Tobacco Documents Online

~ ung Cancer Mortality A ~ising, J James E. Enstrom, Ph.D.2 ABSTRACT-On the basis of data availaW from t fat,ve samples of lung cancer deaths In the United S cs national mortality statistics and other epidernioloft*1111E'dies, the lung cancer monality rate has risen substanlially between 1914 and 1968 among pern ti who never smoked cigarettes. For wnite males the relative In.•.:as• for ages 3S-tl4 years has been a5out 15-told; the relative ncrease for ages 65-84 years has been :t~out 30-fold. For white lemates the relative increase for ages 35-84 years has been about sevenfold. Most of the relative In- crease occurred before 1935 and was probably due to changes In d:agnostic criteria. However, Increases have continued up to the present for male nonsmokers, who now apparently have an an- nual age-adlusted lung cancer death rate of about 25 per 100,000 persons between the ages 35-84 years. The rising lung cancei nte amony nonsmokers Indicates that factors In'sddilron to per- sonai ci9arett**-smoktng tiav`e had a signiftcant effect on the mor- tatlty rate from this dfsesse. In spite of the limited quality of these data, they suggest that a more complete understanding of lung cancer etiology Is needed.-J Natl Cancer Inst 62: 755-760, 1979. >fuch effort over the past 30 years has gone into the collection of epidemiologic evidence establishing and confirming a statistical relationship between lung can- cer and cigarette smoking. To date the c.idence appears so o•er..helming that there is general acceptance of the conclusion in the 1964 Surgeon General's report: "Cig- arette smoking is causally related to lung cancer in men; the magnitude of the effect of cigarette smoking far outweighs all other factors. The data for women, although less extensive, point in the same direction.•" (1) Nevertheless: the-,~cltuion•of catuality~s=still sub- j ected to ~ti ~sm-(2): ~Iisdeed.-iio-m ec3zariisfa for h turnan tobaao ca?cinogenesi`s has,xet-been-successftiIl~?6iiriu; lated an-d ie-sted-and no randomized'ti'urrian iRal-has ev er been, conducted _ to . measure° che-speafu ffect: of either theinitiation 'or' the'"cessatiori -oE mo'Vng''on subsequent lung cancer mortalit~'` A1so, many factors other than personal cigarette smoking have not been examined in gTeat detail in their relationship to lung cancer. In view of this, it is useful to examine lung cancer :nortality among nonsmokers who are subject only to factors other than their.own cigarette smok:ng l3, ~). «'ith this brief introduction, an analysis will be made of existing data on lung cancer mortality trends among nonsmokers as well as among the total popu- lation in the United States. One probable reason this analysis has not been done previously is that data on nonsmokers are limited. Because of the low lung cancer rates of nonsmokers, large groups of nonsmokers must be studied for long periods before sufficient deaths from lung cancer occur. Fortunately, however, two large representative surveys of the smoking history of Arnericans who died of lung cancer are available. These surveys and concurrent surveys of the smoking %:Of.. 62. SO. 4. APRIL 1979 Nonsmokers t 0 v l T0615b2 s of living Amcricans ,lllow the calculation of esti- mated age•specific mortality rates for nonsmokers in the entire U.S. population. These rates can then be compared with early data on U.S. lung cancer mortality when cigarette smoking was a rclativcly rare habit and with other lung cancer data from epidemiologic studies of nonsmokers. In this paper. nonsmokers are dcfincd as persons who have never smoked cigarettes. Although nonsmokers here do not include fonner cigarette smok- ers, they may include present or former cigzir andior ripe smokers. MATERIALS AND ME THODS The earliest available national lung cancer data are for 1914, based on a death registration system composed of 24 States mvith 62 e of the total U.S. population (1, 5, 6). Only 185 white male and 143 white femaie lung mn- ccr deaths were recorded in 1914, and problcros with the accuracy and representativeness of these data ha.e been discussed in detail elsewhere (6-8). For instance, in recent years these 24 registration Statcs have had signifi- cantly higher lung cancer mortality rates than the other States (7). More serious are the difficulties that existcd in :914 of correctly diagnosing and classifying lung cnncer 8). as discussed later. Beginning in 1933 the dc:3,h rcgis- :ration sl-stem included all States and 100' of the U.S. population, and age-specific death rates ha%e since been available on an annual basis (9, 10). t\'ationat .4lortality Survey, 1938-1959.