Tobacco Documents Online

..E\"ViRt)hMEATAL RESEARCH~ 12, '63-274(i1976)' 'Tar" and Nicotine Content of Cigarette Smoke in R'edation to Death R'ates' E. CUYLER HiAM.Mt)ND,Z LAWRENCE CiARFIXb:Eii,3 HERBERT SEIDMA.Y,s ~- MJD EDWARD A. LEtWs Department of Epidemiology and Stotiatics; American Cancer Siociety, New Ynrk, New York 10017 Received September 10, 1976 Over 1,t100t000~menand women whoenrolted in an epidemiological study in 1'939-1960were (rovith few exceptions) traced for 12' years. They all answered' questionnaries on cigarette smoking and various other factors at time of enrollment: and survivors answered repeat questionnaires on three later occasions: Inithis analysis. cigarette smokers were c3assiHed by the amount'of tar, and nicotine delivered by the brand they usually smoked at the start of each of two 6.year periods. Among subjects who smoked the same number ofcigarettes a day, total' death rates, death rates from coronary heart,disease; and death,rates from lung cancer were somewhac:lower for those who 1 smoked "low" tar-nicotine cigarettes than for those who smoked "high" tar-nicotine cigarettes. The:death rates of'subjects who smoked! "low"' tar-aicotinr cigarettes were far higher than the death rates of subjects who never smoked regularly. INTRODUCTION .. Many years have passed since the following, was firrnly established by a large number ofepidcmioliogical studies carried o'utby independent investigators in this country and'abroad (U.S. Public Health Service, 1964): First, and most important, death rates are higher'in smokers than ininonsmokers and' increase with degree of exposure to tobacco smoke. Among the diseases involved in this relationship are: (Y) lung cancer and cancer of several other sites, including the Iip, tongue, mouth, larynx, pharynx, esophagus, and urinary bladder; (2) coronary heart. diszase,. : stroke, and aortic aneurysm; (3) chronic bronchitis and emphysema; and (4) several' . other diseases including peptic ulcers. Thz agz-speeiFc Iun'g cancerdeath rates of men whosmoked cigarettes regularly was found to be about ten times as high as the lung cancer death rate of men who never smoked; and this ratio was considerably higher among men who smoked 40 or more cigarettes a~ day. Expressed in the same terms (i,e., mortality ratios), the coronary, heart disease death rates of male'cigarette smokers were found to be about. 1.5'to 3.0 times as highidepending,uponage and'arnount of smokirngas the coronary heart disease death rate of nonsmokers. From this it might be concluded that lung "From a paper given at the Conferenee, on the Origins of Human Caneer at Cold Springs I°arbor Laboratory, P1ew York, September 14, 1976, I Sc:D.. Vice President 3 Nt A'., Assistant Vice President. • M.B.A., Chief of Statistical Analyses. s F.S.A:,,Consultant: Copyriqhc'y'. 1976 brArademie Press. tnc. AU ingius of hrpra/uction iin any. form resernM. 263

HAk41io1V D' ET AL. " cancer is far more highly associated with cigarette smoking than is coronary heart .'disease. While this is true, it tells only half the story. . Lung cancer is a rare disease in nonsmokers, whereas coronary heart disease iss the leading,cause of death among male nonsmokers in the.United States. luluitiply- ing the high CHD death rate of nonsmokers by 1.5' results in far more "excess deaths" than multiplying the very low lung cancer death rate of nonsmokers by 110. Thus in terms of reduction in life expectancy the association between cigarette smoking and coronary heart, disease is far more important than the association between cigarette smoking and lung cancer. While such information is of scientific interest, it is of no value to the public unless it can be utilized to protect health and extend life expectancy. At the outset, several possible ways of putting, the knowledge to practical! use were discussed. Among these were: (1) Simply acquaint the public with the facu. According,to the proponents of this idea; knowledge of the facts would be sufficient to persuade people not to smoke cigarettes. (2) Conduct extensive public education prognams to persuade people not to smoke cigarettes. (3) Attempt to develop a new type of cigarette which would have no harmful effect&(or at least~ greatly reduced harmful effzcts) and yet would be pleasurable to smoke. All three of these things have been, done. Presumably, due to application of'the first two ideas, a great many erstwhile cigarette smokers have given up1he habit (U.S. Public Health Service, 1976). However, millions of'adult cigarette