suitable markers for the carcinogenic effect of ETS, despite their own apparent lack of car- cinogenic activity." Nicotine exposure model predic- tions are validated by pharmacokinetic comparison with clinical epidemiologic studies of cotinine in the body fluids of nonsmokers. This also permits calculation of the risk to the typical and most-exposed individuals in the population. Current exposure levels for various classes of workers are assessed from recent personal and area monitoring studies of nicotine in workplace air. 3. MODELING NONSMOKERS' EXPOSURE' TO NICOTINE 1. Estimates of population ETS nicotine exposure may be simply derived from earlier ETS RSP population ex posure estimates by Repace and Lowrey,t18t using new information- desenbing empirical relationships between ETS RSP and nicotine concentrations. 3.L Exposure of the Nonsmoldng Population to RSP ftvni ETS Repace and Lowreytuas1°ast in assessing the risk of passive smoking, developed a model to estimate the exposure of the U.S. nonsmoking population to RSP from ETS. This model assumed that nonsmokers' ETS exposures (and hence lung cancer risk) are dominated by two microenvironments, homes and workplaces, in which time-budget studies have shown the U.S. popu- lation averages 88% of its time.10aA-M This assumption has been subsequently verified by ETS epidemiology in nonsmokersa"-'0t Repace and Lowreytl"t estimated typ- ical population exposures to RSP from ETS for U.S. nonsmoking adults of working age by using indoor air concentration models('9) together with time-budget stud-

Indoor Air Standard for Tobacco Smoke ies, tables of respiration rates, standards for naturaL and mechanical ventilation, surveys of smoking prevalence in~ the home and workplace, and epidemiological studies of passive smoking. Repace and l,owrey(38) estimated the typical daily inhaled lung exposure of the U.S. non- smoking population to RSP from ETS during the 1980s to be Q,,, = 1.43 milligrams per day, and the lung ex- posure of most-exposed nonsmokers was estimated to be 10-fold'~ the typical, at Q. = 14.3 milligrams per day. 3.2. Relating ETS RSP to Nicotine Using Field Studies There is a constant relationship between the total amount of RSP and nicotine emitted in mainstream and sidestream smoke.31t Since ETS RSP and nicotine differ in decay rate, instantaneous ratios of RSP and nicotine concentrations during growth and decay of ETS will be variable in timeaM However, in time-averaged mea- surements, a constant ratio of RSP to nicotine in ETS reemerges. This is demonstrated by regression relation- ships derived by Leaderer and Hammond,t16) and by Miesner et aL (z•) from field measurements of RSP and nicotine in homes and workplaces, respectively. Lead- erer and Hammond(16) have studied the relationship be- tween nicotine and RSP from smoking in 47 homes where cigarette consumption was reported, in the NYSERDA Study of indoor air pollution in New York State. The weekly average RSP concentration, R, (in units of mi- crograms per cubic meter) measured in these homes was regressed against the weekly average nicotine concen- tration, N. (3z) The resulting regression relationship is given by: R = 9.8N + 22.9 (u.g/m') (1) where the coefficient of determination, r2=0.64. By comparison, Leaderer.and Hammond(16) found the RSP- to-nicotine ratio at high smoking densities in an exper- imental-chamber with 4 brands of commercial cigarettes smoked by smokers to be only 30% higher,14.1 ± 1.9, than in the NYSERDA field study. M'iesner et al tu) studied workday average RSP and nicotine concentra- tions in 21 commercial and institutional buildings chosen to represent a variety of places where people. spend time indoors other than at home. ~kesner et aL2't reported an R-to-N ratio of 9.1 to 1 for these buildings (r2 = 0.78) in agreement with the results of Leaderer and Hammond.(i6) Similarly, Nagda et aL (33) obtained an R- to-N ratio of 10.3 to 1, from in fiight measurements in the smoking sections of 61 U.S. aircraft flights. We adopt an ETS RSP-to-nicotine ratio of 10:1. 