-The Current Mortality sample, a representative 10% sample of all deaths in the United States, provided the records of lung cancer deaths for 2.381 white males 35 years old and over during 1958 (11) and 749 white females 35 years o;d and over during 1958 and 19359 (12). A questionnaire on the residence and smoking history of each decedent «as sent to the family informant listed on the death certifi- cate. and the overall response rate was 90%. The Bureau of the Census collected comparable information on the 'r' S population as a supplement to the Current Popu- ation Suney for Jfay 1958. From a representati%e na- ,ional sample of approximately 35,000 households, his- :orles were obtained from 31,516 %+-hite males and 34,339 N~htte females at least 35 years of age (11, 12). N'auona! Mortality Survey, 1966-68.-This survey ,..as conducted in a manner similar to that for the ;958-59 survey. Information on deceased persons 35-84 Nears old in the United States in 1966-68 comes from 1a11IRV1ATION l1StD: At:S=Amerion Cancer Society. Received May 19, 1978; accepted September 3, 1978l r School of Public Health, University of Glifornia, Los Angeles. Calif. 900211. 755 -;..n J NATL CA.ICER INST T.(NY 0005803

75b Ensirom the National Mortality Survey, a follow-back survey of a representative sample of 19,526 death registration records, including 1,464 white males and 319 White female lung cancer deaths.3 By mail questionnaire, sur- sitiing family members and others named on the dc;ith ccrtificate provided smoking histories and social char- dcteristics of the deceased person; the overall response rate was 92%, Then, through the Current Population Sur.cy conducted by the Bureau of the Census in Au- gust 1967, smol:ing and socioeconomic information comparable to that for the decedents was obtained for a rcprescntative national sample of 61,000 adulu 35-84 ~ears old, including 25,266 white males and 29,308 white females. (13, 14),' Other epidemiologic studies of nonsmokers.-The Dorn U.S. ~cterans study prospcctively followed a co- hort of 248,195 U.S. veterans who held actiIe U S. Government life insurance policies in 1953 (15). NSost of the enrollees were healthy whiie male veterans of World War 1; 82q were white-collar or skilled workers. Mortality data are available on 54,344 men 35-84 years old who never smoked regularly and who were followed from July 1, 1954, to December 31, 1962 (15). The ACS "cancer prevention study" prospectively followed a cohort of 440,558 men and 562,671 women who were enrolled between October 1959 and February 1960 by 68,116 ACS %olunteer workers (16). The study area included 1,121 counties of all sizes and types in 25 States, and the enrollees were essentially all %, hite, not seriously ill, and generally above average in sociocco- nomic status. Mortality data are available on 95,849 males 35-84 }ears old who never smoked regularly and %.ho were followed from date of enrollment to Septem- ber 30, 1963. Age-specific death rates have been calcu- lated from the deaths and person-years accumulated after an average 46 months of follow-up (16). The pub- lished data are given according to age at start of study and have been modified with a life table correction to make them comparable With data given according to attained age at death.' Lung cancer rates are also pub- lished for this same cohort followed through June 30, 1972 (17), but these rates are not available in sufficient detail to warrant presentation here. Inasmuch as no detailed data are available for na- tional mortaliry surveys or large-scale epidemiologic studies since 1968, one source for more recent lung can- cer mortality rates among %.