smokers have continued to smoke and each year, tens of thousands of children and youths take up the habit. Many attempts have been made to develop a'"safe" or'"less hazardous'"'type of cigarette which smokers would! find satisfactory. The only procedure along these lines which has as yet gained wide acceptance among cigarette smokers is reduction in amount of tar and nicotine in the mainstream smoke of cigarettes. Cigarettes with considerably reduced tar and nicotine (ofteni referred to as "low" T/N ciga- rettes) have now been available for many years. We are concerned here with the question as to whether or not such cigarettes are actually "less hazardous" than, are '"high." T/N cigarettes. ,< Some years ago, a small committee of experts on the subject came to the conclusion that "the preponderance of scientific evidence strongly sug~ests 'that the lowerthe'tar' and'nicotine contenUofcigarette smoke the less harmful would be the effects" (U.S. Public Health Service, 1968). (A short, time later this was reported by the then Surgeon-Generali of' 'tlte U.S. Public Health Service.), Their reasoning was about as follows: (1) Death rates from lung cancer, cancer of several other sites, coronary heart disease, and several other diseases increase witih degree of'exposure to cigarette smoke. (2) Many experimental studies had shown that material condensed from cigarette smoke (usually called "tar") ") is carcinogenic when applied to animals. ('3)The known acute effects of nicotine upon the heart and circulatory system suggested thatthe nicotine content of cigarette smoke was partly if not entirefy responsible forthe fact that age-specific death rates are higher among cigarette s suppose tt the harm ( The tern conclusior (1) If th, smoke mc turned out T/N to "!c Hammonc (2) Smc inhale the their effec exposure (3) It cc harmful tls that, in cc certain ga might incr Therefc in age-spe Since t. stuoiies (B that, peopi people wh panied by disease) lG Betwee can Canc epidemiol a lengthy for 6 year raaire cont during the questionn series, Q4 after June The So I!96+t„ and traced in dIscontim eighth an. Durine by Divisir

"Te#R" AND NLCOTI\iE CONTENT OF eLGARETTES' 265 cigarette smokers than amo'ng,nonsmo'kers: (4) Therefore, it'seerned'reasonable to suppose that if the tar and nicotine content of cigarette smoke were reduced, then the harm done per cigarette smoke'd' would be correspondingly reduced. The term "'strongly suggests'"' included in the above statement iniplied that the conclusion might be intorrect. The major counter speculations were: (1) If the tar and nicotine content of'the smoke were reduced most smokers might smoke more cigarettes per day and thereby cancel the benefit. (This speculation turned out to be incorrect wheniit was found that, smokers who switch from "high" Tlli`N to "'low" T/N cigarettes do not usually increase the number smoked per'day; Hammond and Garfinkel, 1'964.) (2) Smokers of "'low" T/N cigarettes' might (consciously or unconsciously) inhale the smoke more deeply than smokers of''"high" T11`1' cigarettes. If'so, then their effective exposure to tar an'd'i nicotine might not be reduced while their exposure to the gases in cigarette smoke would be increased. (3) It couldibe that gases contained in cigarette smoke are as harmfttl'if'not more harmful than the''tar" and nicotine content of the smoke. Furthermore, it, could be that, in certain circumstances, reduction in TIN is accompanied by' an increase in certain gases, most notably' carbon monoxide. An increase in carbon monoxide _tnight increase the risk of coronary heart' disease. Therefore, ifal7 this is true, the net effect ofreductiion inT/N mightbe anincrease in age-specific death rates. Since that time, published, evidence from two retrospective epidemiological studies (Bross and Gibson, 1968 and Wynder et at.. 19'10)' was such as to indicate that people who smoke filtertipcigarettes have lbwer lung cancer death, rates than people who smoke nonfilter cigarettes. Concern lest this desirable effect be acco'm- panied by an increased' risk of some other disease (especially coronary heart disease) led~ us to carry out this investigation. MATE R'IAL I can Cancer Society enrolled over 1',000,00b' men and' women in a prospective epidemiological study (Hammond,1964),. Upon enrollment, each subject answered a lengthy questionnaire (hereafter referred to'as'Ql),. Once every 2 years thereafter for 6 years, surviving subjects were requested to, answer a brief repeat question- naire containing questions on cigarettesmoking, hospitalization, diseases incurred during the interval between ques'tionnaires, an'diseveral'otherfactors. Tttese repeat questionnaires will be referred to as Q2, Q3, and Q4. The last' questionnaires of this series, Q4, were distributed on October 1, 1965; but some were not answered until after June 1, 1966. The Society successfully traced 99.6% of the subjects through September 30,. 