465 3.3. Assessing Nicotine Exposure of the Nonsmoking Population The field-measured ETS RSP-to-nicotine ratio of ten-to-one is now used to obtain an estimate of the non- smoking population's nicotine exposure, E, derivative from our ETS RSP population exposure model.'41 is simply obtained by dividing the ETS RSP values Q,,, and Q. by 10, yielding the values E,,, = 143 µg/day, for the ETS nicotine exposure of the population-average nonsmokers, and E., = 1430 µg/day, for the most- exposed nonsmokers. In general, the average workplace nicotine air concentration, N (in units of wg/m') is re- lated to an individual's exposure, E (in units of µg(day), by the expression: E = pHN (peday) (2) where p is the nonsmolker's respiration rate (in units of m'!hr) during exposure, and H is the duration of that exposure, (in units of hr/day). Respiration rates range from about 0.4 m3/hr for resting adults, to 1 m'Ihr for alternate sitting and light work, to 2 m'/hr for heavy work.ta`3 4. RELATING NICOTINE EXPOSURE TO COTMIIHE DOSE Cotinine levels for the median and most-exposed individuals in the population can be assessed from clin- ical epidemiological studies. Thus, if estimated ETS ni- cotine exposure is related to cotinine dose using pharmacoldnetic modeling, the accuracy of these expo- sure estimates can be assessed. Accordingly, we define the following parameters (and their units) for'use in a classicaIm single-compartment pharmacokinetic model translatinj estimated nicotine axpcuttre into oortespoad- ing estimated cotinine levels in body fluids: E, daily nicotine lung exposure (µg,/day); a, efficiency of ab- sotption of nicotine by the lung (diineusionIess); D, daily absorbed nicotine dose (µg/day); F, efficiency of con- version of nicotine to cotinine by the body (dimension- less); P, steady state. plasma cotinine.concentration (ng( ml); CL7, rate of cotinine clearance from plasma by he- patic, renal, and other mechanisms (ml/min); CLR, rate of cotiniae removal from .p)asma by renal clearance only (mllmin); A,,, the mass of cotinine excreted into urine daily (ng); V, daily tuine volume (ml); II, steady-state urinary cotinine concentration (ngfml); td, number of minutes in a day. ~ 0 ~ r.: ~ ~ ~ ~ ~

466 4.1. Model Development Next, we derive a pharmacokinetic model relating daily lung nicotine exposure, E to daily average ab- sorbed nicotine dose, D, and to steady-state levels of plasma cotinine concentration, P, and urinary cotinine concentration, U. We predict the population-average and upper extreme values for P and U for adult nonsmokers from passive smoking corresponding to our estimates of nicotine exposure. To assess the usefulness of this model, we compare model predictions, valid for conditions in the mid-1980s, to the results of several large ciinical epidemiological studies of the levels of eotinine in the body fluids of nonsmokers performed in the late 1980s. That fraction of nonsmokers' daily nicotine lung exposure E absorbed by the lung is transformed into a plasma nicotine dose D. Plasma nicotine dose distributes in the body rapidly, with a distribution half-life of 5-10 min, and an apparent volume of distribution of 180 LM Nicotine is metabolized to codnine by the liver with an efficienry F= 86%, and the remainder is partly metab- olized to nicotine-N-oxide, or excreted unchanged in sa- Iiva and utine.(") Elimination follows first-0rder kinetics, with a 2-hr elimination half-life.m-') Cotinine distributes in smokers with a half-life of about 1.6 hr, into an apparent volume of distribution of about 88 L.t39y Cotinine elimination also follows first- order kinetics, and has an elimination half-life of the order of a day in nonsmokers, and therefore yields an index of average exposure to ETS.t'001> Measurements of cotinine in plasma, urine, or saliva are sufficiently sensitive and specific to identify passive smok- ers.t`-37-39•"-ut Passive smokers studied in Western Eu- rope and North America are found to have cotinine levels between about 0.1% and 1% of those in active smok- ers.(a2) Recent findings suggest that nicotine and cotinine kinetic parameters appear to be extrapolatable from smokers to nottsmokers.