•hite male nonsmokers :s data on active Mormons in California and Utah. Active Mormons here are the Church leaders known as High Priests and Seventies. These men can be characterized as lifetime or long-term Mormons who adhere strict:y to the "Word of Wisdom," a Church doctrine forbid- ding the use of all forms of tobacco, alcohol, coffee, and tea and recommending dietary moderation (18). s 1966-R13 National Mortality Survew. National Center for Hea!ch Statistics. Roct;ville, Md. Unpublished iechnical notes and compu:er iapes• 1916. ' Codley FH: Cirarette Smoking, Social Factors, and )forulity: Ner, Esrim.tes (rom Represenuuve National Samples. L'npublished hesis, t;ni.rr,ity of S1anland. College Par4. Sid., 1974. J ', At1 CGNCtiR tNST TOE<<~3.. TThey can be effectively considcrcd to be a cohort of males who never smoked; this assuinption h;,s been tentati~ely confirmed by actual suncy data on a sinall sample of High Priests and Scventics (19). Cs:ng .'•for- mon Church records, we obtained data on both ~ne deaths and the population at risk, as described clsc%, here (18). The lung cancer mortality rates ha%e been calcu- Iated for an average 13,300 California High Priesis and Seventies during 1968-75 and 50,000 Utah High Pricss and Seventies during 1970 and 1975. A total of 63 cancer deaths ha,.r occurred among 2C8.000 person••~ars of mortality experience for ages 35 to 84 ,cars. T!-,ese data are detailed elsewhcre [(19); Enstrom JE: t'npub- iished data]. Active Slormons-are, nor a-.representacve*-_ sample of all U.S. whiie_malSs:~'hofiavenc•.er srnoi:ed;' but tFtey are:stmilar in_total mortalicy to the LrS:.lete an =~ and-ACS cbhorts of nonsmokers; as ',vtll be-discr.iss6d; late` r-_'' RESULTS Lung Cancer Mortality Rates The age•specific U.S. lung cancer morca}ity rates ror nonsmokers and the total population are shown in tabie I for white males and in table 2 for white fcrna':es. Table 3 shows lung cancer mortality rates from U S. cpidemiologic studies for %:hite male nonsmokers •, ~o are not necessarily represcntative of the nation bu: •.o ho provide additional data on nonsmokers. The rates are given in 10•%-ear intci-vals from ages 35 to 84 %c;trs. ••.ith an overall age-adjusted rate standardized to the '.:50 U.S. population. Se%eral comments below etipla:n ;;-e data. The 1914 age-specific rates are for all U.S. %:hites in the death registration area. Inasmuch as no bre:>,::l-?own is available by smoking status, the assumption •.:i1l be made that in 1914 nonsmokers had the same rates as the total population. This probably overestimates the true nonsmoker rates, but is reasonable on the basis that cigarette smoking was a relalively rare and minor habit in 1914; only a small percentage of the men and essentially no women had e%er smoked cigarettes, and the annual cigarette consumption N%as 300 per person at least 15 Nears of age (20, 21). Bv ccmparison, :n 19» about i0A of the men and 40 a of the %.omen had e.er smoked cigarettes (20), and the annual cigarette con- sumption was 3,500 per person at least 15 %ears of age (21). Use of cigars, pipes, and other tobacco products has declined greatly since 1914 (21) and hence has no major influence on recent lung cancer rates compared with the influence of cigarette smoking. Rapid changes in nonsmoker rates after 1914 can be seen by a comparison of the 1914 and 1935 age•specific rates for all U.S. whites. In 1935, males 65 years old and above and females 45 years old and above can ef- [ectively be considered to have never smoked cigarettes on the basis of the 1955 survey (20). The relati~e in- creases between 1914 and 1935 are appro\imatel}• a fac- tor of 10 for ages 65 years and above. Age-specific lung cancer rates are published for 1958 VOL. 6^-. \n. 1. AI'x:L t9"9 TINY 0005804

Rising Lung Cancer Mcrt=llty Ar,tong hcnsmakers 75 T~eLE 1.-.4nnual aye•spcctfu funq cancer mortality rates Jvr US. uAitt mafes in 1911. 1915. 1958, nand l96G-6d accordinq to smoking status (dcaths/100.000 perscnal Never smoked cigarett,es 'ever smoked. Never smoked cigarettes Total U S. population ~ Age. Yr ' 1958, ' ' ' ' ~ ... 1914' 1935 )958 1966-68 :914 1935 1958 :966-68 ~ 35-41 0.5 (20) - 1.8 (3) 1.8 (3) 2.3 (2) 0.5 (20) 4.2 9.0 12.9 45-54 1 4 (43) - 3.4 (5) 3 4 (5) 3 5 (3) 1.4 (43) 13 2 46 5 58 6 ~.7 55-64 . 3.0 (55) - 12.7 (15) 148 (15) 32.2 (24) 30 (55) 22.0 . 131.4 . 170.7 z 65-"4 2.6 (26) 26.7 25.0 (25) 1 337 (25) 656 (43) 26 (26) 26.7 185.2 297.0 ~ '~-34 1.2 (4) 23.3 55.0• (32) 69.7 (32V 899 (36) 1.2 (4) 23.3 147 9 273.1 35-84' 1.6 (148) - 108 (80) 133 (80) 22.8 (108) 1.6 (148) 14.6 776 ::33 s.eferences .. ~ . (1. 5) (9) (1Y) (1J): see (1. 5) (9) (10) 10) xxt foot- notes 3 and 4 ' Based on :914 U.S, death registration area (5); rates for males who never smoked cigarettcs 'nave been assumed to be the same as ^.ose for all males (sce text). Numbers in parcntheses are the Noo of deaths on µhich rate is based. ' For ages 65 and above, rates for males who never smoked cigarettes have been assumed t.o be :he same as :hose .