they could reduce the effect of falls on theolder adult population. El

AcknowledgmentsThis study was supported by the HarborviewInjury Prevention and Research Center, grantCCR 49-002570, Centers for Disease Control.

References1. Hindmarah JJ, Estes EH. Falls in older

persons. Arch Intem Med. 1989;149:2217-2222.

2. Perry BC. Falls among the elderly: a re-view of the methods and conclusions ofepidemiologic studies. JAm Genatr Soc.1982;30:367-371.

3. Fisher ES, Baron JA, Malenka DJ, et al.Hip fracture incidence and mortality inNew England. EpidemioL 1991;2:116-122.

4. Grisso JA, Kelsey JL, Strom BL, et al.Risk factors for falls as a cause of hip frac-ture in women. N Engi J Med. 1991;324:1326-1331.

5. Tinetti ME, Speechley M. Prevention offalls among the elderly. N Engi J Med1989;320:1055-1059.

6. Tinetti ME, Speechley M, Ginter SF. Riskfactors for falls among elderly persons liv-ing in the community. NEngIJMed 1988;319:1701-1707.

7. Ray WA, Griffin MR, Downey W. Benzo-diazepines of long and short eliminationhalf-life and the risk ofhip fracture.JAMA1989;262:3303-3307.

8. Wailer JA. Injury in aged: clinical and ep-idemiological implications. N Y State JMed 1974;74(November):2200-2208.

9. Wild D, Nayak USL, Isaacs B. Prognosisof falls in old people at home. JEpidemiolCommunity Health 1981;35:200-204.

10. Mossey JM, Mutran E, Knott K, Craik R.Determinants of recovery 12 months afterhip fracture: the importance of psychoso-cial factors.Am JPublic Health. 1989;79:279-286.

11. MagazinerJ, SimonsickEM, KashnerTM,Hebel JR, Kenzora JE. Survival experi-

Public Health Briefs

ence of aged hip fracture patients. Am JPublic Health. 1989;79:274-278.

12. From the WONDER computer network.Atlanta, Ga: Centers for Disease Control;1991.

13. Rice DP, Mackenzie EJ, and Associates.Cost of injury in the United States: a reportto Congress, 1989. San Francisco: Institutefor Health and Ageing, University of Cal-ifornia, San Francisco; Baltimore: InjuryPrevention Center, School of Hygiene andPublic Health, The Johns Hopkins Univer-sity; 1989.

14. Centers for Disease Control. MMWR Sur-veillance of major causes of hospitalitionamong the elderly, United States, 1988.MMWR CDC Swveili Swmn. April 1991.

15. Sattin RW, Lambert Huber DA, DeVitoCA, et al. The incidence offall injuryeventsamong the elderly in a defined population.Am JEpidemioL 1990;131:1028-1037.

16. Nickens H. Intrinsic factors in fallingamong the elderly.Arch Intem Med. 1985;145:1089-1093.

Carcinogens in Tobacco Smoke:Benzo[a]pyrene from CanadianCigarettes and Cigarette TobaccoMurray J. Kaisennan, PhD, and William S. Rickert, PhD

...............t

.........?.

Inrodudion

The particulate fraction of tobaccosmoke ("tar") is composed of a complexmire of constituents,1'2 some of whichare regulated under environmentallegislation.3-6 One constituent is ben-zo[a]pyrene (BaP), a polynuclear aromatichydrocarbon formed during the incompletecombustion of organic matter such as gas-oline, garbage, and plants. BaP has beenidentified by the Intemational Agency forResearch on Cancer as an animal carcino-gen and a probable human carcinogen((lass 2A)7 with inhalation, oral ingestion,and dermal absorption as the importantroutes of entry.

Human data are unavailable, but oralingestion of about 7 to 9 mg of BaP perkilogram has produced cancers in labora-tory animals.8 With respect to tobaccosmoke, BaP has been detected in concen-trations ranging from 20 to 40 ng per cig-arette in mainstream cigarette smoke' to40 to 79 ng per cigarette in sidestreamsmoke1 and 96 to 292 ng per cigar in main-stream cigar smoke.9

On January 1, 1989, the TobaccoProducts Control Act came into force inCanada. One of the purposes of this leg-islation is to "enhance public awarenessof the hazards oftobacco use by ensuringthe effective communication of pertinentinformation to consumers of tobaccoproducts."10 To achieve this goal, theHealth Protection Branch of Health andWelfare Canada has undertaken a seriesof studies to develop and validate analyt-ical test methods for toxic constituents.

In this paper we report on the BaPdelivery of 35 brands of commerciallyavailable Canadian cigarettes and 5 brands

Murray J. Kaiserman iswith the Tobacco Prod-ucts Section, Environmental Health Director-ate, Health and Welfare Canada, Ottawa, On-tario. William S. Rickert is president of LabstatInc, Kitchener, Ontario.

Requests for reprints should be sent toMurray J. Kaiserman, PhD, Tobacco ProductsSection, Health and Welfare Canada, Ottawa,Ontario, Canada K1A 012.

This paper was submitted to the JournalMarch 4, 1991, and acceptedwith revisions No-vember 27, 1991.

American Journal of Public Health 1023

Public Health Briefs

of commercially available Canadian fine-cut tobaccos for making roll-your-owncigarettes.

MethodsSmoking Procedures and SampleCollection

The 35 brands of cigarettes and 5brands of fine-cut tobaccos were collectedat the retail level in March and April of 1989in Montreal and Vancouver. Brands werechosen to include some ofthe most popularbrands in the market, including PlayersLight Regular Size (12.9%o market share),ExportARegular Size (5.7% market share),and Rothmans King Size (3.7% marketshare).-1 The preparation of roll-your-owncigarettes and the smoldng conditions for allcigarettes are descrbed elsewhere.12

BaP was collected and analyzed ac-cording to the method described byTomkins et al.13 In this procedure, 14C-

labeled BaP is added to each sample andvalues, as determined by high-pressureliquid chromatography (HPLC), are cor-rected for "'C recovery.

Recovery Study

Five Canadian control cigarettes(Monitor 4) were smoked per port on 10randomly selected ports of a 20-port smok-ing machine and the mainstream particu-late fraction was collected on a Cambridgefilter pad, providing a total of 10 samples.

During cleanup, 100 Lld of tracer so-lutionwas added to six ofthese samples atsix different steps before HPLC isolationand 25 p,l of tracer solution was added totwo of the samples at two different stepsafter isolation and before BPLC quantita-tion. To determine 100%o recovery, two ofthe samples were left unspiked (no tracersolution added) until just prior to injectiononto the BPLC analytical column.

Validation StwdyFive Kentucky Reference 2R1 ciga-

retteswere smokedperporton lOrandomlyselected ports ofa 20-port smokigmachineand the maeam tar was collected on aCambridge filter pad, resulting in another 10samples. All samples were spiked with 100pld of the tracer solution. Five of the padswere further spiked with 107.6 ng of unla-beled BaP. The workup then proceeded asdesCbed by Tomkins et al.13

Resms and DiscussionRecovery and Validation Studies

The recovery and validation studiesprovide a measure of the precision, accu-racy, and efficiency of the methods used.Observed recovery rates of BaP were100%o + 8%. During the course of the ex-periment, the average recovery rate was89% for the control cigarettes and 91% forfine-cut tobacco cigarettes; these rateswere in good agreement with published re-sults of85% BaP recovery.13 The results ofthe validation study provided a value of31.9 _ 3.7ngpercigarette for the unspiked2R1 cigarette; thisvaluewas in good agree-ment with published values of 32 ng percigarette9 and 28.4 ± 3.4ngpercigarette.13

Cigarette and Fine-Cut TobaccoStudies

Table 1 shows declared pack valuesfor tar, nicotine, and carbon monoxide, andvalues for BaP as determined in this study.The results for BaP are averages of fiveobservations per brand (25 cigarettes). (Al-though the current practice is to reportyields on a per-cigarette basis, it should bepointed out that five cigarettes are requiredfor each analyfical value and the per-ciga-rette yield is obtained by division.)

Mean values for BaP range from 3.36ng per cigarette to 28.39 ng per cigaretteand are linearly related to declared tarval-ues (correlation coefficient of 0.89 for allsamples). Because the reported values re-flect smoke concentrations under stan-dard conditions, the smoker of low-yieldcigarettes is probably exposed to higheramounts than those reported in Table 1.

Figure 1 shows the results of a re-gression analysis ofmean BaP on declaredtar. Although the simple linear correlationcoefficient is fairly high at .89, the pointsare widely dispersed about the regressionline, resulting in broad 95% confidencelimits for the regression line. This means,for example, that at 16mg declared tar, theBaP level is predicted to lie between 15and 28 ng per cigarette. Thus, although

1024 American Journal of Public Health July 1992, Vol. 82, No. 7

30-r' '.. 11-111 -......- I.... ..

20t.Xrll20 .. .. .. ... ... .. ... .. ,5 ,. .... .. .. .. .. .. .. .. ..

115 -;........I..

. t #' ~ ~ ~~~~~............. ........O -' _ ',._

iII I | | | |IIiI I i i

0 4 8 12 16 20

Tar Conent, mg/dg

FIGURE 1-Regression of mean benzo[ajpyrene (BaP) on declared tar content

TAUII~~~~~~~~~~~~~~sad 0wIwsdTum~~~~~~~~~~~~~~~~~~~~~~~~~~~~.and Csbon~~~~~~~~~~~~.......

--------------- 00- ----------a----........ ....... ........ .............

7.......QrwanA 10 15 1023~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~.........19 1.5 19221~~~~~~~~~~~~~--------

Nwu*p7 19 1510~~~~~~~~~~~~~~~~.......

BaP is linearly related to declared tar, de-clared tar is not a good predictor of BaP.

Table 2 shows declared tar, nicotine,and carbon monoxide values and meananalyticalBaPvalues for the fivebrands offine-cut tobaccos used in preparation of

July 1992, Vol. 82, No. 7

roll-your-own cigarettes. Once again, BaPyields are averages of five observationsper brand. These values range from 22.92to 26.27 ngper cigarette; the averagevaluewas 24.7 ng per cigarette.

The Occupational Safety and Health

Public Health Briefs

Administration (OSHA) in the UnitedStates has set advisory levels for BaP of0.2 p,g/m3 (8-hour time-weighted averagepermissible exposure limit for occupa-tional exposure).5 Assuming averagebreathing rates of 0.63 m3/hour (women)to 1.08 m3/hour (men) for alternating pe-riods of light work and rest,'4 the permit-tedmaximum amount ofinhaled BaP in anoccupational environment would be from1.01 p,g in 8 hours to 1.73 p,g in 8 hours. Apack-a-day smokerwho consumed 20 cig-arettes in 8 hours could be expected toinhale from 0.067 p,g to 0.568 ,g from thissource. Although the actual amountwould vary depending upon individualsmoking behavior, it is obvious that, forsmokers, cigarette smoke is a significantsource ofexposure to BaP, contributing asmuch as one third the daily permitted ex-posure recommended by OSHA.

To clarify the risk further, the state ofCalifornia mandates a "no significantrisk" level for BaP of 0.06 p,g per day.6 Apack-a-day smoker (20 cigarettes) will in-hale between one and ten times this level.For smokers, current advisory levelsshould be modified to take into accountthe contribution from tobacco smoke.

Although BaP is just one of 60 to 70known toxic constituents of tobaccosmoke,1',7 many smokers continue to be-lieve that tar is the most significant toxicconstituent of tobacco smoke and that cig-arettes with reduced tar are safer.1"15 Onereason for this misconception may be lackof knowledge of the components of to-bacco smoke in a famfliar context. l

References1. US DepartmentofHealth andHuman Serv-

ices. ReducingtheHealth Consequences ofSnmking: 25 Years ofProgress.A Reportofthe Surgeon GeneraL US Dept of Healthand Human Services, Public Health Ser-vice, Centers for Disease Control, Centerfor Chronic Disease Prevention and HealthPromotion, Office on Smoldng and Health.US Dept of Health and Human Servicespublication CDC 89-8411.

2. Dube MF, Green CR. Methods of collec-tion of smoke for analytical purposes. Re-cent Advances in Tobacco Science: For-mation, Analysis and Composition ofTobacco Smoke. 1982;8:42-102.

3. Canadian Environmental Protection Act.RSC C16 (4th suppl); 1988.

4. US Environmental Protection Agency.Ambient Water Quality Criteria for Poly-nuclear Aromatic Hydrocarbons. Wash-ington,DC Office ofWater Regulations andStandards; 1980. EPA document 440/5-80-069. NTIS no. PB81-117809.

5. US Occupational Safety and Health Ad-ministration. Penni5nibleEwoswt Lts.CER 29:1910.1002 and 29:1910.1029; 1985.

6. 22 Cal Code of Regulations § 12711.7. International Agency for Research on Can-

American Journal of Public Health 1025

Public Health Brieif

cer (IARC).Monogrphson the Evalationof Cawciroenic Risky to Hwnans. Lyon,France: LARC; 1987:58. Pub. Suppl 7.

8. Tazcoogical Profile for Benzow[apyne.Atlanta, Ga: US Dept of Health and Hu-man Services, Public Health Service,Agency for Toxic Substances and DiseaseRegistry; 1990. ATSDR/TP-85/05.

9. Appel BR, Guirguis G, Kim I-S, et al. Ben-zene, benzo(a)pyrene, and lead in smokefrom tobacco products other than cigaettes.Am JPublic Heath 1990,80:560-564.

10. The Tobacco Products Control Act. RSCC20; 1988.

11. Maxwell JC. How the brands ranked.Word Tobacco. September 1990:121-124.

12. Kaiserman MJ, Rickert WS. Handmadecigarettes: it's the tube that counts. Am JPublic Health. 1992;82:107-109.

13. TonmkinsBA,Jenkins RA, GriestWH, Rea-gan RR, Holiday SK Liquid chromato-graphic determination of benzo[alpyrene intotal particulate matter of cigaette smoke.JAssoc OffAnal Chemr 1985;68:935-940.

14. Rickert WS (Labstat Incorporated). Someconsiderations when estimating exposureto environmental tobacco smoke (ETS)with particular reference to the home en-vironment. Can JPublic Health. 1988;79:S33-S39.

15. Project Vildng, "WAVE 2." Entered asevidence (AG-021D) in RJR-MacDonaldInc v Attorney General of Canada (Supe-rior Court ofCanada, District ofMontreal).Court docket no. 500-05-009755-883; 1990.

Discrepancies between Self-reportedSmoking and Carboxyhemoglobin: AnAnalysis of the Second NationalHealth and Nutrition SurveyLisa M. Kesges, MS, Robert C. Kiesges, PhD, and Jeffrey A. Cigrang MS

................................

*Xx

Xx.

MPR -m%-.-

K2...........

?iUziV

... ......

IntrdudionBecause of uncertainties inherent in

measurement of smoking behavior usingself-report,' biochemical measures arefrequentlyused tovalidate reported smok-ing exposure. Although cotinine is oftenthe preferred biochemical measure ofsmoking exposure,2 blood carbon monox-ide concentration is useful in many cir-cumstances. It is less expensive to mea-sure and can discriminate smokelesstobaccoproducts from cigarette, pipe, andcigar smoking. It can also verify smokingcessation in treatment programs using nic-otine replacement. Carbon monoxide isparticularly useful in establishing healthrisks of smoking because of its strong re-lationship with cardiovascular disease.3

The current study sought to investi-gate factors associated with discrepanciesbetween self-reported smoking status andcarboxyhemoglobin levels. By identifyingcircumstances underwhich carboxyhemo-globin levels and self-reported smokingsta-tus do not agree, investigators can makemore informed judgments about the valid-ity of these measures of smoking and theassociationsbetween smokingand morbid-ity when carboxyhemoglobin is used.

MehodsThe study sample consisted of 6032

participants from the Second NationalHealth and Nutrition Examination Survey

(NHANES II), aged 18 to 74 years, whocompleted a carboxyhemoglobin assess-ment. Detailed description of the surveymethodology for NHANES II is avail-able.4 Self-reported smoking status wasascertained by asking participants wheth-er they currently smoked cigarettes, howmany they smoked on average, andwhethertheysmoked cigars or pipes. Sub-jects reporting no current use of tobaccoproducts were considered nonsmokers,while those smoking pipes or cigars wereexcluded. Carboxyhemoglobin concen-trations, determinedby spectrographic as-say,5 are expressed as percentages oftotalhemoglobin combined with carbon mon-oxide. Statistical analyses were per-formed using the Statistical AnalysisSystem.6 Collinearity diagnostics, as dis-cussed by Hosmer and Lemeshow,7 didnot reveal unreliable estimates.

Lisa M. Klesges iswith the Department ofBio-statistics and Epidemiology, University ofTen-nessee, Memphis. Robert C. Kiesges and Jef-frey A. Cigrang are with the Center for AppliedPsychological Research, Department of Psy-chology, Memphis State University, Memphis,Tenn.

Requests for reprints should be sent toLisa M. Klesges, MS, Department of Biosta-tistics and Epidemiology, University of Ten-nessee, Memphis, TN 38163.

This paper was submitted to the JournalFebruary 4, 1991, and accepted with revisionsDecember 31, 1991.

July 1992, Vol. 82, No. 7