('s) Daily absorbed nicotine dose, D, is linearlyt") re- lated to the daily average exposure; E. , D a ctE (µ8/day) (3) where a is the nicotine absorption efficiency. The a value for U.S. nonsmokers from E°I'S due to U.S. cig- arettes has not been measured. However, Iwasi et a[. (") have measured nicotine absorption from ETS generated by Japanese Mild Seven cigarettes in 17 nonsmoking Japanese women. They found a nicotine absorption ef ficiency, a= 71.3% ± 10.2%. Measured nicotine ab- sorption efficiencies ranged from a low of 45% to a high of 95%, and were not a function of atmospheric nicotine concentration, which ranged from 40 to 200 µg/m3. We Repace and Lowrey assume the average value for the absorption efficiency for U! S. nonsmokers from U.S. ETS is the same, a=0.71. In equiltbrium, the plasma cotinine concentration P( ^ (typically expressed in units of ng/ml) will be related to the absorbed nicotine dose D by the expression <'-`-39•'s) P = FD/ClA - (4) where D/tQ is the dose rate of nicotine from passive smoking which is assumed to be approximately constant, and CIT is the total clearance of cotinine from plasma (in units of mllmin) due principally to hepatic and renal mechanisms. Aa, the total amount of urinary cotinine excnted, is given by('5): A„ = C11eFDlClr (5) where ClR is the renal cotinine clearance (in units of mI/ min). Dividing both sides of Eq, (5) by V,,, the total amount of urine excreted, where A, JV„ = U, the urine cotinine concentration, using Eq. (4), and solving for C1R, we arrive at an expression for the renal cotinine clearance: CIR = UV„lPts (6A) The total plasma cotinine clearance from Eq. (4) is then: CIr = FD/t,P (6B) Finally, we express U (in units of ng/mI) as a function ~ of E by substituting Eq. (4) into Eq. (6A): U = FCIRaBICh.Y, (7) 4.2. Model Validation Nonsmokers' renal and piasma cotinine clearance (oral and IV dosing) have been measnred by De Schep- per et aL Om at C1R = 7.7 mt/min and Clr = 61 mllmin and by Curvall et aL t'0 at Cl,t a 5.48 ml/min and CIT a 74.4 mUmia. Renal and plasma cotinine clearance may also be estimated (based upon a 70 kg person) from Jatvis u al t'0: Corresponding to a 5-day daily dose of 28 mg of oral nicotine in 3 nonsmokers and 2 occii.sional smokers, the mean steady-state values P_=294 nglml and U=1394 nghnl were measured. Assuming a daily (te = 1440 min) urine elimination V, = 1300 ml,.(`?~_we cal-. culate the clearance values from Eqs. 6A and 6B: CIR = 4.3 ml/min, and Clr = 57 m1/min. Averaging the three sets of data yields for nonsmokers: C1)t = 5.9 ml/ min and Clr = 64 mUmin. (By comparison, for_smokers (n = 28), Cla = 12.3 ± 4.8 nilLmin, and ClT = 72.2 ± 13.1 ml/min).(m By contrast, if the renal-ctearance . 2023668693 )

Indoor Air Standard for Tobacco Smoke for nonsmokers is calculated from the median values U=5.6 ng/mll and P=1.1 ng/ml (see Table II) where nonsmokers' cotinine levels reflect inhalation rather than intestinal absorption, then~using Eq. (6A), C1R=(5.6 x 1300) / (1.1 x 1440) = 4.5 m!/min. Then, correspond- ing to our estimated nonsmokers' lung exposure of E,,, = 143 µglday, we calculate from Eq. (7), U,,, = 0.86 x 5.9 x 0.71 x 143,000 / 64 x 1300 = 6.2 ng/ml, and for E,,. = 1430 µg/day, U,,,,x = 62 ng/ml. Sim- ilarly, we calculate from Eq. (4) P„, =. ForE/tdCi,. = 0.86 x 0.71 x 143,000 / 1440 x 64 = 0.95 ng/ml, and P. = 9.5 ng/ml. These model predictions are summarized in Table I, and compared to typical levels of cotinine measured in nonsmokers' body fluids re- ported in several clinical epidemiological studies as shown in Table II. Our model predictions for cotinine in plasma and urine for the typical and most-exposed individuals are consistent to within 10-15~'o with available data for median and peak levels of cotinine measured in the body fluids of nonsmokers. By extension, our earlier estimates of population exposure to RSP from ETS in the mid 1980s('") also appear to be consistent with the late 1980s ETS population exposure as assessed by cotinine dosi- metry. Table L Model Predictions vs. Observations for Nicotine Lung E;tponue and •Catinine in Body Fluids for U.S. Noasmokers- Niatroae Utinary Pluma Population exposure, E ootinine, U ooaeine, P individual (µB/daY) (nglml) (ng/aii) exposure ' Model predictions 143' 6.2° 0.95' 1430 62 9.5 Observational results No data 5air 1.1• No data 55-90 10-IS Typical Most-ezposed• Median Most-Uposed" ''I3e model (valid bor acposurcs determined by smoking prevalence and workplace smoking policies in the 1980s) givea by F.qs. (4) and (7), predicts plasma and urinary cotinine ~rtations consistent with the results of lue 1980s clinical epidemiological studies, vati. dating both the nicotine espostm modet, derived bae and the earlier RSP e9osue model of Repace and Lowrey finm.vhic6 it is derived. ' Glcadated from estimatcd population average RSP lung exposure Q_ - 1.43 mg/dryu"t osing Eq. (1). • Calcailated from E, using Eq. (7). 11% diPfereace with observed value. ~ Cakulated from E using Eq. (4). 