`or ail ma'.es (~ee text). ~`ever smoked' is defined as "never smoked as many as 5-10 packs of cigarettes. 50-75 ciLars, or 3-5 packapes cf ;,;;.e 'wbacco durir,g entire life' (11). Numbers in parenthcses are the No. of deaths on which rate is based. ' Each age•specific "never smoked cigarettes" rate is obtained by multlplying 1958'never smoked" rate by the ratio of "nc:er srr.a-d cizarettes' rate to "never smoked' rate given in table 3!or 1954-63 pooied cohort. .\'umbers in parent.keses are the \o. of dcatas on which rate is based. '"Never smoked cigarettes' is defined as 'never smoked as many as 5 packs of cigarettes in entire life' (~ee text footnote 3). NumLer.r in parenthr_~cs are the No. of deaths on which rate is based. These 1958 data are for ages 75 yr and over (see text). 'Age•adjusted by the direct method to the 1960 U.S. population 35-84 yr old. T.+BLE 2.-.4nnaal age•spenJu lunp carcer mwrtality rates Jor US. uhite females in J?li. 1935, 1958-59. and 1960-68 accorcinp to smoking ;tatus (death.s; 100.000 pcrsons) Never smoked Total U.S. population Age• yr 1914' 193! ' 1958-59' :966-68' 1914' 1935' 1958-59 :966-68 35-44 05 (19) - 0.7 (7) 05 (1) 0.5 (19) 2.0 (155) 2.8 49 45-34 1.2 (31) 5.0 (315) 3.8 (42) 2.7 (6) 1.2 (31) 50 (315) 8.1 158 55-64 2.2 (41) 98 (419) 10.4 (101) 11.4 (24) 2.2 (41) 3.8 (419) 14.6 268 65-74 2.2 (25) 14.5 (372) 21.0 (152) 19.6 (37) 22 (25) 145(372) 24.5 360 75-84 1.5 (8) 14.5 (136) 34.0 (154)' 38.9 (55) 1.5 (8) 145 (136) 35.9 443 35-34~ 1.3 (124) - 8.3 (456) 8.3 (':,3) 1.3 (124) 6.9 (1,397) 11.7 19 1 References ..... (1. 5) (9) (12) (13); see text (1. 5) (9) (10) (10) footnotn 3 and 4 ' Based on 1914 U.S. death registration area (5); rates for females who ne~er smoked have been assumed to be the same as those for all females (see text). .Vumbers in parent/uses are the No. of deaths on which rate is based. ' For ages 45 and above• rates for females who never smoked have been assumed to be the same as those for all females (see text). .vu,nbers in varentheses are the No. of deaths on which rate is based. '"Never smoked' is defined as "never smoked as many as 5-10 packs of cigarettes during entire life' (12). Nuinbers in puren;hrses are the No. of deaths an which rate is based. '"~ever smoked' equivalent to 'never smoked cigarettes.- •x'nich is defined as "ne\er smoked as many as 5 packs of ci;aret;es ln entire life' (see text). Sumbers in parentheses are the No. cf deaths on which rate is based. These 1958-59 data are for ages 75 yr end over (see text). ~ Age-adjusted by the direet method to the 1960 U.S. population 35-84 yr old. U.S. white males and 1958-59 U.S. white females «ho "never smoked" (11, 12). 'ne age-specific rates for ages 75-84 %ears were assumed to be the same as the pub- lished rates for ages 75 years and above, inasmuch as 85% of the population over 75 years is between ages 75-84 years; the error in this assumption is less than 5% (9, 10). Age•specific lung cancer mortality rates for 1966-68 U.S. N.hite males and white females who never smoked cigarettes were calculated from unpublished national survey data. We obtained age-specific rates by dividing the weighted sample of lung cancer deaths among v:hite nonsmokers and the total white population, as procured from the 1966-68 National Mortajity Surney', by the appropriate number of U.S white nonsmokers and total white population as determined from 1967 Current Population Survey of cigarette smoking habits (13).' Finally, each nonsmoker rate \.•as adjusted slightly when we multiplied the ratio of nonsmoker rate to total population rate as determined from the survey by the actual corresponding 1966-68 U.S. ~ital statistics rate: a somewhat simtlar adjustment procedure was used on the 1958-59 suney data (12). Lung cnnccr deaths in both the 1958-59 a,td 1966-68 surveys are dcfined as VOL 62. NO. 4. ArRIL 1979 J NATL UNt:ER tNST TINY 0005805