1'964', and 98:4G''o through September 30, 1965. Of those who were still living,wheni trace& in the sixth annual follow-up, 94.9%, answered Q4. Annual tracing was' discontinued after the sixth follow-up, but was resumed on October 1, 1971. The eighth and last' tracing was started on October 1, 1972. During the first 6 years, enrollment and tracing of the subjects was administered by Divisions of'the Amenean Cancer Society in 1121 counties in 25 states. When Between October 1, 1959, and ;viarch 3'1, 1960; volunteer workers of the Ameri-

,/itaalt QM aa 0 © OPP a 0 a X . •= rY !] tE 266 HAMMOND ETAL. tracing was resumed in 1971, 3' of these states dropped out for administrative reasons; and we decided'not to attempt t+a trace those few subjects who were "lost" in the sixth follow-up. Thus, on Oictober 1,197'U, we set out to trace 897,825 subjects who had been traced in the sixth follow-up and who were stilUliving as of'September 30, 1965. We traced 98.4% through September 30, 1971, and 92.8575 through September 30, 1972. This report is confined to experience dt2ring two 6-year periods of time: Period 1, July 1~„ 1'960-June301 1966; and Period 2, July 1, 1196'6-Jiane30,1972: It is further eonfnedito subjects 40 years of age or older as of July 1, 1960, who on Q!1' said that they were currently smoking cigarettesregtalarlyand had never smo~kedpipesor cigars regularly; and who on Q1 stated', the number of cigarettes they currently smoked per day and their usual brand of cigarettes.6 Experience during Period2 is further confuned'to subjects who answered Q4 and who on that questionnaire said that they were currently s~moking! cigarettes, stated the number they smoked per day, and the brand of cigarettes they usually smoked' at that time. METHODS Information on the tar andinicotine content of the smoke from various brands of cigarettes was obtained from several sources." On the basis of this information, each subject was classified according to the tar and nicotine content (H, "' high'"; M4 'amed'ium'"; or L, "low") of the cigarettes he usually smoked arthree points in time as indicated in Ql, Q2, and Q4. This was relatively easy for Q1, since atthattime, although some manufacturers marketed two ormore types of cigarettes under the same brand name they could bee distinguished by presence or absence of aflter or by menthol. For purposes of this report„we d'efined"high" Th1 as 2.0'to2.7'mg of nicotine and 25.8'to 35.7 mg of'tar. "'Low" TlIN was defined! as less than 1.2 mg of' nicotine; and with very few k exceptions, cigarettes which meet this qualification also delivered less than 17.6 mg ' of tar. "Medium" IN, was simply defined as intermediate between "high"" and' "low."" _. Insofar as possible, we used. these same definitions for later years. However, some manufactureres marketed, under the same brand name, two or more types of eivrettes.vhica differed in tar and nicotine content. Forthis reason, it is likelythat some of the subjects who were placed in the "a'high" TLP<1 category probably belonged in the "medium"" TINI category and vice versa. There was far less difficulty of this sort in distinguishing the "low" T7N smokers from the other two groups. For the period 1966-i'972,, the three sets were distinguished as follows: (1) "High""' was defined as subjects in the "'high'" category in the 1959-1!960 question- • One additional analysis includes subjects who had never smoked regularly. t Tar and nicotine content of smoke:from various brands of cigarettes in various years waa obtained from analyses madt by Foster D. Snell; Inc.. and!pubtished in the November 1959 issue of Readers Digest; in the July 1961 issue of Readers Digest; and for 1965, interpolated from the Readers Digest August 1963' issue and the Federal Trade Commission ratings published in November 1967. " }•'P T'i~r( ~L•TL+ S -+>I~~T .t:vS. ' _ ~ ..,,~ ~ '11_.~t" ~~;_ ~ dividee all ' of : (white '`.or40+ at tieu fumes ~~. wome - schoo Q34or Q2: Q Thi Ptrioc numb= smoki' srnok, Wit three discarr least' c of sub adjusi was tF L) to smullI less tt corre~ subgr : . 6-yea factot to risl alive : up to the st Ha subje • Pat of the they t inClUv ~..~x;,r,~+c. S*t4Y337~+ww•.~a at••+~r}-t P'n