14% diffetvies with observed value. • It is estimated that most-exposed individuals have 10-fold the ez- pasure of the avenge nonsmoker.a1" Modeled values are consistent with observed ranges. f weighted median value from Table tI. • Median value from tiaddow,Oit as reported in Table II. " Majority of estimated matdmum cut-points diSaentiating most-ex- posed passive smokers from t•igfit smokers, from Table U. 467 Nicotine from dietary sources and from outgassing from ETS-contaminated surfaces in buildings are poten- tial contributing factors for cotinine in body fluids. How- ever, their contribution to cotinine levels in body fluids appears to be negligible by comparison to the contribu- tion of ETS.(`8) 5. MODELING RISK FROM NICOTINE AND COTINIIVE We have validated our exposure model by compar- ing observed median and peak levels of cotinine in body fluids for nonsmokers to levels of estimated nicotine ex- posure which we have associated with the typical and most-exposed passive smokers in the population. We now correspond ETS nicotine etposure with lifetime lung cancer risk from ETS, by transforming the RSP-risk model we developed earlier("•18) into a nicotine%otinine-risk model. A linear exposure-respottse relationship is as- sttmed.'8) 5.1. Risk Estimation from ETS RSP Our previously developed acptuureiesponse model was based on modeling and field studies of nonsmokers' exposure to RSP from ETS and the response observed in a large cohort study of lung cancer in Seventh Day Adventist (SDA) nonsmokers.t`9-*m SDAs, becattse of their lifesryle, experience very little passive smoking rel- ative to demographically comparable nonsmoking nonSDA controls from the general population.('•`9-`*) This model predicted an ETS RSP exposure-Lung Cancer Death (LCD) response relationship, _E=5 LCDs per 10' per- son-years (PY) at risk per milligram of daily exposure to ETS-derived RSP. We validated our model by pre- dicting epidemiologically derived observational data to within 596 P= In the American Cancxr Society cohort studq(-u) of passive smoking in 176,739 nonsmoking women, their observed age-adjusted meaa lung cancer mortality rate was 13.3 LCDs/100,000 PY and the mor- tality ratio for women exposed to spousal smoking was 1.2 relative to women with no spousal smoking. Our exposure : response nwdel. applied - to shis cohort pre- . dicted the values 13.8 LCDs/100,000 PY and 1.19 for the rate and ratio respeciively." Similarly, this model predicts exactly the misclassificadon-adjusted odds ratio for spousal passive smoking and lung cancer derived by the U.S. EPA in its meta-analysis of 11 U.S. epidemio- logical studies of passive smoking and lung cancer extant in 1992,m as well as the odds ratio found in the key

468 Table II. Cotinine in Body Fluids of Nonsmokers (Various Studies)' Repace and Lowrey Study Micmenvironment rt Cone. (median) Comment A. Cotinine in urine ~ Cummings(s2' New York State 227 5 Max: 60-90 ng/mi Fontham• 5 U.S. Metro Areas 728 6.2 Max: 55-99 ngjmg Cr Fontham' • 1108 5.6 nghnl Ribolit"" 10 Countries, U.S. 1369 6 Mw. 55 ng/mg Cr Ha1ey'T=' (males) White collar, Teus 148 6.2 ± 0.5 (mean, SD), nghttg Cr (females) White collar, Tetas 112 8.0 = 0.8 (mean, SD), ng/mg Cr Perez-StabletIm Southwestern U.S. 189 NR° Max 70 ng/mg Cr Wallt•" California 48 4.7 nghng Cr Wtd. Mean 48 7.3 nglmi Wtd. Mean Haddowt•lt (females) Pottland, Maine B. Cotinine in plasma 232 1.1 Maa:10 ng/m1 Van Vunalcisnq Los Angeles 327 NR M= B-3o ng/ml Wagenlmecht<"t Urban adults 18-30 3445 NR M= 14 ngftl Perez-Stablet*4 Southwestern U:S. 189 NR Matc 14 ngfml • A number of studies in the late 1980s have e:omined the ttrine and plasma atxinine levels of adult noasmokers, using radioimmunoassay and gas chromatography. Median valuea found for utiaary ootinine are about 5-6 ng/ml (also sometimes eatpresaed in audely eqnivaient normalized units given by ng/mg Cteatiaine), and the atof[ used to distinguish heavy passive smokers from Gght active smokers is of the order of 55 to 90 ng/ ml or ng/mg Cr. For plasma cotinine, the median is about I npJml, and the peak value is about 10 to 15 ng/mL • Unpublished data from Ref. 29. ` NR, Not reported. epidemiologic study(29) upon which the U.S. EPAm re- lied for quantitative estimates of the impact of passive smoking on U.S. nonsmokers. 