7S8 Enstrom TCE155 i * TAsLE 3.-AnnYal oQe•sycciJie 11unQ caneer mortality rntes and to/.7l mortali./y ratrs arnunQ .elected nant'mvl:inQ u hite malt ).npulrttivu in tAe United Slales (d4oths/10o.000 persons) AQe, yr cigarettes y ulari ttes re g m ked ci ar tt Q re s o e e i ga s lr, c on Lung cancer death rates 35-44 45-54 0 (0) 0 1(0) 5 (0) (1) 4 (2) 5 (9) 3 5 (2) (13) 2 (2) 5 (9) 2 (2) S(14) 0 (0) 10 (6) 5Z-64 65-74 ;5-84 35-84' )t .! 50 9. (25) 12 (49) 38 (4) 60 4 (78) 12.7 (49) (97) (9) (156) 15 (19) 15 (10) 44 (9) 10.4 (49) 18 29 56 13.4 (35) (34) (20) (104) 12 (44) 26 (59) 45 (13) 10.8 (127) 14 (84) 35 (131) 57 (29) 13.1 (260) 28 (14) 54 (19) 145 (2/) 24.5 (63) Toul death rates 35-54' 1.0=8 (6.932) 1.064 (11.727) 1.125 ( 5.184) 1.163 (8,797) 1.085 (12.116) 1.120 (20,524) 1,J29 (2.g29) Ratio of lung 0. 91 1.19 0 92 1.15 1.00 1.17 2.38 cancer to total death rate. % References........ ((1S), appendix table A, [(16), tables 19, 22, appendix (19) pp 30, 58, 881 table 2a; scc text footnote 4, table A-1)) •"Never smoked ciZarettes- is defined as the combination of "never smoked or occasional only- and "current smokers :nd ex•srnokers of pipes and/or cigars only" (15). Numbers in parentheses are the No. of deaths on which rate is based. - '"Never smoked cigarettes" is defined as the combination of "never smoked regularly." %ioe only.-"cigar only,' and "pipe and cigar- (16); published data. given according to age at start of study, have been modified with a life table correction (see text footnote 4) to give them according to age at death like results from the other studies in this table. Numbers in parenthues are the :v'o. of deaths on which rate is based. ` Pooled cohort is defined by a combin:~tion of the deaths and person-years of observation in the U.S. %eteran and ACS c-~orts: this cohort cor.sisu almost entirely of white males. Numbers in parcntheses are the No. of deaths on \khich rate is based. ' Active Mormons are defined to be High Priests and Seventies in California during 1968-75 and in Utah during 1970 and 1975: this cohort can be considered to consist almost entirely of white males who have never smoked (.ece text). No»n1.en in parrnthc..cs are the No. of deaths on which rate is based. ' Are-adjusted by the direct method to the 1960 U S. population 35-84 yr old. those with International Classification of Disease num- bers 162 and 163 (11).s Lung cancer deaths for the earlier years were defined by the international classifi- cation in, effect in 1914 and 1935 (S, 9). Because the 1966-68 mortality survey did not have a "never smoked", classification, lung cancer mortality rates are pi•esented for males who "never smoked ciga- rettes." For females, there is no difference between "ne%er smoked" and "never smoked cigarettes," because the use of cigars and pipes by females is negligible. To estimate the effect of the male smoking classification differences, data from the U.S. veteran and ACS pros- pecti.e cohort studies are presented in table 3. The t,'.S. veceran lung cancer death rates are given for the two categories "never smoked or occasionally only" and "never smoked cigarettes," the latter being defined as the combination of "never smoked or occasionally only" and "current smokers and ex-smokers of pipes and/or cigars only" (1S). The ACS lung cancer death rates are given for the two categories "never smoked regularly" and "never smoked cigarettes," the latter being defined as the combination of "never smoked regularly," "pipe only," "cigar only," and "pipe and cigar" (16). The deaths and person-years at risk from the two cohoru are pooled and presented for the cate- gories "never stnoked" and "never smoked cigarettes." The age-adjusted rate for "never smoked cigarettes" of 13.1 per 100,000 persons is about 20% higher tJtan the J NATi. C.%.';GER INST 1954-62, U.S. veteran cohort 1960-63. ACS cohort 1954-63. pooled cohort \ever smoked 1963-75 active or aeuional Never smoked Never smoked Never smoked Ne~•er ,:ever smoked ~lormons' ' ' ' ' ' rate for "never smoked" of 10.8 per 100,000 persons. Note that the "never smoked" rates for the 1954-63 pooled cohort and for 1958 U.S. white males are iden- tical. Also, the "never smoked" rates for the pooled cohort in table 3 agree with a previous pooling calcu- lation (22). With the use of the ratio of cohort rates, the rate for 1958 U.S. white males t.,ho "never smoked cigarettes" is estimated to be 13.3 per 100,000 persons as explained in table 1. Age-specific lung cancer mortality rates for active Mormons from California and Utah from 1968 to 1975 are also given in table 3. These results are included for comparison «•ith the U.S. veteran and ACS cohorts in lieu of recent