'"'rAR" AND NICOTINE CDIYTEXT OF CLGARETI'ES 267 subgroup, Likewise, the sum of the person-years of exposure to risk (during a "corresponding,adjiustrnent factor to obtain the adjusted number of deaths in that less than 1.00! ) The number of deaths inleach subgroup was then multiplied by the - srrtallests bgroup is 1.00; and for each of the otherttivo subgroups iris either 1.00 or ~: L) to obtain three adjustrnent factvrs. (By def nition, the adjtrstmertt factor for the Q3, or Q4 (yes or no); (9) history of heart disease as reported on Q 1 or as reported on Q2, Q3, or Q4 (yes or no). This matching procedure was carried out' separately for men and women in Periods I and 2. For Period 11 the amount of cigarette smoking was taken as the number smoked per day as specifed on Q1. For Period 2 the amount of cigarette smoking was taken as the higher of these two numbers: the number of cigarettes smoked per day as reported on Ql and the number per day as reported on Q4. Wirthin each matched group, as defined above, the subjects were divided into three subgroups according to tar and nicotine (HI, M, or L). The entire group, was discarded if it did, not contain at least one H subject, at least one M subject, and at least one L subject. Otherwise, the subgroup (H, M, or L) with thesrraallest number of subjects was identif ed. This smallest number will hereafter be referred to as the adjtisted'number of subjects in a specifned group. The adjusted'number of subjects was then divided by the number of subjects in each of the three subgroups ('H, M, or fumes, gases, chemicals, X rays, or radioactive materials (yes or no) (men but not women were matched on this factor); (7) education (no high, school! vs some high school or above); (8) history of lung cancer as reportedon Q1 or as reported on Q2, at time of enrollment (urban or rural); (6) history of occupational expoxure to dust. or40+); (4) age began cigarette smoking (<15,15-24, or 25+); (5) place of residence "low" or "'medium'`on the 1'965-1966 questionnaine (Q4), or as "low" on both Q2 and Q4. "Matched groups'"' analysis was utilized. As a first step, the subjects were divided into groups such that within each group the subjects were alike in respect to all of the following factors: (1) age (same 2-year date of birth cohorts); (2) race (white ornonwhirte); (3) number ofcigaretrces smoked per day (1-9;10-19; 20, 21-39 (2)'"Low" was defined as "1ow""'on the 1959-1960 questionnaire (Q,1), and as either naine (Q 1) and as either"'high" or "medium" on the 1'965-1966 questionnaire (Q4). 6-year period) for each s!ubgroup was multiplied by the corresponding adjustment factor to obtain the adj,usted number of persvn-years. The person-years of exposure to risk of a subject during,a 6'-year period is (a) 6 years if the subject is known to be alive at the end of the 6-year period; or (b) elapsed time from the start of the period up-to the tiime of death if the subject diediduring the period; or (a) elapsed time from the start of the period up until the time the subject was last traced. Having carried out the above procedure separately for each matched group of subjects, the findings were summarized. RESULTS. Part 1 ofTable I shows the number ofmale and female subjects at the start of each of the two time periods classified by the tar and nicotine content'of the cigarettes they usually smoked. Part 2 ofTable 11 shows the number of these subjects who were included in matched groups as previously defined. The difference bztween these on

268 HANiDtOiti D FLtL. TABLE' I TOTAL <tiiuM/nER OF S'CBJECT4,ST. SW.RT~~. OF~ EACH1 oF~Ttti'U~~PERi10Ds., NIaIBE~R'~. LN ~ tf3TCHED CROVPs, Axn~. AnjUSTiED~ 1!tC-WBER~~OF PERS@N-YE4,RS'.OF~~ E:CPOwRE TO. RI5K (BY'"TAR" AND NICOTI\E~ CO%TF--,•T'of ~C[cARE'rrE). "1High" '"il3edium"" "Low" Period' tar-nicotine tar-nicotinc tar-nicotine Male Male Female - Female Male Male Female Female Male Male Female Female' 1. Total'numb,er of'subjeet3 at start of period 1 63,063 54,999, 15,360 2 29,157 40,090 6',832 1 44,137 59,750 32,703 2 22,909 49,193 16,803 2. Number of'subjects in matched groups 1 57,346 50;698' 14,897' 2 25,459 35,112 6,5641 1 43,062 58,538' 32,357' 2' 22,153 47;679' 16,550, 3 1 Adjusted person-years of espoiure to risk 1 82,428' 82:898 83,072 2 35',974 36.051 36.435 1 174;619 175,038 175,7414 2 83,639 89,027 89,129 • Period I: July IL 1960-June 3W. 1966; Period'2: July 1, 1966-June'30, 19%'1. 