5.2. Rlsk Estimatioa from ETS Dlicotine and Cotinine . We now develop conversion factors relating ETS nicotine and cotinine exposure to lung cancer risk from ETS. -Aanosphaic nicotine: Using the value derived for the euposure-response 7, together with standard as- sumptions for of5ce occupancy and ventilation, and as- suming an average 40-year working lifetime (WLT). Repace and Lowrey<14M estimaoed the lifetime lung cancer death (LCD) rate for a person breathing ETS-deriVed RSP (at a rate of 1 m3/6r) in a model workplace to be -0-1.33-x---10-3-i,CD/(WLT µgW RSP): This rela- tionship implies that for a nonsmoking worker, an esti- mated lifetime risk of 1 x 10-6 is generated by breathing an (annualized) 8-hr time-weighted average (TWA) of 75 ng of ETS RSP per cubic meter of workplace air for a working Iifetime.(I`) Correspondingly, by applying the approximately 10:1 RSP-to-aicotine concentration ratio found from field . studies, as e:aempiified in Eq. (1), we derive an esd- ~. mated risk factor -0 =100 for nicotine in the workplace: tl~ = 1.33 x 10-'LCD//(WLT µg/m' Nicotine) (8) In other words, a 1 x 10-6 (de minimis) working lifetime risk is estimated to be generated by nonsmok- ers' exposure to 7.5 nanograms of atmospheric nico- tine from ETS per cubic meter of workplace air (ng/ m') (8-hr TWA). Cotntirte in body fltddL- Using Eq. (2), assn++tng a respiration p a 1 m' per houu for an 8 hr day and an exposure to~ 11T= 7.5 ng/m' of nicotine, we assume the corresponding estimated Iung exposure E= 60 ng of nicotine to be delivered per 24-hr day, since our equa- - tions are for the steady state, yielding aa estimated- de minimis cotinine value UA„6, = 2.6 x-10-a ng/ml urine corresponding to a 1 x 10-s lifetime risk of lung cancer. The estimated risk conversion factor; @=10-61Ujo,o,; - relating lifetime lung cancer risk from EfS exposure to urinary cotinine concentration is then cx;ressed as: ®_ . 4 x 10-` LCD per (WLT-ng cotininelml urine) (9) AA l and similarly tlsing Pt,,,,,, =4 x 10-` ngMll,e the plasma . . J

Indoor Air Standard for Tobacco Smoke cotinine concentration corresponding to de minimis life- time lung cancer risk, the risk conversion factor for plasma cotinine is given by: f2=2.5x 10-'LCD per (WLT-ng cotinine/ml piasma) (10) We now use the values derived for ® and S2 together with data from clinical~ epidemiological studies of cotin- ine in plasma and urine to estimate the risk from passive smoking for the typical and most-exposed nonsmokers. The weighted median urinary cotinine value estimated for female U.S. smokers, based upon existing studies (which are not a national probability sample), is, from Table II, 5.6 ng/ml, with the most-exposed female non- smokers exhibiting maximum urinary cotinine ievels about 10-15 times higher, depending upon the choice of cutoff discriminating heavy passive smokers from light active smokers. Most epidemiologists favor cutoffs in the range 55sU,,,,s90 ng/ml (see Table II). There are limited data on males, with some workers reporting male non- smoker cotinine levels somewhat lower than female 1ev- els, and others somewhat higher. We shall assume the exposures of males are comparable to that of females.tl8y Using Eq. (9), and a median urinary cotinine concentra- tion U.., = 5.6 ng/ml, the median working lifetime lung cancer risk for U.S. nonsmokers is estimated to be eU,,d = 2.2 x 10-', and the risk to the most-exposed (e.g., corresponding to a bar-waitress's exposuretsl) is estimated to be ©U,= = 2.2 x 10-=. Using IARC cotinine studies,c30> the risks appear to be similar abroad within factors of 2. Simflarly, From Table IIB, median female plasma cotinine levels P,,.d have been measured at P..d = 1.1 ng/ml, and peak levels are reported to range approxi- mately from 10 5 P,,.d s 15 ng/m1. We shall assume a 10=fold higher peak exposure, based upon our exposure model.(u-'&1') Using Eq. (10), the median lung cancer risk for U.S. nonsmokers is similarly estimated atfaPd = 2.8 x 10-', and the risk to the most-exposed is estimated at i2P,. = 2.8 x 10-2. These lifetime mor- tality probabilities are associated with current levels of exposure. The magnitude of the estimated risk will now be placed in perspective. 