national mortality data on nonsmokers. Active Morrnons and the two nonsmoking cohorts have nearly identical total deach races, t.hich suggests that these groups are similar in overall health. Hov.'ever. the active Mormons have a lung cancer death rate of 25 per 100,000 persons, which is about double the cohort rates. The ratio of lung cancer to total death rate is shown in table 3 to facilitate comparison of the various cohores. Table I indicates that between 1914 and 1966-68 the lung cancer death rate for white males who "never smoked cigarettes" increased by about 30-fold for ages 65-84 years and by about 15-fold for ages 35-84 years. Most of the relative increase occurred before 1935, but there appears to be a continued increase up to the pre- VOL 62. NO. 4. At'RIL t979 TINY 0005806

Rising Lung Cancer Mortality Among Nonsmokers 759 scnt, ha•cd on the 19f6-68 natic,nal sur%ey and 1968-75 :,ctivc Mormon rates. The results for women :rre similar. hut the changes arc less dr.,matic. Tal,le 2 indiatcs that between 1914 ,,,td 10G6-68 the lung cancer death rate for white females %.I1o "nc%cr smol:cd" increased by about 13-fold for ages 65-g4 ~r;trs nnd by about sevenfold for,ages 35-84 years. Nlnst of the relative increase occurredRbefore 1935, but sotne incrcases continued in the older age groups up to 1,);8-59. However, since 1958-59 no further increase has lxrn noted. primarily bctN.ccn 1920 and 1030: +.ith incrc:tsing use of chest radiology, hronchoscopy, and c%tology and rcfcrr.tl of patients to hosPit;tls for cli:,ynosis (1, 24). For instance, if 3: of the deaths certified to respir;itor}• tubcrculosis in 1914 %.cre rcjah-duc to lung e.tnccr, the reported 1914 lung canccr ratcs would be increased by a factor of 7 for males and a factor of 5 for females; larger errors X•ould further increase the 1914 rates (8). Recent estimates indicate that diagnostic impro%ements may have accounted for up to a tenfold increase in the lung cancer death rates in the L'nited Kingc.+om since 1916 (24). The precise effect of diagnostic improvements on the U.S. lung cancer death rates is unknown. ~ Sources of Error From the data presented in tables 1-3, only the gross &:+nges should be considered definitive, because smaller changes possibly are due to %arious types of error that may be present. For instance, no claim is being made that the lung cancer death rate for U.S. white male nonsmokers was precisely 1.7 tirries as high in 1966-68 as in 1958. only that the 1966-68 rate is much greater than the 1914 and 1935 rates and appears to be some- ,.hat greater than the 1958 rate. First, there is the statistical error associated with sur- ,eys. An example of this is the small number of lung cancer deaths in the 1914 data among persons over 75 ~ears of age and the 1958-39 and 1966-68 "never smoked" categories among persons betw;•-n the ages of 35 and 54 years. Sccond, there is respondent error in the sur.•e~~s due to the fact that the smoking habits of all the deceased and many of the living were determined from proxies, usually the spouses, and not from the affccted individuals themselves (23). This error may be greatest for older women, who are usually not survived by their husbands. However, the error has been shown to be quite small among nonsmokers in one study (11), and it is reasonablt. to assume that whatever error might be present is similar in the 1958-59 and 1966-68 surveys. Third, smoker classification varies. For instance, the 1966-68 survey classifird persons as "never smoked cig- arettes," Ahereas the 1958-59 survey used "never smoked." However, the 1958 rates for males who "never s:no1;ed cigarettes" have been estimated by use of results `:om the 1954-63 cohort studies in table 3. Fourth, there may be undisclosed biases or errors in the 1958-59 and :966-68 mortality and population surveys that make ;`cm unrepresentative of the U.S. population and that could affect the death rates in either direction. But the remarkably good agreement in nonsmoker lung cancer death rates for 1958 U.S. white males and the 1954-63 pooled cohort tends to diminish the likelihood of any large-scale errors in the 1958 survey. The 1966-68 sur- %ev data have been examined in great detail elsewhere and have been found to be quite reliable.' Furthermore, the total white male and white female lung cancer death rates estimated from the two surveys agreed within