6'Includes cases which wereditticult to classify. two sets of numbers indicates the number of subjects who were excluded because'e some oroups did not contain at least one H, one L'v1', and' one L subject. The adjusted number of subjects was: 14,688 for males, Period 1; 6175 for males, Peri od 2;,30,176 fbrfemaIes, Period 1; and 15',342'forfemalzs, Per'pd 2.Sint:e a large proportion of'the ""Iow'"' TIN subjects were matched, the adjusted number of matched subjects was only slightly less than the number of "low"'T!N'subjects in matched groups. On the basis of the adjus.ted numbers of subjects, the mean ages of the subjects were: 53.6 for males and 51.6 fo'r females at the start of Period 1; and '`'58.4 for males and 56.7 for females at the start of Period 2. Fart 3 ofTable I shows the adjusted number of person-years of exposure 2o risk as previously defined. .-As shown in part 1 of Table 2, the adjusted' nttmber of' deaths was 4735.5 for "high" TIN srnokers; 4299.9 for "mediinm" TIN smokers, and 3991.2 for "low'"' TIN smokers. The difference between! the "high'"' group and the "low" group is statistically signiificanC(P<0.0001). Furthermore, for each offiour sets ofcomipari- sons (male, Period 1; male, Period 2; female, Period' 1; and female, Period 2), the adjusted nutnberofdeathswas~hiphest for the "high'"TlN smoker~s and Io~westforthe "1mw"" TLI^l smokers. In each of these four, sets, the difference between the "high" T/N' and the ""low'" TIN groups is statistically significant ('P<D.01005). As also shotivn in part I ofTable 2, the adj usted number of'lung cancer deat'hs was 31'8! 4 for "high'"' TLN' smokers, 285.5 for "medium" T1iV smokers and 235.2 for "'lbw"'T1N smokers. The difference between the'"higtt'"'T/Nigroup and the'"low" TIN t , COm''f ~ canc( TJTI : Pa1 numl: adjus the r Comb mort€ ; forfe Perio ' 1VSk "met: TIN : Adju: ofd'e: to ris Tal numb 14'S3.- ence. signif and f: Se Male Male Fem, Femal, Tote Male Male Femal, Femat, . Tot:;

4'`TAR" AND NICOTINE CDNTFN'T OF CIGARETTES 269 TIN smokers. Part 2 of Table 2 shows 'mortality ratios calculated by dividing the adjusted number of deaths in the "medium" T/ht'and"low" T/Nigrowp by the corresponding adjusted number of deaths in the "`high'" T/N group. As shownon, the bottom line, the mortality ratio for the "low" TIN group was 0.84 for all causes of de'ath combined and 0.74 for lung cancer deaths. It is of interest thai' the lung cancer mortality ratios of "low" TIN smokers were lower for females than for inales (0.57 for females and 0.83 for males in Period 1 a'nd! 0:62 for females and 0.79'for males in Period 2),. _. Mortality ratios were alsoicalculated by dividing the adjusted death rates of t'he "medium" TIN and'"'low" TlN smokers by the adjusted death rates of the "high" T/PJ smokers. These mortality ratios were very close to, those shown in Table 2. Adjusted death rates, if desired, may be calculated by dividing the'adjusted number, of deaths (part 1 of Table 2) by the corresponding adjiusted'person-years ofexposure to risk (part 3 of'Table 1). Table 3' shows corresponding figures for coronary hearG disease. The adjtastedd number of coronary heart disease deaths was: 1616.8 in "'high" TINI smokers, L483.3' in "medium" TIN smokers, and 1392.7'in "low" N smokers. The differ- ence between the "`higia'"' T/N group and the "low" TIN group is statistically significant (P<0.000'li). For each of the four individual sets of'comparisons (males and females in Periods 1 and 2), the adjusted' number of coronary heart disease TIN groupi5 stat'isticallysignificanr(P<0.ID005)L For each of fourindividual sets of' comparisons (males and females in Periods I and 2), the adjusted number of lung cancer deaths was highest for the "high'"'T1'i`N smokers and lowest for the "'low"" TABt:E' 2 /AD3UETF.D i lIU~7BER OF~ IDEATHS ~I(TOTAL A1tiD~ LUNG~. CANCER) '~ AND ~:YIfDRTALITY~. RAIIOS ~ DUR7\G~EACH OFTWO:PERIODS~OF~TiJtE~ BY~SEX'AND~BY~"TAR " AND~. NICO1rirE CO:YTENT OF~. C(GARETTES USUA'L'LY~ SdtO.KED -Total . Malb 1 Malb 2 Female i Female 2' Total Total deaths "High" ..Medium.. ..Lou,., Lung cancer deaths ••Hish•• T/N T/N 1VN T/N "Mediucti "' "Low'"' T/N T/:t i. Adjusted! numoer of'lenths 1.543.0 1,394! 4 1.351.7 1L:4 1'17.4, 101.01 935,.2' 913:7 759.4 89.6 84.5 70.61 1,253.6' 1„117:P 1.053.9 48.3 41.4' 27.4 1,003.7 874!7 826'.2' 58.1 42.2' 36.2' 4,735.5 4,299.9 3:99L._' 318.4 285.5' 23'5.2'. 2. Mortality ratios 1.00 0.90 0.88' 1L001 0:96' 0.83 1.00 0.98 0.81 1100 0:94 0.79 1.00 0.89 0.8+>'! 1'L00 0.86 0.57 1.00 0.87 0.82' L00' 0:73' 0j62 1.00 0.91 0:84I 1100' 0:90 0:7d.