6. EVALUATING ETS RISKS 6.1. Concepts of Regulatory Risk We now consider whether the estimated' popula- tion risks from ETS which we have modeled are "ac- 469 ceptable" by societal standards for permissible human exposures to environmental carcinogens such as in- dustrial chemical emissions and radionuclides in air and water, and carcinogenic molds and pesticide res- idues in food. Several U.S. federal regulatory agencies promul- gate regulations and standards to protect the public from exposure to environmental carcinogens. It is of interest to inquire as to what levels of population cancer risk typically trigger regulation, what levels are beneath reg- ulatory concern, and how consistently are they applied among various federal agencies. Travis et 4ts't re- viewed the use of cancer risk estimates in prevading federal standards and in withdrawn regulatory initiatives, to determine the relationship between risk level and reg- ulatory action in 132 U.S. federal regulatory decisions of record. Travis et aL (n) describe two technical risk assessment terms: de manifestis risk and de mininus risk. A de rnanifesris risk is literally "a risk of obvious or evident concern," and has its roots in the legal definition of an "obvious risk" (i.e., one recognized instantly by a person of ordinary intelligence). De manifesris risks are those that are so high that U.S. federal regulatory agencies almost always acted to reduce them, and de minimis risks are so low that agencies almost never acted to reduce them. For various reasons, risks failing in be- tween these exiremes were regulated in some cases but not in othersts' ; however, residual risks after controltO are generally de rnvuntis. Travis et aL 9M found when the population at risk was large, as with ETS, de ntan- ffads risk oorresponded to 3 x 10-4, and de minirnu risk was 1 x 10-6. - 6.2. Population Risks Compared to Regulatory Levels How Aoes the risk from ETS compare with the fed- eral de nianifestis risk level? Using the available data for late 1980s cotinine concxnirations in nonsmokeis'Tody Quids, we have estimated the aggregate population risk from BTS at - 2 to 3 x 10-' (cMsisoent with esti- mates(6-5`) made using other methods), an order of mag- nitude above the de manffastis risk level. ...Another way of undcrstanding the import of such risk probabilities is to multiply the aggregate risk by the population at risk (in the case of lung cancer, nonsmok- ers aged z 35 years) in gfder to estimate the annual mortality. In 1990, there were - 50.7 millioir lifelong nonsmokers and - 34.6 million ex-smokers in this pop- ulation category.(55) A 2 to 3 x 10-' lifetime risk (as- suming 40 years' exposure to the working population)

~.5.?...., 470 corresponds to a 5 to 7.5 x 10-5 annual risk, and when applied to the nonsmoking population of 85.4 million nonsmokers at risk produces = 4000-6000 lung cancer deaths (LCDs) per year, consistent with the risk esti- mates of 5000 ± 2500 LCDs per year, adjusted'to 1988, produced by other methods.«6t By comparison, the U.S. Environmental Protection Agency has strictly regulated as Hazardous Air Pollutants under Section 112 of the Clean Air Act, airborne human carcinogens involving far lower numbers of estimated deaths, such as benzene (< 8 cancer deaths per year) (CDslyear) arsenic (<5 CDs/year), vinyl chloride (< 27 CDs/year) (all EPA es- timates before control and at the 95% upper confidence limit) ~Q In contrast, de rrrinimis exposure of the entire nonsmoking population at risk for a working lifetime of 40 years, would result in s 2 LCDs/year. 7. A WORKPLACE ETS STANDARD BASED ON MCOTINE AND COTININE 7.1. De M'rnimis and De Manffeslis Levels of Nicotine and Cotinine What are the maximum concentrations of cotinine in body fluids and the corresponding maximum airborne nicotine concentration estimated to be consistent with a de minimrs (10-6) lifetime population risk? From Eqs. 8, 9, and 10, the steady-state levels of cotinine in body fluids corresponding to de minimis risk are calculated to be Um,,;,, = 2.6 picograms of cotinine per milliliter of urine, and' PQ,,,,,, = 0.4 picograms of cotinine per mil- liliter of plasma, while the de nrinimis airborne nicotine level is calculated to be Nd,,;,, = 7.5 nanograms per cubic meter, 8 hr time-weighted-average ('I'WA). These values represent levels corresponding to maximum per- missible daily ETS exposure consistent with de mi,rirnis risk. Routine detection of cotinine levels this low from individual body fluid specimens is probably not possible with present technology, with tadioimmunoassay and standard gas chromatography methods having detection limits of about 0.3 ng/ml and 0.1 nglnil, and quantitation limits of 1 ng(ml, and 0.1 nghml, re.spectively.t"o Spe- cialized isotope-dilution tandem mass spectrometry methods can detect levels as low as 30 pglmi.