a small percentage with the corresponding rates obtained from U.S. vital statistics. Finally, and probably most importantly, there have been changes in the diagnosis and certification of death due to lung cancer. These developments took place L'OL 62. NO. 4. APRIL 1979 DISCUSSION By use of the data a%ailable on the U.S. population from 1914 to 1968, it appears that the rclatitic increase in _the lung cancer-morulity--rate_for non"smokcrs liasd been larm, The increase has occurred in 311 age groups from 35 to 34 years, but it is conccntratcd itt those o%er 65 ~ears. •The increase [oryvhite males 65-84 years olcL,,% has_ been about 30-fold,,,i;hereas the increase for all white males 35-84 years old has been about 15-fold, on the basis of the reasonable assumption that the rates in 1914 describe nonsmokers. The sevenfold increase fc&Z white fe_males 35-84 years._pld,;:ollows the same patternr_j as ": thai :for"~: hite' maies,~lhough - it has been 1817 dramati.Z.N,tost of the relati, e increases apparently have occurred before 1935, but significant increases ha%e occurred since then, particularly in persons 65 ,ears old and above. As seen in tables I and 2, the 1966-68 age-adjusted lung cancer death rate for %%hite male nonsmokers (22.8/100,000 persons) is about one•third the 1958 rate for all white males (77.6/100,000 pcrsons) and the 1966-68 rate for white female nonsmokers (8.3/100,000 persons) is only slightly less than the 1958-39 rate for all .e hite females .(11.7/100,000 per- sons). These comparisons indicate the current magni- tude of lung cancer in persons who ne%er smoked cigarettes. Furthermore, the rising lung cancer rate among nonsmokers implies that other factors must have a substantial influence on lung cancer mortality in addition to personal cigarette smoking. Among tha:,&. influences that could be affecting both smokers and--a nonsmokers are changes in diagnostic aiceria, increases_ in environmental carcinogens,-.and certain constitU20- tional factors (2): No rigorous data are available on the effects of diagnostic changes, but various estimates suggest that the true 1914 rates could be much larger than the reported rates (1, 2). As mentioned earlier, if 5 a of the deaths certified to respirator.• tuberculosis in 1914 were really due to lung cancer, the reported 1914 lung cancer rates would be increased by a factor of 5-7. The early mortality data may also have been influenced by other relationships between tuberculosis and lung cancer, which are explored in detail elsewhere (25). Before 1935, diagnostic changes probably had the most impact on increasing lung cancer rates. More recently, en%ironmental carcinogens and constitutional factors J NATL CaNCER INST ~ ~ ... .., sJ i, i ~ / TINY 0005807

. 760 Enstrom ntay have had a siynificant cffcct. Environmental circinogcns encompass occupational carcinogcns like aslx•Ntos (26, 27) and c•nt•ironmental poiiution including environmental tobacco smoke (28, 29). Constitutional factors include genetics and general host susccptibility (l, 2). Before any final conclusioas can be drawn, the lung canccr rates am~ng nonsmokcrs, especially since 1050, should be cor ~.rrned with further analysis of the existing nation- surveys and cohort studies, as well as t.•ith another t-jonal survey to measure nonsmoker lung cancer rates tn the current decade. But if one takes the a.ailable dat.t as a whole. apparcntly lung cancer rates among nonsmokers have now reached significant lc.els. lndeed, the 1966-68 national survey and 1968-75 active Mormon survey indicate that the annual age- adjusted ;ung cancer death rate among white m3les N. ho never smoked cigarettes is now approximately 25 per 100,000 persons for ages 35-84 years. If one assumes that these findings are 3ccurate, then additional under- standing of lung cancer etiology is necessary to supple- ment the t,elt-escablished knowledge intolving cigarette smoking. Further study of nonsmokers could provide an important means of ascertaining the non•smoktng- rclated aspects of lung cancer etiology. REFERENCES (1) Smoking and Health-Report of the Adtisory Committee to the Surgeon Genenl of the Public Health Service. Public Health Serv Publ No. 1103. Washington, D C.: U.S. Go~t Print Of- fice, 1964 (2) 8ustCt+ PR: The Biology of Cancer. A`ew Approach. Balti- more: Univ Park Press, 1976 (3) CottNFtEt-D J. HatSstEL W, H.sststOND EC, et al: Smoking and lung nncer. Recent evidence and a discussion of some questions. JNatl Cancer Inst 22:173-203, 959 (4) HAEMZEL W: Epidemiological tests of theories of lung cancer etiology. Public Health Rep 71:16l-172, 1956 (S) Mortality from Cancer and other Malignant Tumors in the ReEistntion Area of the United States. 