. HA1'(SDt]NID'ETAiL. Period' Male 1 Male 2 Female 1 Female 2 Total "High" "Medium" • tar-nicotine tar-nicotine ••L.ow ' tar-nicotine ll Adjusted number of CHD deaths 696.5'' 632.5 - 6!45.6'. 336:0 345':6 274:2' ~ 318.7' 277.5 237.4'. 265.6' 228:0 215.5' 1,616.8 1,483.3, 1,39'2.7' 2. Mortality ratios 1.00 0.91 0.93 1.00 1603 0.82 1.00 0.87' 0.81 1.00 0.86 0.311, 1.00 0.92 0~86 deaths was higher for the "'hight' TIN group than, for the '"low" TIN group: J' As shown on the bottom line ofTable 3, zhe coronary heart disease mortality ratio for "low" TJN s(nokers was 0.86. ADDITIONAL ANALYSES Coronary Heart Disease As a further test of the hypothesis as related to coronary heart disease, a second Inatched~groups analysis was carried out including,additional factors of significance or possible significance in relation to that disease. In this analysis, the subjects within each matched group were alilCe with respect to age (same 5-year date of birth cohort, with respect to alI'the other'eight factors included in the first analysis (as previously described), andlwitfi respect to all of ttie folowiiog factors:(l'0) history of stroke (yes or no); (1I) history of diabetes (yes or no); (12) history of highbliood pressure (yes or no); and (as reported on Q1l~~('13)usualarnount ofex~ercise~(idone, slight vs moderate, heavy); (14) obesity (?0~'0 or more over average weight, yes or no); (15) coffee-teal (six + cups a day, yes or no); (1I6)' whiskey, gin, etc. (none,, occasional vs dailydrindcing); (1'7) aspirin(oftzn vs not often); and (18) occupation (doctor, lawyer, teacher, nurse andi other professionals vs other). Because of the larger number of factors on which'the subjects were matched,, there were fewer matched groups containing at least one H, at least one M', and att least one L subject. The ad'justed numbers of subjects were: 11,599 men, Period 1; 4996 men, Period 2; 23,584 women, Period 1; and 11,450 women, Period 2. Since these adjusted' numbers of subjects were less than the corresponding adjusted numbers of'subjects in'the first analysis (as previously described) there were fewer TABLE 3 AD)uSTED'NU:IIBER OF DEAT'HS'(COROTA'RY HEART'DtSEASE) AYD'.%YCDRTALITY Ft.ifflOS ~ ~ DCRoC~EACH OF~Two ~PER'tODS OF T(SYE.BY~S~EX~~AX~D~~By~**T'11~" AYD~ 1HtCOTtxE' C.OYTE.VT~OF, CItwIMETriE3 U'SUALLX~ S?tOKED~. ~ adjust result! . The was: 1 for "k group compz naryh 8roup. 0;83. Select III c distritt than o couldd ofsub Pe rioc there4 therez very c up sm Many As cance `''1o„r• nttttcf: vari© t' studie cigare subje, or hig cigare tions cigare able r subje carrie were factoi befor time. Perio Th

' "'TaR~' AIID NIICDTtNE CONTENT OF CIG'.iRETPES ,271 adjitsteddeathsd Nevertheless, the results of this second analysis were close tothe results of the frst analysis. :-Tho ; adjusted number of coronary, heart disease deaths (in this second analysis). was: 1007:5for'"high" T/NI smokers, 92'9.31for"rnediiam'"T/N smokers, and 834.5' for "'low" T/N 'smokers: The difference between the ""higtt'" group and the ""ibw'"' group is statistically significant (P<pi001). For each of the four individual'sets of' comparisons (males and females in Periods 1 and 2) the adjusted number of'coro- -nary heart disease deaths was higher in the"high" TINgroup than inthe "low" TIN 'group. The coronary heart disease mortality ratio! for the "low" TlN group was 0.83. Se lecttve Ejjj "ect of Giving Up Snlnking, In comparing smoking habits as reported on Qi', Q2, Q34 Q4, and Q5 (which was distributed in 1972), we found that a larger proportion of the "'Iow'"'TJN smokers than of'the'"high" TIN smokers gave up the habit at a~laterdate. Conceivably, this could have accounted forthe difference between the death rates ofxhese two groups of subjects.To check this possibility, wemade anothermatched'-groups analysis for Period 1. In this analysis, subjects who were not smoking as reported on Q2 weree thereafter excl>aded; Likewise, those who were not smoking as reported oniQ3 were thereafter excluded. The results of this analysis (in terms of tnortal'ity ratios) were very cliose to the res uits shown in Tables 2 and 3: Fromithis, we conclbde that giving up, smoking did'not account for the findings as shown on those two tables. Many "Low" TIN Cigarettes versus Fewer "High" TIN Cigarettes As shown.in Tables 2 and 3, the adjusted numbers of deaths (total deaths, lung cancer deaths, and coronary heart disease deaths) were consistently lower among -"low" T1N smokers than among "'high"' TIiV smokers when the subjects were matched on number of cigrzrettes smoked per day (as well as being matchedi on various other factors). There is abundant evidence fromithis study and, many other studies (U.S. Ptbblic$ealth, Service; 1971) that death rates increase with:number of cigarettes smoked' per day. For this