(In How- ever, nicotine at this level in workplace -air could be detected at realistic sampling rates over a weekly period. The de mmtifestis cotinine and nicotine levels are estimated respectively at 0.8 ng/mI (steady-state urine cotinine) 0.12 nglml (steady-state plasma cotinine), and Repace and Lotvrey - 2.3 µg/m3 (air nicotine, 8-hr TWA). Standard methods will detect levels of this order in body fluids or air.ta'_'1s:'a- 6°t (in order to properly assess de minintis levels of ni- cotine in air or cotinine in body fluids, background levels(- due, respectively, to outgassing of surfaces or dietary or nonworkplace exposures to nicotine should be subtracted off.) 7.2 ETS Risks in the Workplace As Table III shows, nonsmoking workers of all cat- egories appear to be currently reeeiving nicotine expo- sures from ETS in offices, in industrial plants, in aircraft, and in restaurants with lax smoking policies, which ex- ceed the de manifestis level, sometimes by as much as an order of magaitude. Figure 1 gives a plot of ttte es- timated lifetime risk from breathing ETS at a rate of I m3/hr, as a function of daily wortplace nicotine concen- tration (8-hr TWA) at all practical levels of exposure. Achieving de stinimis risk requires several orders of magnitude reduction in exposure, according to measure- ments reported in Table III. Such reductions are not pos- stble using venti7ation or air deaning.tl`t Because of leaks driven by pressure imbalances in buildings induced by convection and piston effects from elevator motion, even in large buildings, it is improb,abie that separation of smokers and nonsmokers on different ventilation sys-. tems will readily achieve an acceptable level of exposure ( to ETS. In summary, based upon the available information on current ETS nicotine exposure levels shown in Table III, and the nicotine-risk model presented here, it ap- pears that the risks to white collar, blue collar, and ser- vice workers from ETS in many workplaces consider- ably exceed the de manifertis level which triggers strict federal regulation of carcinogens. Control of nonsmok- ers' ET5 exposure to levels of de niinirnis risk using methods short of complete elimination of smoking in the workplace does not appear likely-particularly if the ad- ditional risk of heart disease mortality from ETS is con- sidered. 8. CONCLUSIONS 1. We have estimated that the typical' U.S. non- smoker of working age during the 1980s ap- peared to be exposed to a nicotine intake of ~ approximately 143 µg per day, while the most- )

Indoor Air Standard for Tobacco Smoke Table 111. Nicotine in Personal Air of Nonsmokers and in Building Air (Various Studies)' Study %tic.roenvironment rt µg/m' Comment A. Nicotine in nonsmokers"air. Personal monitors 471 Schenker'"' CoultasO" Mattson'"I Railroad eierks, N:E. 40 White Collar N:M_ 15 Flight Attendants 4 (Air Canada) 6.9 20.4 = 20.6 4.7 = 4.0 Workshift median Workshift mean = SD 4 F7ights. mean - SD B. Nicotine in building air. Area monitors LeadererCt"' Homes, N:Y. State 47 .2.17 = 2.43 7-day av., smoking Hatnmondt"" Mass. Industrial White collar 24 60 21,5 :t 40.2 9-hr workiLift av. (uaasmoker's air, Blue coliar 123 8.9 t 16.8 smoking allowed Food service 51 10.3 s 12.6 on preatises) Carsont7Oj Offices, Canada 31 11 Workday samples Miesne&'t Wtxkplatxs, Mass. 11 6.6 = 7.6 Workweek average 0ldaker('t1 Restaurants, N.C. 33 10.5 (.1-35) 1-hr iv. (range) leakins<'" Knoavi11e,1N'Metro Restaurants 7 3.4 s 2.3 Z 1-hr average Cocktail lounges 8 17.6 s 22.8 Bowling alleys 4 10.7 s 5.1 Gaming Parlors 2 10.7 = 3.0 Laundromats 3 2.0 s 0.7 Airport gates 2 6.0 = 5.5 Office 1 6.0 s 2.9 Nagda-i U~S. Aircraft (All flights) 69 13.4 t 14.7 In-flight average Smoking section (Domestic) 61 0.11 i- 0.13 Nonsmaking section (International) 8 0.33 z 0.23 Nonsmoking section Vaughnt9jt Highrise office bldg. 1 2.0 z 0.15 Nottsm. air; 9•hr av. • Measurements of workplace tticotine aoi>oeutratioas in the late 1980s using both petsottal and area mtto:itors in workplaces where smoking is permitted, demoasaate (from the tneans and standard deviations), a range in nicotine concentration N of about 1 µglm' to more than 100 µghn', indicating corrcspondiag workplace nicotine exposures E ranging from about 10 pWday to more than 1000 µg/day, using Eq. (2), assuming workdays of the otder of 8 + hr and respiration rates of the order of 1 m'/hr. Thisis consistent with our predicted values for the typical nonsmoker of E,,, - 143 µg/day and E_ - 1443 µg/day, for the tlmst-eposed, which also include domestic exposures to ETS. exposed nonsmokers appeared to have an in- take of about. 1430 µg per day. These values should decline in the 1990s as U.S. smoking prevalence decreases and restrictions on smok- ing increase. 2. For urinary cotinine, our n•lodel predicts a corre- sponding median value of about 6.2 ng of co- tinine per ml of urine for the typical U.S. nonsmoker from passive smoking during the 1980s, and'a.value of about 62 ng/nil for the most-exposed nonsmokers. For plasma cotinine, we predict a corresponding median value of 0.95 ng of cotinine per ml of plasma for the median nonsmoker, and 9.5 ngrtml for the most-ex- posed. These predicted values are consistent to within 15% of the results of late 1980s clinical epidemiological studies relating passive smok- ing and cotinine in body fluids. 3. We have derived a health-based standard for en- vironmental tobacco smoke based on its sunro- gates, atmospheric nicotine and cotinine in body fluids of nonsmokers. For atmospheric nicotine - in the workplace, the de ntinimis or "accepta- ble" lifetime risk level of 1 lung cancer death per million nonsmokers at risk occurs at 7.5 nan- ograms per cubic meter (8-hr time-weighted av- erage.) For cotinine,in body fluids, de minimis risk occurs at a daily average level of 2.6 pi-_._ cograms of cotinine per milliliter of urine ex- creted, or at a level of 0.4 picograms of plasma

472 E :. .:: ~ 10 -7 L_ de minimis concentration K .001 .01 .i 1 10 ETS Nicotine Concentration (micrograms, per cubic meter) Repace and Lowrey too Fig. L Emironmental tobacco smoke induced lung aneer risk vs. workplace nicotine air coocentntion (8 hr TWA). ETS Risks to workers may be evaluated using this model on the basis of nicatine measurements in individual aaorl•sites. Workplace niaotiae mncenttations typically appear in the range l to 100 µg/m' as shown in Table III. The de maiufestir kwel of carcinogenic risk occurs at an exposure wnesntration of 2.3 µ®Im' daily workshift ezposure persisting over a working lifetime. The etposune concentration corresponding to de minvnis risk occurs at 0.0075 µgfm', determining the level'of the standard. (1 ppb nicorine in dry air @ STP - 1.3 µg/m'j. cotinine. De manifestis risk occurs at 2.3 micro- grams per cubic meter of workplace air, 8-hr time weighted average (TWA). For cotinine in body fluids, de manifesrfr risk occurs at 0.8 nan- ograms of cotinine per mililliter of urine, or 0.12 nanograms of cotinine per milliliter of plasma. The exposure- and: dose-response relationships upon which these standards are based are de-_ rived from our previously published exposure- response model, and have been validated by physical, clinical, and epidemiological data. .4. We estimate the median 1980s U.S. lung cancer population risk from passive smoking to be about 2 x 10-', and estimate that the most heavily exposed nonsmokers have a lifetime lung cancer risk from ETS of about 2%. IARC data on uri- nary cotinine from 10 countries suggests that ETS risks appear to be similar abroad. Based on recent field studies of nicotine in workplaces, it appears that average ETS exposure levels of U.S. white-coilar, blue collar, and service workers may exceed by an order of magnitude the de manifes7fs risklevei of 3 x 10-`, beyond which environmental carcinogens historically have beeni strictly regulated by U.S. federal regulatory agencies. 5. Using the information presented in this work, it is possible for the first time to make an enforce- able cancer-risk standard for environmental to- bacco smoke in the workplace, using nicotine in air, and cotinine in body fluids, based on the concept of de minimis risk. Using this de mia- imrs risk standard for ETS, together with newly developed active and passive area and personal nicotine monitors and sensitive cotinine assays, it is now possible to monitor individual work- places or individual workers to determine risk from the workplace carcinogen ETS using standard industrial hygiene techniques. This proposed standard permits the efficacy of any, risk management strategy (i.e., workplace smoking policy) to be evaluated at the work- site and compared to federal regulatory criteria for acceptable risk for exposure to environ- mental carcinogens. N - . ~ N ~