1914. Bu-eau of the Census, U.S. Dept Commerce. Washington. D C.; U.S. Govt Print Off. 1916 (6) Cot'ESt M: Cancer Mortality in the L'niced States. 1. Trcnd of Recorded Cancer Mortality in the Death Registration Suces ot 1900, from 1900 to 1935. Public Health Bulletin No 248 . Washington, D.C.: U.S. Govt Print Off, 1939 (7) I•ltusoaE BK: Trend of lung-cancer mortality :n rhe States: Some limitations of available statistia. JNac! t=sncer Inst 16:267-284. 1955 (!) GuLU.+st AG: Trends of mortality attributed to carcinoma 31 he lung: Possiblt effeeu of faulty certification of deaths eo o~her respiratory dis;easa. Cancer 8:1130-1136, 1955 (9) CottDorc T. CRITTENDLr M. HAE\szEL W: Cancer moru itr trends in the United States, 1930-1955. Nail Cancer Inu Nfonosr 6:133-350, 1961 J NATI. C:ANCER INST ME2~87 (!0) N.munnl Center fur llralth Sutistics: 1'ical Scatisucs of the l.'nucd States, Annual Relsnrt% lor 1958. lyi9. 1966, 1967, 1968. R,-clvtlle. \td.: Nal center licalth Stat. t~alungion, D C: U.S. Gust Print Off (11) Ii.ENS/LL t'. 1.ocrt .xn DB. Sutt,t. NIC: Lung cancer morulIty as trlricd to residcnce and stnr,ling historcrs. 1. Whtte m3ies. J Nail Cancer Inst _3:947-1001. 1962 (12) HAE.NSZEL W. TAet ur.1t KE: Lung cuncer mortality as related io residt-nce and smoking histurics. 11. White fcmales. J\'atl Cancer Inst 32 803-838. 1964 (13) National Center for Health Stitistics: Cigarette Smoking Status-June 1966, August 1967, and August 1969. Von hly Vital Sutistics Report. vai 18. No. 9 Isuppl Dec. 5. 1969). Rocf.%ille. Md.: Natl Ccnter Health Sut, 19G9 (1I) Cont..E1 F, fCsct Er'.EL oL: Cigarrtce Smoking and Diffcrennai Mortality. New Esumates from Rcprescni:tnve \auonal Samples. in He,tlth. l:ncced Sc.irs. i975. DIiE'r•' ^ubl No. (HRA) 76-1232. Rockcille. Md.: Natl Center Hcalth SUL Washington, D C.: U.S. Covt Print Of(, 1976, p;63 (13) Kntir; HA: The Dorn study of smoking and mortaliiy amons U.S. veterans: Report on eight tnd one half r.rs u( obscr- 1ation. ,;atl CGncer lnst NlonoSr 191-1:6, f966 (16) HAI-+.~+O%D EC. Smoking in relation to the dcjth j,es of one million men and women. Nail Cancer fnsc.cute `.lunogr 19: 127-204, f966 (17) H.,ststoND EC. GAstFtNs;EL L. LLw EA: Longesity, sclecti•e mortality, and cotnperitive risks in relauon to c!tcminl carcinogsnesis. Environ Res 16:153-173. 1978 (id) E.NSTRONt JE: Cancer mortality among Mormons. Cancer 36: 825-841, 1975 (19) Cancer and coul monality among acti%e Nformons. Cancer 42.1943-1951, 1978 (20) H!.E.NSZEL W, Sttt~+s t~ NIB, MtLLER HP: Tnbacco Smoking 1.uerns in the United States. Public Health Nfonogr 45:1-105, 1956 (21) \11LMORE BK, CoNOt'Ett .4kG: Tobacco consumption in chc United Sutes, 1880 to 1955. Public Health Ntonor,r 45 10'- I l t; 1956 (22) DOLL R: The age-distribution of cancer Implicaiions for models of carcinogenesis. J R Sut Soctesy, Series A General) 134:133-166, 1971 (23) AHr.tLD P1, CLttsOn GA: Changes in Cigarette Smok:ng Habits Between 1955 and 1966. Public Health Serv Pubi No. 110,00, Series 10, No. 59. Rockrille. Nid.: Nat) Center Hcalth SuL Washington, D.C. U.S. Covt Print Oft, 1970 (24) Royal College of Physicians: Smoking or Heafch: The T'nird Report '-om the Ro5a1 College of Phrsicians of London. Londo: Pitman Medical Publ Co., 1977, p 53 (25) AoKt K, lrstx J. SruN' SC Studies in the eptdemiologs of !ung cancer tn relauon to pulmonary tuberculosis. in Epiderns- ology Aspects of Tuberculosis in Japan and in Israel (Kltngbetg AtA, Turner 1, eds). Ness-Ziona: Israel :nstiwte for Siological Research, 1969, pp 30-56 (26) SAFStOTn U, WAOONEx J, eds: Occupational Carcir,ogenesis. Ann NY Acad Sci 271:1-516. 1976 (27) SLUKOFF 1J: Cancer risk of asbestos exposure. In Origins of Human Cancer (Hiatt HH, Watson JD, Winsten JA, eds). Cold Spring Harbor, New York: Cold Spring Harbor L.abora- cory, 1977, pp 1765-1784 {2d) RowrLizi C. Dat'ts W, eds: Ensironmental pollution and car- cinogenic risks. LARC Sd Publ 13:1-454, 1976 (29) Editorial: Breathing other people's smoke. Br Med J 3:{53-454, 1978 1'OL 4". NO. 4. .iPRIL 1979 TINY 0005808