reason, we wished to determine whether sinbjects.who smoked a relati'vely large number of "'low" T/v cigarettes had as higlnh : or higher death rates than persons who smoked a, smaller number of " "high" TIN cigarettes. Because of linlitations in the number of subjects with varipuscombina- tions of number of cigarettes smoked per day and tar and nicotine content of thee cigarettes smoked„ we were unable to make fine distinctions. However, we weree able to compare subjects who smoked, 1 to 19 "high"" T/h( cigarettes a day with subjects who smoked 20 to 39 '°low'"'Tml cigarettes a day. For this purpose, we carried out atnatched groups analysis such that the subjects in each matched group were alike with respect to age (same 5-year date of birth cohor2) ~and all of the other factors used in the first analysis exceprfor number of cigarettes smoked per day. As before, this was done separately for men and women inieach of the two periods of' time. The adjusted number of subjects was: 7971 males, Period' 1; 2785 males, Periqd 2; 12,275 females,, Period ll; and 484'1 females, Period 2. The adjusted number of deaths (all causes of death combined) was U826'.3' for

aais;r_~, . r i r :.t . 272 HAMMOND ET.#G. subjects smoking 1V to 19 "high" TIN cigarettes a day and 1923.9 for subjects smoking 20 to 39 "low" TJN cigarettes a day (mortality ratio 11.05), This difference ,; isnotstatistiaallysignificant.Theadjusted'inutnbero'fcoronaryheartdiseasedeaths was: 670.6 for subjects smoking I to 1'9 "`high'"'T/N cigarettes a day and 736.6 for subjects smoking,20 to 39 "1ow""T ~lY cigarettes a day(mortadity ratio I.10). This difference is statistically significant (P<0:05). The adjusted' number of lung cancer deaths was 75:8'for subjects smoking l to 19'"higb'"'TLN cigarettes a day and 129:51 for subjects smoking 20 to 39 "lnw'"'TIN cigarettes aiday(rnortal'ityratRo 1.71). This difference is statisticall signifucant(P<0.000t). For each of the four individual sets of comparisons (males and females in Periods I and 2), the adjusted number of lung cancerdeaths'waslower in subjects who smoked, 11 to 119 "high" TlY cigarettes a day than in subjects who smoked 20 to 39 "low"' T/N cigarettes a day. "Low" TlaV Smokers versus Nonsmokers ; The next question which we posed was whether the death rates of subjects who. smoked "low'"' T~l~i cigarettes were appreciably different from the death rates of'f subjects who had never smoked regularly. To this end, we undertook a, match'edd groups analysis matching,on age (2-year date of birth cohort) and all of the factors included in the first analysis except number of cigarettes smoked per day and age subjects began smoking. Since subjects who hadi never smoked greatly outnum- bered "low" TIN subjects, all but a very few of the latter subjects were matched. ,The adjusted numbers of subjects were: 15,346 men, Period, 1; 6822 men, Period'2; ;. 32,702 women, Period 1; and 16,803 women4 Period 2. On the basis of the adjusted' ' numbers of subjects, the mean ages were 53.81for men and152:3 fo'rwomen at start of Period 1 and 58.7'for men and 57.3 for women at start of Period 2: The results are shown in Table 4. The adjusted number of deaths (all causes of ti-.`death combined) was 4670.3 for'low" TIN smokers and 3099.0 for subjects who never smoked regularly. This difference is statistically signiificant'(P<0:00'0i1). The adjusted number of coronary heart disease deaths was 1I674'.3' for "low" T ~l .~, smokers and 1008.3 for subjects who never smoked regularly: This difference is statistically significant (P<0.000i1): The adjusted'number of lung cancer deaths was 25'ff:0 for low T1N smokers and 39.4 for subjects who never smoked regularly. This difference is statistically significant (P<0.0i001). For each of the fmurindividual sets of comparisons (males and females in Periods l and 2) for all causes ofdeath combined, for coronary heart disease deaths and'for lung cancer deaths; the adjusted number of deaths was higher for "low" TlN smokers than for subjects who never smoked regularly and each of these differ- ences is statistically significant. (The value of"P'"ranged from <0.01 to <0:0001). As shown on the bottom line of'1'abie 4, the mortality ratio (adjusted number of deaths of subjects who never smoked regularly divided by adjusted nurnber of deaths of "low" TIh1i smokers) was: 0.66 for'all'causes'of dea'th combined, 0.60 for coronary heart disease deaths, and 0.15 for lung cancer deaths. s..^.. Male =' D+tsle Femal Fema1, Tot_ . bfate 3tale Feasal, ... Fetnat Tot: . Nt It smok anal} Cir, (folio 1952; majo befor smo, "meL rettes incur incre to sa step quit W Tr, man• heal; cigar

6, TAR'" AND NICOTINE Cb:YTENT OF CICAR'F:T'rPS 273 TABLE 4 MO~RTALITY~R.CTIOS DtiRt\IG~E.ICH OF~Two~PERiODS~OF~TIl1E~~ B7~SEX', -Lour- T/:I~~ ~ SltO1:ERS; AN~D~ SR'.BJEICTS wwo~. NEVER S?lOKED ~ REGULA'RLY ° ApJUSTED~. NU>tBER~~ OF' DEATHS~(TOTAL. CORONARY HEART DISE.ALSE, AND, ~. LUNG CANCER):.a\~D~~ Total deaths Coronary heart disease I ung cancer Male . _: 1 l1522.3 Male 2 - 853.0 Female 1 11288.0. Female 2 1,007.0 Total' 4,670.3 !! I100' 2 1L00' 1 i 1:00 2 1.00 1.00 Never smokedp ••'Low" T f Y Never smoked* "Low° T1N Never smoked• I. Adjusted number of deaths 86,u 8 742.3 399.11 107.0' 9.8' 542.11 3I111i.0 238:4 77,0 7.1 979.0 343.0' 205:6 30.0 12.8' 713:1' 273.0 165:2 44.0 9.7' 3,099.0 1,674.3 1.008;31 258.0 39.4 2. Mortality ratios 0:57 1:00 0!54' 1.00 0.09 0i64! 1.00 0:77' 1.00 0.09 0:76I 1.00 0.60 1.00 0.43 0.71 1.00 0.59 1.00 0.22 0:66I 1.00 0.60 1.00 0.15 • llever smoked cigarettes, pipes, or cigars regularll+. CONCLUSIONS ' It is quite apparent that reduction in the tar and nicotine content of cigarette. smoke did not make cigarette smoking "safe" for the men and women in this analysis, all of whom were over the age of'4I0 in 1959. . .- Cigaret'tes with reduced tar and nicotine were not introduced until the mid 1950s . `((following the retrospective studies of Wynder and Graham, 1950; Dolll and Hill, 19521; Levin er at., 1950). Almost alI' of'the male cigarette smokers and the great majority of the female cigarette smokers in our study be;an'smo'king cigarettes long . before that date. Therefore, the subjects here classified as "low" TINI cigarette smvrers, were, with few zx'ceptions, persons who smoked ""high" T.';?i or "medium" T/N cigarettes for many years and then switched to "low" TIN ciga- rettes. It' appears that by so doing they somewhat reduced the serious risksthey int:urred by smoking. (This does not'apply to the relatively few who aCthe same time increased the number of cigarettes they srnokediper d'ay.) Therefore, we think it fair to say, thatsw'itching from "high" T7N to "'low" TIN cigarettes was at least a srnau' step in the right direction for those who continued to smoke cigarettes. Those whoo quit smoking fared considerably better. W'hat' of youths who have not yet taken up the habit of cigarette srnoking?' They would be welI'advised never tio do so. However, in spite of all the warnings, many thousands ofyoung people do in facttake up tihe habit. The threatito'thz future health of those who make this youthful decision would be reduced' if "high" T/N ' cigarettes were remov'ed from the rnarket. Manufacturers may be willing,to'db so

E.V vitRO?. HA41MONID ETAL voluntarily in the light of'the fact that long-term trends have been in this di'rectimn.. We will end with a word of caution, ' In producing,eigarettes with extremely little tar and nicotine, some manufactur- ers may use additives for flavor or for some other purpose. In our opinion~ both additives and the cigarette smoke condensate (tar) should be tested' for car- cinogenicity before such cigarettes ~ are put on the market. Brossi L D. J:, and Gibsonl R. (1968). Riaks of lung cancer in smokers who switch to filter cigarettes. Amer. J. Pub. Health 58, 1096. Doll. R., and Hill. A. B. (1952). A study of the aetiology of carcinoma of the lung. Brrt. Med. J. -2 127t. Hammond, E. C. (1964). Smoking inxelationito mortality and morbidity. Findings in first thirty-four, months of follow-up in a prospective study started' in 1959. J. Nat. Cancer lnst. 32. 1!161.. Hammond. E. C:, and Garfnkel L. (196a). Changes in cigarette smoking.J: Nat. CancerInst. 33, 49. Levin, M. L.. Goldstein. H.. and Gerhardt, P. R. (1950). Cancer and tobacco smoking. A preliminary report. J. Amer. 1+lyd. Assoc: 143, 336. U. S. Public Health Service (1964). "Smoking and Health Report of the Advisory Committee to the Surgeon-General of the Public Health Service:" Public Health Service Publication No. 1103. U. S. Dept. of HL E. W., Washington. D.C. U. S. Public Health Service ('196,3). Public health service technical report on "tar" and nicotine. In "'Hearings before the Consumer Subaommitterof the Committee on Commerce," United States Senate.,August 23-25. 1967; p. 7. U.S. Government PtintingAffice. Washington. D.C. U.S. Public Health Service ('197'1): "The Health Consequences of'Smoking. A Report of the Surgeon- Gtneral," DHEW Publication No. 714513. U.S. Dept. of H.E.W. Washington, D.C. U.S. Public Health Service (1976). "Adult Use ofTobacco-1975." U.S. Dept. of H.E.W.. Bethesda, Md. Wynder, E. L., and Graham. E. A: (1950). Tobacco smoking as a possible etiologic factor in bron- chogenic carcinoma, 1. Amer. Med. Assoc. 143, 329. Wynder, E. L.. Mabuchi. K:, and Beattie. E. J.,,Jr. (1970). The epidemiology of lung cancer. J. Amer. AYed: Assoc. 213, 2221. . REFERENCES M H m. . In ,'. Iran. pests about. Garg: )mver. detec the f' po'llu Sana~: Tt: provi samF Pahic after tion Prep, Th meat resu, cont. pow soxli Acet . Ti in a of'p: