Tobacco Documents Online

20006

LLS. DEP~ OF HEALTH AND I"IUMA, N SERVICt~ Pub|lc Hea|th Service National Institutes of Health Populations at L~ Ris}~ of Cancer NCI Mo, nograph T109371645

Journal of the National Cancer Institute l~In~ 1980 Volume 65 Number 5 Table of Contents Vol. 65, No. 5~November 1980 Monograph: Populations at Low Risk of Cancer Cancer and Mortality in Religious Groups 1055 Cancer incidence in Mormons and non- Mormons in Utah during 1967-1975 1063 Cancer in Utah: Risk by religlon and place of residence 1075 Cancer mortality among Mormons in California during 1968-75 1083 Cancer risk factors: An analysis of Utah Mormons and non-Mormom I097 Mortality among California Seventh-Day Adventists for selected cancer sites 1109 Cancer mortality in a human isolate !115 American white Protestant clergy as a low- risk population for mortality research Cancer Patterns in Ethnic Croups 11.2"/ Cmcrr lxlltems or rour ethnic grotlps in Ilawaii IIII C~ncrr mo,tallty among Chinese i.n the I.'nltc~ Stat,:s Cmcrr moltality risk among .Japanese in the Unltcd States 1157 C.ttlo.-r in Alaskan Iudians, Eskimos, and J. L. L),on, ]. W. Gardner, D. W. West ]. L. Lyon, ]. W. Gardner, D. W. West James E. Enstrom D. W. West, ]. L. Lyon, ]. W. Gardner Roland L. Phillips, Lawrence Garfinkel, ]. W. Kuzma, W. Lawrence Beeson, Tero Lotz, Burton Brin Alice O. Martin, Judith K. Dunn, ]pc L. Simpson, Carolyn L. Olsen, Sam Kernel, Michael Grace, Sherman Elias~ Gloria E. Saree, Bion Smaller, Arthur G. Steinberg Haitung King, Frances B. Locke La.u, rence N. Kolonel Haitung King, Frances B. Locke Frances B. Locke, Haitung King dnne P. Lanier, William ]. Blot, Thomas R. Bender, ]oseph F. Fraumeni, It, Cancer Mortality in Nonsmokers in the United States 1163 Cancer mortality among nonsmokers in an E. Rogot, .I. Murray insured group o~ U.S. veterans 1169 Cancer mortality in nonsmokers: Prospective Lawrence Carfinkel study by the American Cancer Society 1175 Cancer mortality among a representative James E. Enstronf, Frank H. Godley sample of nousmokers in the United States during 1966-68 Summary and Comments on the Workshop 1187 High-rate cancers among ]owqisk popola- tip,s ll~l Appendix i I~B Pa~t icipams Ioseph F. Fraum.eni. Jr. T109371646

Populations at Low Risk of Cancer Sponsored B~: Field Studies and Statistics Pro~a'am Division of Cancer Cause and Prevention National Cancer Institute National Institutes of Health Beth~la. Maryland A Workshop held in Snowbird, Utah August 25--o5, 1978 Editorial Committee Chairman: Brian MacMahon Gcnro~e Cooley Joseph F. Fraumeni, Jr. Pete~ Grcenwald William T109371647

Cancer and Mortality in Religious Groups Moderators: Mark Skolnick, Ph.D. Mary C. King, Ph.D. TIO~7 lr:~t8

i Cancer Incidence in Mormons and Non-Mormons in Utah During 1967-75 J. L Lyon, M.D., M.P.H., ~" 4 j. W. Gardner, M.D.,. M.P,H., ~ and D. W. West, Ph.D, ABSTRACT~Oatl from the Utah Cancer Registry w~m uxd to compare cancer incidence In Mormons and non-Mormons In Utah k)r the I~dOd 1967-75, Chul~h membership was Identified for gT.~ of the 20,379 ¢al~s In Utah by a torch of the c~ntral m~nb~r~hlp fllel of the ChurCh of Jesus Christ of LatteP-Day Saintl (or Mormon Church). Sites associated with smoking (lung, lewnx, pharynx, orel cav'd~, e~ophagu~, and urinary bledder) Ihowld an Incidence in Mormons at about one-half thet of non- Mormons. RIt~'a of centers of the breast, cervix, and ovary wore low in Mormon wench: the rate for cervical cancer wls about one-half of that obl~rvld in non-Mormons. Cancers of the ltomach, colon-rectum, and pencreas were about one-third lower in Morrnona than in others who ere not members of thia religious group, Mo~t of the differences sHn in clncor incidence can be explained by Mormon teachings regarding sexual activity and • lc0hol and tobacco ux, but some differences (e.g., colon and stomach) remain unexpl~in~cL---JNCI ~ 10~5-10$I, 1980, That personal habits have a significant influence on the development of various malignant neoplasms has been established, and the association of these factors with tumor development continues to be a main focus of epidemiologic research. Large groups whose mem- ber~ have similar known life-styles offer opportunities for epidemiologic study of certain risk lactors and their consequent association with malignant diseases. We have used the data accumulated by the Utah Cancer Registry to describe cancer incidence in Utah (1) and in the large (900,000 or over 70%) Mormon population of the State (2). Thus the Utah Cancer llegistry is the . largest single source of cancer incidence data on the Mormon population available in the world. The Church of Jesus Christ of Latter-Day Saints (Mormon) was founded in 1850. From the beg/nning, it has been an evangelical Church, and its membership has increased to include over 4 million meml~rs worldwide. Beginning in 1847, the Mormons settled in the Great Basin area, includir~ Utah, for the next 50 years. Tens of thousands of converts immigrated to tlds area from all parts of the world, with most of them coming from Canada, Northern Europe, and other parts of the United States. Thus the history of the Great Basin and of the State of Utah has been shaped by the Mormon Church and has produced a large, homogeneous population with a fairly uniform li~e- style (3). Some beliefs and practices of the Mormon Church that may relate to its members' state of health include emphasis on family flee, education, strict sexual mores, and abstinence Erom alcohol, toixacco, tea, coffee, and nonmedicinal drug~. Emphasis on the family places high value on children and homemaking, which likely explains why the State has had the highest birth rate in the nation for many decades H). The emphasis on education has put the Sta~e ahead of other Sta~es in educational achievement; therefore, Morrnous are dis- proportionately overrepresented in the science and business worlds (~). Because the strict sexual mores proscribe all premarital and extramarital sexual inter- course for both men and women, Utah has been among the lowest of all States in the incidence of venereal diseases and percentage of il]eg/dmate births for many years (6). The use of alcohol, tobacco, coffee, tea, and nonmedicinal drugs is prohibited by the "Word of Wisdom," a law of health given by Joseph Smith in 18S$ and to which active Mormons adhere (3, 2). Our pre~ous studies demonstrated that cancer inci- dence in Utah is £0--S0% below that expected from national rates and this is primarily due to the lower incidence in its Mormon population. Cancer incidence for all smoking-related sites and for the female sites of breast, cervix, and ovary in Mormons was significantly lower compared wid~ that in non-Mormon and na- tional populations. Colorectal cancer was also seen to have a lower incidence in Utah. We have instituted several studies o~ specific cancer sites in Utah; these include reviews of cancer of the cervix (~), malignant melanoma (8), and prostate cancer (9) and case-control studies of cervical, colon, ovarian, bladder, and brain cancers. Most of these investigations are still in prog- ress, and other detailed ones are planned. We have also used the Utah Bureau of Vital Statistics to in- vestigate cardiovascular disease (I0) and cerebrovascu- lar disease (Lyon JL, Bishop C, Nielsen N: Unpub- lished observations) in Mormons and non-Mormous. This report details cancer incidence in Mormons and non-Mormons in Utah for 1967 through 1975. ~ Presented at the Workshop on Populadons at ~w ~k ~ ~r ~ at Snowbi~, Utah, August 2]~. l~& ~ Sup~ in ~ by Public H~I~ ~ ~ntmct ~1- ~ from ~e H~d St~i~ and Statistics ~o~m. ~si~ d ~n~ ~u~ a~ ~don, ~tlonal ~n~r lmdtu~. ~ ~visi~ ~ Ep~em/ol~. ~ent of Family ~ ~ ~nhy M~i~. ~ ~e U~h ~r R~s~. Uni~ity of U~h ~l~e~ ~ M~i~ne. ~h ~e City. U~h ~ " Add~ wp6nt ~s~ to Dr. Lyon at the ~r~ent ~ Family and ~mm~hy ~icine. J'NCL VOL. ~5. NO. 5. .'~01¢~Ma£g I~ T109371649

MA~ ANOD The Utah Cancer Regisu--y was established in 1966 to register cancer cases for the entire population of the State. In 1973, the Registry became a member of the NCI SEER Program of cancer registries. SpedaI efforts to ensure completeness of reporting were described in (/, 2). In 1967-75, the 20,$79 case~ of caner d~agnosed in Utah residents were identified by the Registry. Table I gives the percentage of cancers hhtologically verified by site and those identified only by death certificate. Data for IM6, incIuded in our previous reports, were excluded from dds study because of probable overstate- merit of incidence dm-ing the F~ y~r o~ ~m ~ll~- don ~ ~e esmb~t of ~e ~pu~fion-~ ~ ca~ ~s clarified a~ng ~ ~m~rship in • e Me.on ~h. ~rminm~t d mem~rship ~s made ~ssible with ~e u~ of the central mem~r- s~p file, wh~h ~ d~ in devil (10). ~e ma~i~ ~du~ i~lud~ ~rch of all ~n~l ~u~h mem~hip fiI~ of ~e living ~d d~d; mashing ~s b~d on name, county of redone, y~r of bird, and ad~do~l i~o~ation when n~ed. ICD axie No? Site Percent ICD eocle No.~ Site Percent 140 Lip 98 180 Cervix uteri 08 141 Tongue 97 181 Plaeenta 142 Salivary gland 95 182 ~erpu, uteri 145-145 Mouth 97 183 Ovm'y'-tubt-llgament 97 M8"-149 Pharynx 97 184 Female genital system lf~0 F~ophteus 91 18~ Prmtat~ gland 94 151 Stem~eh 88 186 T~t~ 97 lf~, Small intestine 97 187 Male genital s3mt~m 98 1~8 Colen 98 188 Bladder 98 1~4 Rtetura-tnu~ 98 189 Kidney 98 l~ Liver 78 190 Eye lf~ Gallbl~dder 94 191-19~ Brain and central nerveu~ 91 157 Paner,~a 73 ~era 160 Ns~tl eavit~-sinus 98 193 Thyroid gland 98 l~t L~r~x 9"/ 19~ Other end~rine glands ~62 Lun~ 88 Ill.ddln~! sites 16~ Pleura 98 1~ Unknown prim~r~ site 80 164 Media~tinum 95 Lymph, non-Hod~kin's 95 1~ Respiratory ~n-aet 100 Hodt4dn'e disease 98 170 Bone 98 Multiple mysloma 91 171 Connective tissue 97 Aeute leukemia 172 Skin, nmlanoma 98 Chrenie leukemia 98 174 Brea~ (female) 97 Other hematopo[efie sy~mm 83 175 Breast (male) 94 Total site~ 94 The percentage of cases confirmed by death certificate only was 3.2. Coda numbers am from International Classification of DiseMe, (Eighth Revision). Total Utah population Mormon population NowMormon population Mal~ Females Male, Female, Male, Female 0--4 59,916 57,100 43,§72 41,667 16,844 5-9 59,610 57,494 $6,608 $7,$81 28.007 !9,918 10-14 fr2.881 60.489 42.117 44.111 20.764 !6.~78 !5-19 80.084 ~9.$8~ 48,205 48.679. 11.859 " 10.90~ 20~24 49,68~ 5&9~8 40J)98 41,8fl4 9,~89 • 25-29 ~ 40,440 29.865 29.5~8 8.989 ~0-~4 ~0.'/0']' 31.188 21,40~ 22,917 9.894 8,249 85-~9 27,116 27,970 18.8"/6 19,481 8,240 3.f~9 4044 ~.988 27,847 17,~/7 I&790 9.911 8,f~7 45-49 ~6,~91 28.948 17,421 18.~0 8,8~0 8,6"/8 M ~4~ 25~231 14~.779 16,787 9,648 8,444 M9 21.145 21.847 142,43 15685 ¢90£ 6.18~ 60--64 17.549 18. 73"/" 12.0~ 12.881 5J;27 5.8/6 65-69 1~,~16 l&,~6 8~06 11,418 ¢410 3.808 "~0-74 9.720 12.151 5.706 8,173 ¢015 3.9"/8 >-75 12All 18.~.$ 10.0fll 18.,~42 2.380 Total ~0.575 ~&72~ ~80.8~I 400.914 159.754 I,~2.M8 J'NCL VOL ~5. ,~0. 5. .~OVEMBf.~ ]e~e T109371650

C.amm¢ In Mornm~ and Hon-Mormmm 1057 Religiom pt~erence was also verW~-d by search of f~. ~e U~ Moron ~ia~on for July ~nsm ~r~ ~e ~e di~n ~ estimat~ ~m a 5~ ~ys~c ~mple of the ~ur~h ~ken in Aunt 1971 (2, ~0). ~e non-Moron la~on ~ ob~in~ ~ ~ub~on of the Moron ~pula6on from ~he m~l S~te ~ns~, which a~ by liner int~la~on ~ the 1~0 U.S. ~nsm and a 1975 ~pulafion estimate from ~e Utah State Bur~u of Vim] S~tistics. Table 2 gives ~e July 1971 ~pulations by a~. We com~ ~e in~dence of ~n~r in ~e Moron and non-Moron populations of the State wi~h the TN~ mt~ for whit~ usi~ SIR (11). ~e SIR, expre~d in ~rcent, is ~e ~tio of ~ n~r of ~s~ ob~ in ~e ~tive Utah ~pulafion divided by ~e humor exacted in the ~ ~t~ held for that ~pulation. ~e ex~ ~s were ~l~iat~ by appli~on of ~e ~ a~-s~ific inci~nce mt~ m the res~fve Utah ~pulation (with the 5-yr age ~ou~). Smtisti~l s~ifi~n~ of the ~fferences for ~ch ~pula~on from the ~ was ~te~ined ac- cording to ~e me~ of ~lar and Ede~r (12). ~e- adjust~ in~dence ~tes were ~lculac~ by the direct meth~ with ~e 1~0 U.S. white ~pulation used as • e smn~. Diffe~n~s ~tw~n rotes for the 2 ~oups studi~ were ~st~ with a m~ifi~tion of the Mantel- Haenszel pr~dure, which controlled for sex and age in 5-~ age in~s (10, 13). RESULTS Of the 20,379 cases, 439 (2.2%) were excluded because of inability to ascertain religion. Of the 19,940 cases remaining, 12,175 were classified as Mormon and 7,765, as non-Mormon. Text-figure, 1-4 show the SIR for both populations in Utah during 1967-75 for selected sites. Age-adjusted incidence rates are given in table 3. Utah's comparative advantage in cancer incidence. which i~ primarily for Mormom, is clearly reflected. The favorable cancer experience for the State, com- pared with the nation as a whole, has been reported for mortality data that sl~3w Utah has 22~ less cancer mortality than does the United States (14). Previous studies on cancer incidence show Utah had 19.% lower rates than the national, and Mormon and non-Mormon rotes were 25 and 7%, respectively (1, 2). The currem resul, are similar in magnitude for the Church group (26% below TNC,$ rate,), but the non-Mormon rates are now 11% greater than those of the TNCS. Cancer sites having a strong a~ociation with ciga- rette use in other studies (15), e.g., lung, larynx. pharynx, oral cavity, esophagus, and bladder, demon- strated the largest difference between Mormons mad non-Mormons, with the incidence in Mormon men and women 54% below that in the non-Mormons (text- fig. 2). This difference varied from a 69~, difference between Mormons and others for" emOmgeai and laryngeal cancers m a ~ differem~ f~r bladder cancer. Lung cancer, which contributed one-half of the total smoking-associated cases m each group, was 55~ less in the Mormons compared with the non-Mormons. How- ever. the nonoMormons still had rates 18% below those expected from the TNCS. Thu~ the overall rates for non-Mormons in smoking-associated sites were 6% below the TNCS expectation. Rates for breast cancer in Mormon women were 18.% below those expected from the TNCS and ~0% below their counterparts (text-fig. 3). The same pattern was seen for ovarian cancer, but little difference was found in cancer rates for the corpus uteri. Cancer o| the uterine cervix (invasive) followed the same pattern previously reported (2, 6), with rates in Mormon women ~5~ below national rates and 65~ below other whites. The unusual finding of low breast and cervical cancer rates in the same (Mormon) population is still seen° Although the rates for cancers of the prostate gland and testh among Mormons were slighdy higher than expected from the TNCS, they were similar to or lower than the corresponding rates for other whites. Mela- noma was significantly above TNCS expectation in both groups with non-Mormon rates higher than Mormon rates. Lip cancer also presented a similar pattern, though" at a level about three to five times higher than that in the TNCS, which shows that Utah has one of the highest rates for lip cancer in the world (16). Other digestive tract cancer sites showed significant differences between the two groups (text-fig. 4). Can- cers of the stomach, colon, rectum, and pancreas each had rates in the Mormon population that were 80-40% below those expected from the TNCS, whereas rates in non-Mormons differed from those in the TNCS only for female rectal cancer. Cancer of the pancreas has been associated with cigarette smoking, but this is not true for stomach and colon cancers. These two sites usually have an inverse relationship; i.e., if one is high, the other is low, rather than both being low in the same population (17). DISCUSSION Between 1967 and 1975. cancer incidence in Utah was significantly below that of the United States as estimated by both the TNSC and the SEEK Program (19). Virt~ually all of this difference is attributable to the lower rates in the Mormon portion of the Utah population. Thi, difference may be real or an artifact o~ the case-ascertainment procedure~ of the Utah Can- cer Registry and/or the matching pro~'dure~ used to ascertain religion. The R~gistry has gathered data throughout the State since 1966 and, since 1973, it has been a pard.dpant in the SEEK Program. Cancer cases were usually identi- fied by. the $8 hospitals in the State, but additional annual searches were made of the 21 pathology Inborn- JNCL VOL. ~. NO. 5. NOVEMaER T109371651

1058 Lyon, Gardner. m~d We~ I I I I I I ! ! 0 LIP I f ! t I I ! ! ! ! o cIrRYI X II~IIIII~NIJ)I ~N~D! I MELANOMA STOMACH COLON- ~ECTUN M~ANQMA STOMACH PANCREAS NECTUM ! I I ! NLO$ T~x:r.~GUR~.S I-4.----31R for sek.c~ed cancer sites by rcli[ion ;nd sex, utah, ]967-75, LDS=Mormon; NLDS=non.Mormon. torics and 2 radiation therapy centers serving the Share. AI! death certificates with mention of .cancer are routinely forwarded to the Registry. For this period, only 5.L~ of all cancers were ascertained with the death certificate as the sole source. A l-year survey of the records of these decedents revealed d~.¢ most of them were elderly individuals who were diagnosed and treated in doctors' o[fices or in nursing homes. O.m-of-state diagnosis and treatment do not represent a serious problem in Utah, with an average of four to five cases located out of state annually. We have no reason to suppose that a Mormon would be less likely to be identified and diagnosed with cancer than a non- Mormon. In fact, cancers of unknown sites were significantly more likely in the non-Mormons than in the Mormons. The cenu'al Church files used for identification of membership contained approximately 5.5 million rec- ords a~ the time of this study. The file of living members was computerized in 1975 and has trans- JNcI. g'OL. ~..-N'O. 5..~01rF.M~Eg 1900 T109371652

ICD eerie M~e Femzle M~e Female 140 ~p 11.~ ~ 0~ 26 l~ M 1~ 141 To~e 1~ ~1 0.~ lS &~ 46 1.78 I~ G~m a~ mou~ 1.01 ~ 0,~ 16 ~.~ ~ ~1 147 N~,x 0~ 14 0.~ 9 0,~ S 0,~9 I46, 1~149 O~er pha~x LIT ~ 0~ 8 &01 38 .0~ IS0 ~ I~ 42 0~ T 4.~ ST I.~ 152 S~ll in.fine 0.~ ~ 0.74 ~ 1.~ 13 1.18 15 l~ ~l~ 21.~ ~9 18,~ ~2 2818 311 30.~ 372 1~ ~m 11.~ 257 7.70 2~ 12.14 1~ 9.75 1~.0 ~ver 1.~ 32 0.~ 21 1.~ 18 0.78 10 1~.0 ~llb~dder ~ ~ 1.~ ~ 1.~ 15 2.51 1~6.1. 156.1- ~her bilia~ p~ 1~ 28 1~ 31 0.~ 10 0.~ 11 1~.9 167 P~e~ 8.47 1~ 4.17 1~ 9.~ 108 7.~ 1~1~ ~er di~fi~ or~ 0.71 18 0~ 26 0.~ 8 0.~ 8 161 ~x 3.~ ~ 0.~ 10 9.~ 118 1.27 1~ Lung ~.~ ~ 4.71 137 ~8.~ ~ 10.74 1~, 1~ O~er mpi~WW ~ 1.~ ~ 0.~ 19 1.~ 27 1.~ 18 170 ~n~ ~ join~ 1.14 41 0.~ 19 1.49 19 0.~ 171 S~ fi~u~ 2.19 ~ 1.~ ~ 3.71 46 2.57 1~ S~n, ~l~m 5.~ I~ 5.~ I~ 7.~ 89 8.5 ~ [ 1 174 B~t 0.73 17 61~ 1.7~ 0.~ 9 ~.~6 1,0~ In sire~ 1~.18 517 48.~ In~i~ 8.~ ~ 17.~ I~.0 Corp~ u~ri 21.19 ~ ~.40 1~.9 U~ HOSr L0I ~ 0.~ ll 1~,0 ~ I0.~ 3~ 14.75 1~.0 Va~na 0.17 5 0.7I 9 1~,~1~.3 VuI~ eli~s 0.75 ~ 1.69 181. I~.~ ~her fem~e genial 1.~ 33 1.~ 18 1~.9. s~m 1~,9 186 T~tia ~.~ ~ 3.74 1~.0 Penis 0.~ 7 0.78 , T 1~.1-187,9 Se~mm a~ o~er 0,~ 2 0,~ 0 ~e ~ni~l 1~ Bl~der le.~ ~0 4.46 1~ 24.10 ~ Z~ 1~.0 Kidney ~.~ 1~ 2.~ 77 7.~ 89 3.~ 1~.~I89.8 O~r urin~ o~n, 0.~ 14 0.~ l0 0,~ 10 0.~ 7 l~ ~e and orbi~ 1.~ ~ 0.~ 29 LIT 15 0.~ 13 ~91-1~ B~in a~ o~er 4,~ I~ 4.46 141 5.~ 71 4.~ ',,3 T~id ~knd I.~ ~7 ~,~ 182 1.~ ~ T.~ ~'q ~he~ e~rine glands 0~ 10 0.g4 I1 0.~ Y 0.~ 8 ~ph~ma aM ~ 7.~ I~ 4.~ 1~ 7.71 94 5.~ 74 tleul~ e*ll H~fkin~ d~m 4.31 I~ 1.~ 6~ 8.~ 49 4.~ ~her ~phom~ LT5 "~ 1.37 4~ L49 18 g.70 Mu!~ple meylomt 8.~ 78 g~ 66 ~ ~ 8.~0 Acu~ ~ph~c 1.~ ~ 1.18 46 1.~ ~ L~ le~e~u . ~nie ~ph~e ]e~ 2.~ 47 0.~ 28 3~ ~5 1.24 kemi~ ~er ~ph~e l~ 0.~ 8 0.17 5 0.~ 7 ~13 2 m~ ~er ~ le~emM 2,91 75 2~ 74 ~ ~ LS1 ~ber ¢~nie leukem~ 1.~ ~ 1,~ ~ 2.12 ~ L46 18 ~her ]e~emi~ 0.~ ~ 030 ~ 0.99 14 ~4~ 6 • Code numbers ~tre from the Internaficmal Cla~ifieatlon of s Site ~ not included in th, All sites" ¢ NOS=not otherwis* Di~J~e~ (Eifhth Revision). J.~cL vot_ e..xo. 5, NOVEMBER T109371653

acziom on o~er one-dz/rd of ~ ~ ~By. ~e ~ ~r n~ ~nS ~s ~d by r~ular ~y bu~ simply ~ find as ~ny ~ ~ ~ible the mem~rship files and ~ use ~, bi~h date, and ~unty of r~idence at time o~ diagn~sis as ~tching ~gisw~ su~r (when p~ible) who obtained mo~ info~a~on, s~h as nam~ of s~s and ~n~. A rumple of nam~ ~ in o~ prior study (2) also submitt~ ~r u~ in this study; a ~ment of 1.5~ ~U~ ~t~n the o~naI match and ~e ~t~. ~is e~r ~s ~istent for ~th popula- tions. ~o~ dec~enu identified as Morons from obitu~ no~ces we~ includ~ in that ~li~ous group ~ess of Church mem~ship [ile-ma~ching r~ults, which ~y have inflated ~e Moron ~pulation (~ey pm~bly repre~nted non-Mormons with spouts who ~Ion~d to the ~u~h). In p~Hng the population estimates for u~ in this study, we ~s~ver~ ~at an e~or had ~en made ~rlier in the over-~ Moron ~pulatlons, which un~rs~ ~he ~u~h~ffiliat~ pro~rtion and over- s~t~ &at of the non-Moron. ~e co~fion of ~his e~or d~ Moron ~tes by a~u~ ~I0% but i~d &o~ of other whites by 15-30%. We us~ cor~t~ populations in this ~r, with the ~sult ~ some of the sit~ t~t now show si~ifi~nt differences ~w~n ~e 2 ~oups were not s~n in the ~rlier ~port (2). ~e ~fia~ons in ~ncer incidence for smoking- a~ia~d sites (lung, la~nx, pha~x, o~l cavity, ~opha~s. and bladder), which we ~d previously ~rted, ~rsls~ and we~ in some ~s~n~ ~ter (2). ~ncers of the pancr~s and ~dn~ did not show signifi~nt ~fferences in the ~rlier study, but here • ~ ~d a 30-40% Io~er inciden~ in the Moron population. ~at ~h~ noplasms have b~n weakly ass~ia~d wi~ ci~rette smo~ng in other stu~es U5) may expl~n the di[[e~nces in rotes in Utah, With its la~ nonsmoking ~pulation. Rat~ for lung ~n~r for the non-Morons were still ~low tho~ of the TN~. but retie rates we~ app~hi~ thee for ~l~ (SIR=74 ~d ~, tively). ~e ~ason for the low in~den~ of lung ~n~r among non-Morons was n~ cl~r, but it may ~flec~ that ~ey ~ve l~s op~rtu~ty m smo~ in ~i~I settin~ and in ~eir workpl~e. Da~ on oncer of the esophagus (which h~ ~en s[rong]y ass~at~ with a1~hol u~) ~ ~flsistent ~th subs~ntiaHy low~ al~hol ~nsumption in the Mo~ ~pu- ~affon. W~n we sub~c~ ~th ~e obse~ and ~ for smo~ss~ia~ ~n~r sites (including ~ and kidney) from the total of aH ~ for each popu]afion, we ob~ned an SIR of 84 for ~e Morons a~ I17 ~or ~e non-Mo~o~. Th~ when smo~ng- ~on still h~ m~s 16~ ~low ~e na~on~ ~ and 28~ less than the nowMormon. D/fferences berwcen the 2 populations therefore persisted at a number of sites not associated with cigarette smoking, including cancers of the stomach, colon-rectum, uterine cervix, ovary, and female breast. Differences in breast cancer rates may be associated with the high birth rate in the Mormon population. Preliminary data of studies on c~ncer of the uterine cervix indicate marked differences in promiscuity be- tween the 2 groups (/9). This finding may explain the unusual finding of low rates of breast and cervix caneers in the Mormon population. The etiology of ovarian cancer is less clear, but its incidence seems to parallel that of breast cancer. The finding of low rates of colon and stomach cancers in the same population is also unusual, e~pecially for men. In other areas where colon cancer rates are low, those for stomach cancer are often high (]5, 17). The higher than expected rates in Mormons for malignant melanoma of the sk/n and cancers of the lip, prostate gland, and testis are also unexplained. Several case-control studies are nearing completion and will provide information on some etiol6gic factors; other investigations are planned to elucidate further etiologic relationships in cancer sites that show large differences between the rates for Mormons and non-Mormons. REFERENCES (/) LvO~ JL, GAr, anzlt JW, KLAvs£~t MR. i~ide~e a~ ~mllty "in Ulah. ~er 1977:59:2~2618. (2) Lvos ~ ~caCt MR, G~o~t JW, SMUT CR. ~ncer ci~n~ in Morons a~ non-Morons in Utah !~1~0. (~) ~mN~ON LJ. ~ D. The Mo~ ex~fie~a hism~ (4) S~c~ JC, ~Avt's SO. Moron feniliw ~mu~h ~If a ~nm~: ~o~er ~t of the A~ni~on hy~thoh. Biol l~4;21:7~76. (5) ~n~ ~. ~1 o~n~ of Ame~n ~iemim and ~ho~n. ~en~ 1974;185:4~-5~. (6) ~to~ JW, LYOS JL. ~w inc~e of ~i~i ~er in ~n~. S~t ~ ~ty: Ch~h of J~ no~ in U~. ~ J Epide~oI 1~8:!~:~5. (P) ]~ DW. PoWSLL J. Pms~[ic ~n~ J Epi~miol Uo) LYO~ ~. W~g HP. G~o~£a JW. ~t'a[~ M~ WI~A~s RK. ~i~u~r ~mllty in. Morons a~ non-Morons (12) ~ JC III, ~ F. S~nifi~ facton for the ~fio oi a 6ons ~ ~e Mame/-~en~l ~. J Am (17) 3~ TJ. Mc~v ~V. U.S. ~n~r ~lity by count~ pub]~6~ ~ (NI~74~]5]. (15) S~n ~L S~ a~ ~: A ~rt T109371654

~ in Mormons and t'6on-M~ 11)61 JNCI. VOL. ~. NO 5, NOI~%M~ZR 19@0

Cancer in Utah: Risk by Religion and Place of Residence,., J. L Lyon, M.D., M.P.H.. ~" ~ J. W. Gardner, M.D., M.P.H., 3 and D. W, We#~, Ph.D. AB,.RTRACT~We ¢oml~red ¢=rm~ in¢iden¢~ dudng tg67-75 between Mormonl and no~'PMormon= tMng In urban end rural • f~ll Of Utlh. The non-Mor~o~ urban men had a 34% higher dsk of cancer compared with their rural counterparts. Moat of th~ Incmaao in dak o~urred in aitel all, elated with tobacco end for ¢an¢om of fie storn,~ch, colon, and pfostato gland. Url~an Mormon el/e= had no significant Ineres=~ in risk. Tho urban femalo populltlon was at higher risk than win tho rural reglrd~ I~=* of rellglcm. TPm Increase wee not a= striking as that ob=erwd in no~-Mormon men (8 vs. ~4%); cancare of ga~trolnt~atlnll tract= I~d famlla genitalia contributed to the elevated risk. We concluded that personal habits such il ImOk- Iflg and drinking and reproductive factor= warn poer~bla explana- tions for the previously observed urban-rural gradlents in cancer dlko---JNCI aS: 10¢3-1071, 19~0. Generally speaking, conditions related to the place in which we live, including size, location, and in- dustry, have an important bearing on our risk of developing cancer. Life-long city dwellers experience higher risks of developing cancers of the respiratory, gastrointestinal, and genitourinary tracts and of the female breast than do residents of rural areas. This observation has been made in a number of studies (1-5), and the variation reported is likely to be a~od- ated with multiple factors including differences in external environment, diet, occupation, and personal habits. Approximately equal proportions of the Mormon (76.8%) and non-Mormon (78.4%) residents of Utah reside in the urban areas of the State and share the same external environment, but they may have differ- ent dietary and occupadonal patterns (5). Mormons and non-Mormons differ in their risk of acquiring certain cancers, and significant differences in cancer have been found for sites associated with cigarette smoking and alcohol ingestion and for malignant tumors of the female breast, ovary, and uterine cervix (7, 8). tk-cause of these differences in cancer risk between Mormons and non-Mormons and the asseda- tion of many of these cancers with differing risks for urban and rural residents, we report cancer incidence by place of residence for Mormons and non-Mormons in Utah. Members of the Church of Jesus Christ of Latter-Day Saints (or Mormons) settled the region including the State of Utah between 1847 and 1900. This settlement created an ethnically homogeneous population with a high degree of uniformity in social practices, and, in 1970, approximately 70~ of the Utah population were members of this Church. Church beliefs, both formal and informal, continue to shape such diverse character- istics as marriage customs, family relationships, and educational levels for a large segment of the State's population. Church proscriptions include strict sexual mores for both men and women and abstention from the use of tobacco, alcohol, coffee, and tea METHODS The 20,379 cases of cancer diagnosed in Utah resi- dents in 1967-75 were identified by the Utah Cancer Registry, a population-based registry covering the en- tire State since 1966. Special efforts made by Registry personnel to ensure completeness of reporting were described elsewhere (6, 7). Data for I965, included in the authors' previous reports, were excluded from this study because of probable overstatement of incidence during the first year of data collection in the establish- ment of the Registry. Using Church records, we clas- sified each case according to membership in the Mor- mon Church. Ascertainment of membership in the Church wa~ made possible by the existence of a central Church membership file, which also was described in detail elsewhere (7, 8, 10). Verification of membership by means of obituary notices from local newspapers showed only 1% disagreement. In 1970, the U.S. Bureau of the Census designated three Standard Metropolitan Statistical Areas in Utah, which comprised four contiguous counties containing 77.6~ of the State's population (Salt Lake, Weber, Utah, and Davis Counties). These four counties were defined as "urban," and the other 25 counties in the State were defined as "rural." County of residence at the dee of diagnosis is recorded in Registry files and was available for 98.5% of the cancer patients; the remainder were excluded from this analysis. The Mormon population in Utah for July 1971 was obtained by linear interpolation of annual Church census reports. The age distribution was estimated from a 5% systematic sample of the Church population taken in Augns¢ 1971 (7, 8, I0). The non-Mormon population was obtained by subtraction of the Mor- mons from the total State population, which was derived by linear in~rpolation of the U.S. I970 census A~a~U~VlA~OUS usz~ SIR=standardi~,'d incidence ratio(s); T~CS= Thi~ Nadoc, al Cancer Survey. .., ' Pre~ented at the W'ork~hop on Populations at Low Risk o~ Cancer conducted at Snowbird, Utah, Au~mt PJ-25, 1978. = $uppocced in part by Public Health Service conuact N01. CP4~ from J.he Field Studies and S(atL~fic= Progcaro, Division of Cancer C~ule and Prevention, National Cancer In~titme. ~ Dividon of Epidemiology, Deparument of Family and Com. mtmity Medidne. and the Utah Cancer Registry, University o| Utah C~llege ol Medicine. Salt Lake City. Utah $41~2. 4 .4ddress reprint wqu~sL~ to Dr. Lyon at the Department o[ Family and Community Medicine. jI~:I, V~L ~. NO. 5. NOVY.MI£R I~ T109371656

0-4 ~4.~)8 ~.~39 9.~64 8~ 1~,~8 1~,1~ ~9 ~1 ~0,210 S~ 7~1 1T.~2 14,~9 ~145 ~4 ~ ~.5~ 9~ 9.~ 8.~5 10,~ 1,1~ ~ 2~ ~8 6,~ ~965 7~0 9,1~ 1,~9 1,716 8~ 16,~1 18.311 4,6~2 4.~ 7,~6 6,S~ 1,~8 4~ 1~9 1~118 S,M8 4.~ 7,~ 7,171 2,118 5~ 1~ 12,~9 8.479 4,118 ~ 6,~5 2,~8 1,~9 ~59 10.~2 11,159 ~,761 4,~ 4,~5 5,~3 1,~7 1.~9 6~ 8,~0 9~?9 3~ ~,~ ~,~ 4,~9 1,~4 1.~7 To~i ~I.~ ~9,182 89.~ 91,7~ 124.477 120.579 ~ ~9 ICD~ Cancer site Mormon Non-Mormon Urban¢ Rural¢ Differ- Urbane Rural~ Differ- ~II~~ en~'~ 140-207 All sit~s <~ 7~t ~ -- 1151" g 781" 67t 0.~ II~ I0~ -- 141, 1~ To~l,~ si~ ~ ~T ~T -- 1~ 1~, 161- ~ ~? ~T 0.01 ~ 162, 1~ i41, 14~ Onl ~ ~ pha~x ~ 39~ ~ -- 108 ~t 0.~ 149 ~ 41t ~t -- ~ ~ -- 150 E~hz~ ~ 36? ~? -- 1~ 6a 0.~ ~ 21T ~ -- 79 "- ~ -- 161 ~ ~ ~T ~7T ~ 1~* ~ 0.~ 162 Lung ~ 37T ~ -- ~ ~T 0.~ ~ ~T 78 -- 95 ~ 69? ~T -- 95 70 157 P~ ~ 69? 6~ ~ 81 91 ~ l~t 108 -- I~T 1~ P~ ~]and ~ 108" 111" -- 1~ 98 ~0~ ~ ~t 52T -- 98 • The SIR were c~mp~ed with the TNCS, by urbzn co ~ res/dence. • ICD-In~erna~ions! Clsss~fieation o~ Disems (Eighth Eev/sion). ~ T, /:~<0.0i; ". P~O.08. Both indicate significance ~f difference from TNCS. • P-values are f~m the modified Man~el-Hsenazel chi~quare test of difference by urban-rursl residence. Da~h-P>O.08. T109371657

Mormon Urban-Rtwal Camc~r Rlsk 1065 and a 1975 population e~dmate from ~he Utah ~ureau ~ Viral S~fi~ U~n ~ ~1 ~ mann~, ~ ~e use of da~ fr~ ~e 4 urea ar~ ~ ~ ~m~s (ruble I). ~e in~en~ of ~ in the ~ and non- Moron ~pulatio~ residi~ in the ur~n and rural a~ of the Sm~ was ~m~r~ with the ~ whi~e m~ if1) by m~ns of SIR, and age~dj~ incidence m~ we~ olculat~ for ~ch ~pu~tion by ~e me~ with 5-~r age inter~ls s~zed ~ the U.S. 1~0 population. ~e SIR, exp~ as a is the ratio of o~e~ ca~s ~vided by the num~r ~ exacted if ~e TN~ mt~ for whit~ held for this population. ~e exacted ca~s were ~lcula~d by appli~on of the ~N~ age-s~ific inci~nce ~tes the ~s~ctive Utah populatiom. Statisd~l si~ifi~nce of the differen~ for each ~pulation f~m the was dete~ined acco~ng to the meth~ of ~ilar and Ede~r (12). For ~ch religious group, diffe~nces in ~ncer incidence ~tw~n urban and ru~l r~idents were te~'d with a modif'u:adon o[ the Mantei-Haens~l procedure with control for age (in 5-yr intervals) and sex (I0, 13). RESULTS Of the 20,379 patients, 439 (2.2%) were excluded because of inability to ascertain religion, and an additional 187 {0.9~) patients were excluded because no county of residence was amilable (63 Mormons and 12-1 non-Mormons). Of the 19,753 remaining. 12,112 were iVlormons (9,002 urban and 3,110 rural residents), and 7,641 were non-Mormons (6,158 urban and 1,483 rural residents). Tables 2 and $ show the SIR by site for the Mormon and non-Mormon urban and rural populations of Utah: the corresponding age-adjusted incidence rates are given in table 4. Urban Versus Rural Cancer Incidence As we have reported (6), the urban population of the T~a~ 8.--~a~e~r incidence in ~tah, Z~67-75, b~ .rtli~ion and r~id~" Com~r¢~o~ of Mormons to n6a-~lormon~" ICDb Cancer site Sex Urban Rural Non. Differ- Mormont Non- Differ- M°rm°n¢ Mormone enced Mormone enee' 140-207 All sites ~ 73T llST <10"s 72T T6T -- ¢ 78T 116~ <1~s 67T 105 0.~3 141, 14~ Tob~relat~ si~ ~ ~T 106 <lff~ 43f 5~ 0.0~ 150, 161- ~ 46T ~ <lff~ ~T 86 i~~ 162, 1~ 141, 14~ Oral ea~ ~ pha~x ~ 39T 108 <1~s 3~ MT 149 ¢ 41~ 128 < 1~~ ~0~ 1~ 0.~ 150 E~pha~ ~ 36~ 1~ < I~s 88f ~ 161 ~ ~ ~ 1~0" <1~~ 37T ~ 0.~ ~ Lun[ ~ 41T 157 0.~ ~ $6 ~ 37T , 96 <1~~ 39T ~T 0.08 ~ ~T 76T <lff~ ~ ~* 0.009 I~ Bl~der ~ 6~ ~T <lff~ 5~ ~T ~ 8~ ~ <IY~ 41T 107 0.02 140 ~p ~ 2~ ~T -- 87~ 4~T ~ 2~T 431~ -- 357T 808T 151 8~m~h ~ 6~ " ~ 0.05 ~ $1 ~ 72T 95 -- T8 ~ I~ ~I~ % 6T$ !04 0.~1 4~ ~$ ~ 66T 105 <lff~ 53~ 112 0.0~ 154 ~ ~ 6~ 78" -- 55~ 57~ ~ 6~ 9~ 0.0~ 68~ 70 ~ ~ 98 0,005 ~ ~44 0.~ 173 S~n ~ I~* I~T 0.~ 99 1~ ~ l~2T 194T 0.04 I08 218 0~01 174 B~t ~ ~? ~IT <1~~ 74, 97 180 Ce~ inv~i~ ~ ~T ~* <1~~ 6~ 111 0.~8 182.0 Cc~us u~ ~ 104 13~ 0.~ 91 107 1~0 Ov~ ~ 7~ 108 0.~ 65T 9[ 1~ ~m~ gland ~ 1~ 14~ 0.Or II1" ~ 189.0. 189.1 Kidn~ ~ ~ 92 -- 68" ~ ~ ~ ~ -- 52~ 109 0.~ The SIR were compared with the TNCS: Mormons to n0n-Mormons. ICD=Int~rnational Cla~ifieation ~t DL~.~ (Eighth Revision). +. P<0.0h *. P<O.O& ~oLh iadie,,~ ~'~iff, cance of difference [rom TNCS. P-values at~ from the modified 3~aatel-Haensz¢l ebi-squar¢ test o~ difference in religious status. Da~k=P>O.O~. JNCL VOL ~,. NO. 5. NOVEMBER I~' T109371658

ICI~ Ca~cersi~e Sex Mormon 140-207 All =i~e, ~ ~4~S (4,7e4) ~44~ (;~l) ~ ~ (~) ~ (;,,9) ~ (~) ~0 (~) 1~ Lip ~ 10.1 (1~) 1~I (1~) ~ (1~) I?.S (~) ~ 0.8 (1~ 1.1 (8) 1.0 (11) L9 (~) 141 Tonic ~ 1.~ (~) 0,4 (~) 5.7 (~) 1.1 (~) ~ 0.4 (9) 0.5 (4) 1.9 (I8) 1~ (4) 1~ S~i~ gla~ ~ 0.8 (16) 1.6 (1~) 2~ (14) 1~ (4) ~ 0.6 (15) 0.4 (~) 1.1 (11) O.8 (~) 1~45 ~m a~ m~ ~ 1.1 (19) ~6 (1~ 4,0 (~) ~ 0,6 (~) 0.5 (4) ~ (21) ~,7 (7) 14~ ~phz~x ~ 0.6 (1~) 0,3 (2) 0.5 (5) 0 (0) ~ 0.4 (9) 0 (0) 0.4 (4) 0.6 (1) 146. 1~ Other pha~x ~ 1~ (21) 1.1 (8) ~.6 (~) 2.7 (8) 149 ~ 0.~ (6) 0.3 (~) 0.9 (9) 0.6 (1) 1~ E~pha~s ~ 1.6 (~) 1.8 (~) 6.6 (48) ~.i (9) ~ O~ (6) 0.1 (1) 1.1 (10) 1.8 (3) 1~1 S~a~ ~ 8.6 (142) 11.2 (74) 19.1 (120) 12.9 (~) ~ 4.6 (1~) ~.1 (42) 6.0 (59) ~.8 (9) 152 S~ll interne ~ 0.9 (~8) 1.0 (6) 1.9 (11) 0.6 (2) ~ 0.9 (19) 0.1 (1) 1.3 (1~) 0.6 (1) 1~ Colon ~ ~.3 (~79) 16.~ (110) ~.6 (~) 19.9 (52) ~ 19.4 (414) 15.7 (1~) 1~ ~mm ~ 11~ (188) 0.9 (~) 15.1 (I~) 10.1 (~1) ~ 7.6 (162) 7.8 (6~) 10.5 (1~) 8.1 (18) 15~.0 Liar ~ 1.S (2~) 1.0 (7) 2.1 (14) 1.2 (4) ~ 0.9 (18) 0.~ (~) 03 (~ 1.~ (~) 1~.0 Ga]lbkdder ~ 0,8 (14) 1.9 (8) 1.8 (11) 1.4 (4) ~ 1.9 (41) 1.4 (12) 1~.1, ~er bJlia~ pamg~ ~ 1~ (19) 1.4 (9) 1.4 (9) 0.~ (1) ~6.1- ~ 1.0 (~) ~.~ (9) 0.S (S) 1.9 (~) 156.9 157 Pa~ ~ 8~ (t42) 8.4 (~) 10.6 (75) 12.8 (~) ~ 4,~ (~) 3.7 (~) 7.I (~) 10.7 (~) ~ne~ and ~ri~neum ~ 0.6 (11) 0.5 (6) 1.0 (8) 0.3 (1) ~ 1.0 (21) 0.2 (4) 0.6 (9) 03 (S) 161 ~x ~ 3~ (~) 3.t (~0) 11.9 (91) 7.1 (24) ~ 0.4 (8) 03 (2) 1.4 (14) 1.1 (2) 16~ Lung ~ ~.1 (45~ 27.3 (181) 7~,7 (~) 40.3 (~) ~ 4~ (101) 4.4 (~) 11~ (1~) 9.4 (~) 160, 1~ ~her ~epi~ o~ ~ I.~ (19) 1.5 (10) 2.4 (gl) 1.3 (4) . ~ 0.6 (14) 0.6 (~) 1,~ (1~) 1.4 (~) 170 ~ne and joint ~ 1~ (~) 0.9 (8) 1.6 (16) 0.9 (~) ~ 0.4 (11) 0.9 (8) 1.1 (11) 0,6 (1) 171 ~ft ~s ~ 2.1 (48) 2.1 (17) ~5 (~) ~,4 (10) ~ L7 (4~) 2.0 (~6) 2.6 (~ 9.5 (8) 178 S~n ~ 5.9 (119) 4A (81) EI (70) ~8 (18) ~ 6~ (14~ 4.8 (86) 8.8 (~) 10.0 (~) 174 Brmt ~ 0~ (14) 0.4 (8) 1.3 (8) 0;8 (l) ~ ~.5 (1~) ~5 (425) ~.9 (~) ~,8 (185) 1~ C~, i~ ~" ~ 15.9 (4~) ~.9 (9~) ~.4 (4~) ~.7 (9!) C~ ~v~ ~ 8,1 (1~) 9.4 (67) 17.7 (181) 18A (~) ~9.0 Co~ u~ ~ ~.9 (4~) 19.1 (148) 1~.9 U~, NOS~ ' ~ 1.0 (~) 1.1 (7) 0.7 (7). 1,0 I~.0 ~ ~ 10.7 (2~) 9.6 (7~) 15.a (l~) I~4 (~) !84.0 V~ ~ 0~ (4) 0.~ (1) 0.7 (7) 0:6 (1) I~, Z~.a V~ c~r~ • ~ 0.8 (17) 0.7 (6) 1.7 (17) 1.9 (4) 1~, 1~.~ Other retie ~ni~l ~ 1.0 (~) i.i (8) 1.0 (11) 1.7 (4} I~.9. I~.9 I~ T~ ~ ~ (102) ~Z (~) S3 (37) ~6 (9) ~ST.0 P~is 6 ~ (5) 0.~ (~) L4 (6) ~5 (1) " Number~ of ~ are given is par~ke~s. Incidence m~l~.~ ~pulation we~ a~e ~us~ by the direr meth~ ~ t~e U.S. ~ [CD=Jn~on~ Cl~ifiea~on ~ Dimm (~h~ ~ion~ T109371659

ICD* Cancer site Sex Mormon ?~o~Morrnon Urban gural Ur~ Ru~l 1873-187.9 189,0 189_~'-189.8 190 191-19£ 194 Scrotum and other male genital c~ 0.1 (2) 0 (0) 0 (0) 0 (0) Bladder c~ 16.5 (277) 13.8 (93) ~.0 (217) I~9 (46) ~ 5.1 (ltO) 2.5 (21) 7.4 (71) 6.5 (15) Kidney c~ 5~ (96) 5,8 (~) 9.0 (~6) 7.2 (~) ~ 2.7 (60) 2.1 (I7) $.8 (37) 4.9 (11) Other urinary organs ~ 0.6 (10) 0.5 (:~) 1.6 (10) 0 (0) ~ 0.4 (9) 0.2 (1) 0.6 (6) 0 (0) Eye and orbit c~ 1.1 (21) 0.9 (6) 1.4 (IS) 0.'/(2) ~ 0.9 (I9) 1.$ (I0) 0.9 (9) 1.5 (3) Br~in and other nervous system ~ 4.7 (I01) 5.4 (38) 6.3 (57) 4.5 (13) ~ 4.5 (108) 4.4 (~) 4.7 (47) 4.1 (I0) Thyroid ctand ~ 2.1 (46) 1.6 (11) 2.1 (18) 1.5 (4) ~ 5.5 (141) 5.4 (41) 7.0 (73) 10.7 (24) Other endocrine ~ 0.8 (6) 0.2 (2) 0.4 (4} 0.9 ~ 0.3 (9) 0 (0) 0.6 (6) 0.6 (I) Lymphmarcom~ and ret/eulum ~ 8.1 (155) 6.2 (42) 9.9 (77) 5.4 (15) cell sarcoma ~ 5,2 (113) 3.2 (25) 5.6 (56) 8~ (17) l-lodgkin's d~seMe ~ 4.4 (I01) 4.0 (29) 3.9 (87) 4.2 (12) ~ 1.9 (55) 1.4 (12) 4.4 (46) 2.5 (7) Other lymphomas <~ 2.0 (28) 1.2 (8) 1.8 (12) 1.1 (4) ~ I~ (2T) 1,8 (14) 2.6 (~6) 4.0 (9) MuIUple myeloma ~ 3.4 (~6) 3.1 (21) 5.4 (37) 2.9 (7) ~ 0-.4 (61) L6 (13) 8.8 (~) 2.6 (6) Leukemins A~ute lymphocytic c~ 1.5 (41) 1.7 (13) 1.7 (21) 0.6 (2) ~ I.$ (~r) 0.9 (8) I.S (1~) 0.9 (8) Chronic lymphocytic ~ 1.8 (81) 2.5 (16) 4.4 (28) 2.3 (7) ~ 1.0 (29) 0.7 (6) 1.2 (II) 1.7 (4) Other lymphocytic ~ 0.3 (5) 0.2 (1) 0.7 (4) 0.8 (2) ~ 09. (4) 0.I (I) 0.2 (2) 0 (0) Other ~cut~ ~ 2.9 (55) 2.9 (20) 4~ (31) 2.5 (7) ~ 2.4 (58) 1.9 (16) 1.8 (13) 2.0 (5) Other chronic ~ 1.6 (27) 1.6 (11) 3.2 (20) 1.0 (3) ~ 1.1 (25) 1.1 (9) 1.5 (14) 2.2 (4) Other ~ 0.8 (13} I.~ (9) 1.3 (9) 1.0 (2) ~ 0.6 (14) 0.8 (7) 0.3 (3) 1.8 (3) Other or unknown sites ~ 10.T (182) 10.4 (70) 24.1 (159) 13.6 ~ 8.5 (18~) 7.8 (65) 14.9 (142) 11.0 (24) • N'umbers of case~ are given i~ par~thezee. Ineldenee ~@I~,~ ~lation were a~ a~us~ by the dir~ meth~ ~ the 1~0 a~nda~ ~pulation, " ICD=In~afional ~fion d Di~ (Eigh~ R~sion), ~ Ce~x in si~ ~ n~ ~elud~ in ~s for aH si~s. ~ NOS - ~t o~e~ State showed higher cancer incidence that did the rural population at those sites reported by others to demon- strafe this difference (1-3). However, when classified by religion, a different pattern emerged (table 2). Mormon~.--The most striking finding was the pau- c/ty of an urban-rural difference in cancer incidence for Mormon men (text-fiBs. !-4). This was espec/ally noticeable at tobacco- and alcohol-a~iated sites where no significant d/fference at any cancer site was found (text-fig. 2). The colon was the only major site where a significantly higbe~ cancer inc/dence was seen in urban Mormon males. Mormon women demonstrated a small (8%), but significant, difference in cancer incidence by residence. Most of this difference was contributed by the excess rates noted in urban females for cancers o~ the colon, breast, and tobacco-related si~es. Non-Mormor~.--ln contrast, the non-Mormon men had a substantial (34%) gradient between urban and rural cancer incidence (text-figs. 5-8). Much of this gradient was contributed by the tobacco-related cancer sites, particularly lung cancer (text-fig. 6). Large differ- ences were also seen for cancers of the stomach, colon, and prostate gland. Non-Mormon women also demonstrated a difference in incidence by residence, although it was not nearly as striking as that of their male counterparts. The small number of cases and the less striking differences made this evaluation more difficult because no female site attained statistical significance. Nevertheless. gradients were present for tobacco-related sites and cancers of the stomach, rectum (although not the colon), breast, uterine cervix and corpus, and ovary.. Lip cancer, traditionally higher in rural than urban JNCL VOL. ~6. .~0, 5. NOVEMBER 19~e T109371660

West K~DNEY I I I I I I I LYNPH~ LEUKEIIIIA NERVOUS SYSTEM ALL SIT~S KIDNEY LYMPHQMA LEUKEmiA SY3TEM I I I N~rml I I ! ! T[.~r.Fm~'a£ ].--SI11 for the U~ah Mormon population by urban and rural residence for selected cancer sites, O LARYNX ESOPHAGUS BLADDER ! ~Urllen FEMALE LARYNX LUNG ESOPHAGUS BLADDER el SIR "r$ ~ 125 1 ! I I T£XT.FIGUK£ ~mSIK for the U(ah Mormon population by urban and rural residence for selected cancer sites. 1~7-75. areas (1), also behaved in the same way here, but at a much higher level than expected for both religious groups. Prostate cancer showed no urban-rural gradi- ent for Mormons, but it showed a significant one for others. - Mormon VersuB Non-Mormon Cancer Incidence This comparison" demonstrates that most DE the previously reported differences in cancer incidence between the Mormon and non-Mormon populations arise from the urban segments .o[ each population (table ~). The urban Mormon population experienced a cancer incidence [or all sites about 35% less than that of the non-Mormon population.. This was most st~ng at tobacco-related cancer sites where the Mormon rates were about 55% lower than the non-Mormon tares, but this lower incidence was also seen at the major sites of the stomach, colon, rectum, pancreas, female b~ast, uterine cervix, and prostate gland. A much smaller difference existed for the rural areas and was more pronounced in women than men. The tobacco-related cancer sites demonstrated a gradient, which, again, was g~ater for women than men. Such major sites as the colon, s~omach, and prostate gland had no significant difference [or men. Significant gradients between Mormon and non-Mormon women were found fox melanoma and cancers o£ the colon, pancreas, breast, and the uterine cervix. The non-;~Ioxmon population fits the pattern de- scribed earlier of higher incidence in the urban areas JNCI. YOL 65. NO. ). .~OV~.M~£~, i~0 when compared with the rural (I-3), but the Mormon population does not fit the pattern, especially males. Whereas this observation is most striking at sites usually associated with tobacco and alcohol use, it is also seen at other major sites with little connection to these well-established carcinogens. DISCUSSION Validity of Data The da~a-gathering system, used by the Utah Cancer Registry and described in (6-8, I0), includes routine surveillance of all pathology ]aboxamries serving the 6EN ITA~.IA ~IIEAST CERVIX (I NVASIVE| CORPUS UTEri OVARY PROSTATE SIR 0 ~ 50 75 I00 IZ~ ~0 I T~XT-I~Gt'RE ~.~$IP,. for the U~ah Mormon population by urban and rural residence [or the s~lcc~'d cancer si~cs. ~?-75. T109371661

the Utah Mormon pc~puiz~ion b.~ ml~n and rural ~'~idev~ for se|ec~ed c~m:er sites. I ur~n :rod rural residence for ~elecxcd c'~ncer ,~ite~, t~7-75, T109371662

Lyre% Gardner, and West OVARY I, PROSTATE I T~.'t'T-RGt:I~£ 7.~SIR for the Utah non-Mormon population by urban and nwal residence for ~elecied cancer sites. 1967-75. State, with 95% of the cases histologically confirmed and 3.2~ reported by death certificates only. Our use of metropolitan counties to define urban areas differs from some other investigations. Haenszel et al. (1), in their study of urban and rural cancer in Iowa, used the "old" U.S. Bureau of the Census definition of incorporated places of greater than 2,500 persons, whereas more recent studies have used the metropolitan counties to define urban areas. Haensze] and his associates (14, 15) who reported on lung and colon cancers and Levin et al. (3) in their study on urban-rural cancer differences in upstate New York Migration between urban and rural areas may intro- duce bias. However, ac~rding to Church records, the proportion o~ the Utah rural population who are Mormons has remained constant (at ~,75%) for at least 30 years. The similarity of lung cancer incidence in the entire rural population suggests liule migration, and an explanation for a Mormon urban-rural gradient like that found for non-Mormons would require an urban migration of rural Mormons in excess of the total rural population of the State. The clustering of dds predominant religious group in rural areas of the urban counties might also artificially lower urban Mormon rates. However. ac- cording to census data, if these rural areas were totally inhabited by Mormons, the remaining urban Mormon population would be reduced by only 4%. Urban-Rural Cancer Differences The failure tO find an urban-rural difference in cancer risk in Mormon men is the most striking finding of this study and contrast~ sharply with those in the non-Mormon population and in all other studies. Because the~e populations share the .same external environment, the observed differences in can. cer risk by place of residence are probably attributable to differences in life-style and personal habits (with the M|I.INOMA STOMACH PANCREAS C0L.QN-- RECTUM kip MEI.ANGMA SI'~MACH PANCREAS RECTUM ~III 0~5 SO ?~ I00 ll~ 150 I l I I I i • i ! I I I I I Rural I TLx'r~;u'~ 8.--SIR for the Utah no¢~Mormea populad~ by urlma and rut'a] residence for ~i~t_~l cancer si~e~, 1967-75. JNCI. YOL ~5. .~0. ~. .~I01"F..MBt~'I, I~ T109371663

Mocmo~ Urben-Rerd Caeeer Flbk 1071 a~umptkm that di~erer~es in ~~ ~ ~ ~e). D~ ~ ~ ~ ~n M~s ~ ~~ ~~ ~ in ~ ~d a~ol u~, ~mily s~, ~d, ~ha~ ~ a 1~ ~. wm~ cmto~ edu~tionM levi. a~ ~u~fion (~9, 16). ~me ~s~h~ ha~ ~ high ur~n versus ~I lung ~n~r i~n~ as e~den~ for ~e bagful ~f~ of ur~n air ~llution if. I7-19). ~e four- ~,~ty metm~Hmn a~. ahho~h not as ur~n in ~e s~se ~at New York City, ~mn. and ~it a~ ur~n, bas a [~rly den~ population of over ~,~0 and ~eml major industries including a cop~r smelter, a st~l mill, and an oil ~[ine~. However, we found no ~ter lung ~ncer risk for ~e Morons living in ~e ur~ than in ~e ~ml areas. ~erefo~, the increa~ lung ~n~r incidence among ur~n non-Mo~o~ is probably attributable to differen~s in smo~ng habim or to syn~gism ~tw~n smo~ng and air pollution or ~cu~fio~! ex~sure ~, 14, 15). ~e indden~ of ~n~r in other sites a~iated with ci~tte smo~ng {o~l ~vity and phar~x, l~nx, ~ophagus, and ~na~ bladder) also was higher in ur~n non-Moron males, which ~rhaps ~s similar explana~ons. Higher urban ve~m ru~ indden~ of breast ~ncer s~n in non-Moron women is consis~nt with other studies and may ~flect differen~s by ~sidenc¢ in repr~uctive pat~ms. ~e crop.sis by their Church on large families may decr~ ~sidential differences in repr~uctive patte~s for Morons and explain ~eir similar ur~n and ~ral br~st ~ncer mt~. ~ncer of the ute~ne ce~ix pre~n~ a more complex pattern. ~rlier studies have shown a higher in~den~ of in.sire ~i~l cancer in ur~n ~an in ~1 a~s (1-3). However, we found the op~site in Utah. Higher inciden~ of in situ ~noma of the ~ix was obeyed in the ur~n areas of the State and likely ~flec~ ~tter accepmn~ of ~tolo~c s~ning in these ar~s (20); also, the lower incidence of invasive ~ical ~ncer might ~en ~ at~bumble to this diffe~n~. No explanations a~ ~dily a~ilable for ~e urban ~fferen~s found ~tw~n Moron and non-Mor~on men for ~n~ of the prorate gland, stomach, ~d ~lon and for Moron women for ~ncer of ~e corpus ut~. ~ncer of the lip ~ in ex~s in Utah, ~rdless of rel~ion ~r ~siden~. However. h ~s slightly ~gher in rural ~1~. which ~y reflect a higher proportion of ru~l men employ~ in a~i~l- In the urban ar~s of U~, ~e Morons. in con~ast to other ~oups, ap~r to have a patte~ of living which p~u~s lower ~n~r ~sk similar to ~t s~n in r~l ~p~ations. ~e te~s "~n" and "~1" ~siden~ ~n only classify br~d ~ffe~n~s in ~- havior, and ~e s~ifi~ of the~ ~fferen~s ~ve not tion and ciga~tte smoldag (5, 1L I5. 17). Invesdgaticm of a wide vark-ty of behavioral difference~, iacluding diet. occupation, leisure activities, and pm'umal habits in urban and rural populations, may be productive. REFERENCES (1) HA~S2~.t. W, M^gcvs SC, ZtUM[~t ET. ~n~r ~rbi~ty in u~n a~ m~i ~. Public ~1~ Mon~ ~) ~L~ J, ~ A. ~m~fi~n of i~de~ ~ in ~ and the U~t~ S~t~. J Natl ~n~ ~t 1~7;19:9~8. (3) L~IN M~ H~NS~L W. ~ggO~ VH, I~G~U~U SC I~ ~r ~nci~nce in a~s ~ N~ York S~te. J ~tl ~ncer Inst ~4:1~3- !~7. (4) ~L W. DAW~N ~. A ~e on ~r~lhy • e colon a~ ~mm in ~e Uni~ States. ~r 1972;18: 2~-~2. ~) BUgLL P. ~NN JE. Reh~ im~t ~ ~oklng and air ~llu, ti~ on tung ~. ~h ~vimn H~I~ (6) LY~ J~ ~g ~, ~ MR, ~ART i~e a~ mor~lhy in U~h. ~r 1978~9:~618. (7) L~ON JL, ~tg M~ ~aVNgg JW, S~gT C~ ~ i~- ~e in ~ons a~ n~-Mo~ons in U~h, i~70. N (8) LYON JL, ~Ngt JW, W~ DW. ~er incid~ce in ~or- mons and ~n-M~ in Utah d~ng 1~7-~5. ~N~ 1980; (9) ~RIN~TON L~, B~N D. ~ Moron ex~He~a histow of the ~t~r.~y Sa~. New York: ~opL 1979. frO) L~ON J~ W~g HP, G~N~ JW, ~. ~io~r mm~lhy in Mo~s a~ n~-M~ons in U~, 19~-1971. ~ J EpM~iol 1~8;108:35~-~. (H) ~ SJ, Yo~ ~L Jg, ~, T~ Na~onal ~nc~ S~: l~dence ~. Nad ~ ~ Mon~ (~) ~ JC IlL ~aga F. ~gnifi~e lacmn for the ~tio of a Poison ~ble m i~ ~mti~. Bi~e~ 19~:~0:6S~ ff~) ~ N. ~i-~ te~ wi~ one de~ of f~om: ~ten- ~ of ~e Mamel-Haen~l !~8:~7~. ~ ~ to ~M~ a~ ~o~ng hhto~. L Whi~ ~1~. (1~) ~L~L W, TAggeR KE. Lun~nc~ mor~lhy ~sid~e a~ smo~ng hi~to~. II. White female. J Natl (1#) Spic~R JO. Gt'~AvUS ~. Mo~ genility ~ro~h half a cen- tu~; ano~ ~st of ~e Ammunition h~is. ~ Biol (17) ~ P. ~ ~e ~la~on~ ~t~n atmmphe~c ~}lutlon mp~, ~wilium, ~ly~num, ~d~ a~ ar~nic. Br (18) SHY ~M. Lun~ cancer a~ ~e ~n envlro~ent, a ~view. H~el A~. ~1 W~ ~. ~i~l impli~6~s ~ air ~]lu- t/on re~h. ~ton. ~ss: Publ~ing ~ien~ Group, 1~ (IY) D~N G. Lu~ ~n~r and bm~i~s in No~em ~h~, (~0) G~g ~W. L~ J~ ~w [~id~ og ~1 ~ in Um~ Gyne~! On~l T109371664

Cancer Mortality Amo, ng Mormons in California During 1968-75 Jame~ E. Ensh~m, Ph.D. 3, 4. s ABSTRACT--On t~e be=is of Church record=, detailed and total death rate= were determined for an average of California Mormonr during lg~-75, f~r an average of 70(L000 Utlh Mormons during 1970 i.,xl 1975. Ind for I subgroup of acthte Mormon mama known t= High Prle=ts and Sevontlel, For e.tncer t= = whole, the standardized mortality ratio wM 68% for all California Mormon male=, ~3% for ~11 California females, and • 3% for acthre Mormon mal~ in California and Utah compared with 1970 U.$. whitee. Ag~-specifl¢ and age..adJuat~d total mop tality rates were substantially lower in Mormons than in 1970 U,S. whiter, with the greatest differences occurring between 35 ind 64 y~ira of lOS, whale the rates for ~ctive Mormon males were reduced by more than 60~. Methodolo~Ic IMue$ end source= of error were dlscu=lod, and the overall qullity of the data wll good. Some health-related characteristic= of Mormons are also summadzod.mJNC! 65: 1073-10~2, 1980. The Church of Jesus Christ of Latter-Day Saints (more popularly known as the Mormon Church) has approximately £.5 million members in the United States and about 4 million worldwide. Church mem- bers are located mainly in the Rocky Mountain States, particularly Utah where they comprise about 900,000 of the State's 1.3 million population; about 400,000 Mor- morn reside in California (I, 2). They are interesting from an epidemiologic standpoint because their "Word of Wisdom" advises against the use of tobacco, alcohol, coffee, tea, and addictive drugs (l, 2). Also, the Church recommends a well-balanced diet, particularly the use of whole grains and fresh fruits and vegetables, moder- ation in the eadng of meat, and good health habits in general. Adherence m these practices varies depending on how acdvely members participate in the Church. I previously examined cancer and total mortality rates among all Mormous in California during 1970-72 (l) and among active Mormon males in California during 1968-75 and in Utah during 1970 and 1975 (2); currendy, I am examining the life-style characteristics of a well-defined cohort o[ active California Mormons (5). Lyon et el. studied cancer incidence among all Mormons in Utah during 1966-70 (4) and cardiovascu- lar mortality among Mormon white adults in Utah during 1969-71 (~). This pa~ will present cancer mortality and total mortality for all Morrnons in California during 1968-75 and for active Mormon males both in California during 1968-75 and in Utah during 1970 and 1975. Acdve Mormon males are defined in this paper as those Church members designated as High Priests and Seventies in the Church's lay priesthood. High Priests are the local Church leaders, such as bishops, patri- archs, and members of various quorums and councils; Seventies are the adult missionaries who proselytize and attempt to convert nonmembers to the Mormon faith (2). These active Mormon males are most likely to adhere to the Church-advocato:l tiff.style, and they have been selected as active Mormons in lieu of more direct information on individual Church members. No equivalent classification exists for Mormon females. Various questionnaire data from Alameda County, California (1), and throughout California (5) indicated that acdve Mormon males abstain almost completely from tobacco, alcohol, coffee, and tea in accord with the Word of Wisdom. Mormon males and females as a whole appear to use about one-half as much tobacco and alcohol as does the general population (I, 5). These data are only suggestive and not necessarily represents- dye of all California Mormons. In addition, the socio- economic status and urban-rural distribution of the active Mormon males (2, 3) and Mormons as a whole (1) appear to be fairly similar to those o[ the California white population. Furthermore, the overall Mormon diet is fairly similar to the average American diet (2, 5), but differences could occur in some specific dietary" components when extensive Mormon dietary histories are collected. The sole emphasis of this paper will be an analysis of Mormon mortality data and a discussion of errors associated with these data. For additional information, reference can be made to previously published aspects of this subject (I-5). Asilttv~o~s t,,si~. ICD=International Clas, ilicadon o~ Di~ea~: NCHS=,National Center (or Health Statistio: SMR==tandardized mortality t Pre*~nted at the Workshop on Popu)ations at Low Risk oE Can- car conducted st Snowbird. IJtah. August 23-25. 19'/8. ~ Suppotteti by grants PDT-31 and PDT-,51A from the American • C;mcer Society and by cancer research funds of the Univeni~ California. Computer assistance was obtained from the Health Sciences Computing Facility of the University M California at Los z $chool o~ Public Health and Jons~on Comprehensive Cancer Center, University M California. Los Angola,. Calif. 900£4. * ,ddd*¢ct reprint requests to Dr. Enstrom at the School OI Public Health. = I am gt-ateIul to all the people who have assisted me in this t,e,~earth. In wticular, l thank the sta~ in the FIL~tor~cal and Mernbenhip Departments and in the Office cd the Presiding Bishopric of the Mormon Church and Dr. Letter California. without w~ose cooperation and help this work would not have been po~ible. I ~lso thank Virginia Hono~e~ McCarthy. Eva Operskalski. F~rard Robertson. Donrm Scid. and John V,'right for technical 1073 JNCI. VOL. 6& NO. ft. .~O~:EMflER 19S0 TI09371665

1074 MATERhM.S At~D M~'HODS :I'he mortality rates determined in this paper were based on Mormon Church membership and death ~.~ords stored in Salt Lake City, Utah. The record- kceping system was described in detail in (1, 2). The Church clerks throughout the world are instructed to file complete and accurate records in an annual ward report, which includes membership and death data. Both the bishop and clerk of each ward (local church) must certify that their report is complete and accurate before mailing it to Salt Lake City. The ward reports are stored in the Church Historical Department, and copies of the membership statistics and death secdons o| these reports were obtained for essentially all the California wards during 1968 through 1975 and all the Utah wards during 1970 and 1975. During these periods, the number of wards averaged approximately 700 for California and 1,700 for Utah. The only ward reports omitted were those that were not sent to Salt Lake City or were misfiled; these amounted to only a small percentage of the total. The procedure for the analysis of these data was also described in (I, 2). For each deceased person listed in the ward reports, the full name and other identifying information, including sex, date and place of birth, date and place of death, reported cause of death, and the priesthood level, were transferred to computer records and processed, For California, the identifying information for each deceased Mormon was compared with the State indices, which alphabetically list the names of and equivalent identifying information on all deceased residents. The State file number was obtained whenever a match occurred between the Mormon and the State lists. Of a total of 8,820 male and 6,~86 female deaths identified in the ward reports, a total of 6,4~0 male and 6,10g female deaths was confirmed with California State death records. Of the 5% of the deaths not confirmed, about one-half occurred out of California according to the ward reports. These out-eft-state deaths included deaths of young men killed in military action • in Southeast Asia and persons who often' died suddenly from accidents, heart attacks, and other causes ac- cording to the Ghurch ward report. Only rarely was cancer listed as the primary cause of the unconfirmed out-of-state deaths. Most of the out-of-state decedents were not California residents at the time of their deaths according to the information on the non-California death c~tificates obtained [or a sample of these per- sore. About one-.half of the uncorffirmed deaths ap- peared to be probable California deaths, and why they were not listed in the State files is hard to understand. Most likely, the name of the deceased was misspelled or the incorrect date and/or place of death was recorded in the ward report. We will make further attempts to clarify these unconfirraed deaths. Inasmuch as the calculations of death rates were made with only the confirmed California deaths, the true death rate might be slightly higher if most of the unresolved deaths are evenumlly determined to .be genuine California deaths. Also, some additional deaths among California Mor- mons may not have been listed in the ward reports; this point is discussed later. Among the total of 6,820 Mormon male deaths in California was a subset of 1,552 active Mormon male deaths. Of this subset, 1,327 (98%) were confirmed with California death records, and additional attempts are being made for verification of the remaining 25 deaths. The percentage of unconfirmed deaths was lower among the active Mormon males because the ward report informadon is more accurate and complete than it is for inactive Mormons (all other Mormon males). Computer tape records containing death certificate information were obtained from the California Depart- ment of Health for all confirmed California deaths, including out-of-state deaths occurring among Cali- fornia residents. The underlying cause of death for these Californians was assigned by the Department of Health nosologist who used the ICD (Eighth Re- vision) (I). For Utah, deaths were verified differently. The 2,577 deaths among Mormon males were compared with the 4,104 deaths among Utah male residents in 1970, as summarized on a national mortality tape obtained from the NCHS (6). Because most individual identi- fying information, such as name and date and place of birth, was removed from the NCHS tape, the matching was done with the data that remained on the tape, Specifically, the sex, month of death, age at death, county of residence, and underlying cause o[ death available for each Utah decedent were matched with corresponding information available for most of the Mormon decedents. Inasmuch as a nearly one-to-one correspondence was found in the comparison of the 2,577 deaths among Mormon males with the 4,104 total deaths among Utah males, unique and unequivocal matches could be made without name and date of birth for 1,824 Mormons, including 703 of the 904 High Priests and Seventies, For the remaining 201 active Mormon males who could not be unequivocally iden- tified with the use of this matching procedure, death certificates were requested from the Utah Department of Health for verification. To date, 884 deaths (98~) among the High Priests and Seventies have, been correctly identified, but the remaining 20 Church- reported deaths have not been found in the State files. A similar matching procedure was used for 1975. The 2,555 deaths among Mormon males were com- pared with the 4,333 deaths among ma/e residents of Utah in 1975 that were summarized on the NCHS national mortality tape (6). The matching for this year was done by sex, month and day of death, age at death, and county of residence. Again, unique and unequivo- cal matches could be made for 2,041 Mormons, in- cluding 814 of the 987 High Priests and Seventies; death certificates were requested for the 173 unmatched. "1o date, 972 deaths (98%) among the High Priests and Seventies have been correctly identified, but the maining 15 Church-reported deaths have not been T109371666

State nosologi~t [ICD (Eighth l~wision)]. The population-at-ri~k was de~ermined by the tabu- latioa of the number of Ca~nia and Utah Mormon males, by p~iesthood level, listed in the same annual ward reports that contained the information on de- ceased members. The number of members reported for each ward was a~tm'aed to repre~nt d~e precise poI:m- lation for which deatha were also reported. Averaged over the S-year period (January 1, H~8, to December 31, 1975), there were in California about 1~0,000 Mormon females and about 170,000 Mormon males, including 11.800 High Priests and 3,700 Seventies for a total of 15,500 active males. During 1970, Mormon membership in Utah amounted to about $94,000 females and $73,000 males, including 40,~00 High Priest~ and 9,700 Seventies for a total of 50,600 active males. In 1975. membership was increased to about 450,000 females and 410,000 males, including 50,$00 High Priests and 9,100 Seventies for a total of 59,400 active males. The averages for 1970 and 1973 were 43,~0 High Priests and 9A00 Seventies for a total of 55,000 active Mormon males in Utah. The age distribution of California and Utah Mor- mon males by priesthood level was determined from the computerized Church membership files as of July 1, 1975, and September 15, 1976. In lieu of direct questionnaire data on individual Mormom, the com- puterized records provided the best available estimate of their age distribution. The accuracy of the compu- terized records was limited because of certain errors e.g., a small percentage of the members was not included, a small percentage was included twice under different addresses, and some members were entered incorrectly. However, the records for the acdve mem- be~ were presumably more accurate than were those for members as a whole because inactive members were more likely to be in lint or unknown statns. The 1975 computerized records for all members have been dis- erased in (I). The 1975 and 1976 age distributions were averaged and then assumed to be the actual age distribution for the |90,000 Mov-mon females and 170,000 males, irw.luding 15,500 active males in Cali- fornia during 19~8o75 and 55,000 active Mormons in Utah durir~ 1970 and 1975. For all California Mormon males and females, this was equivalent to what was previously called the "modified" age distribution in the earlier paper (I). In each 10.year age group, the 1975 and 197fi distributions differed from their com- birwi average by at mo~t ~,~, which was a reasonable estimate of the error pre~ent. The number of High Priests and Seventies at least 35 years of age averaged 15,880 in California and 50,590 in Utah. Although race was not recorded in the membership records, Mormons can be considered to be all white; evidence for this is the fact that about 99~ of their deceased members are categorized as white in the State records. The assump- tion that the average age distribudom in 1975-76 were the sa~e a~ tho~ ia 1968-7~ and 1~'/0 a~d ~fl~ ~ outgo of ~m~ d~ ~is #~, al- ~0~ ~ ~ ~ sexily ~ a~ut 1~8 to 1~5. Fmth~o~, ~ a~ ~m~ution M de- ~ ~lifornia Morons ~, ~ain~ ,taMe from 1968 ~ 1975, whi~ in~t~ ~t ~e a~ ~bu6on of ~e ~puMdon~t-Hsk is pin,My unchan#, if one a~um~ t~t the ~s~[ic M~n ~ ~t~ m~n~ ~e mine from 1~ ~u# 1~5. However, • ~ ~ ~butiom ~y ~nmin ~me as yet unde- t~min~ ~,t~tic e~or; ~ns they mint ~ q~lified as t~ ~st c~ntly available. I~smuch as ~e Moron morality ~m~ end~ly on Chu~h ~ords, using av~laMe dea~ and ~pu~don~vrisk ~m, one must ~i~ ~e of ~ morality ~tes wi~ a wHIMefin~ cohort of in~d~lly idendfi~ Mo~o~. For ~is pu~, a histoH~ bio~#i~ b~k ~tten ~ound 1950 J. Muir a~ut active ~lifomia Mo~om of ~t e~ ~s us~ (7). For ~h Moron included, #yes demo~phi¢ info~ation, inclu~ng full name, dam and pl~e of bkth, ~ents' ~mes, ~tion, and r~iden~, and allo~ a ~mch-in~ndent his- tmi~l pr~dve dete~inafion of mo~lity for a 30- y~ ~H~. Listed in ~e b~k a~ 426 roles ~d 407 females, primarily hns~nd and wife ~in, who we~ alive and at l~st 35 y~ of a~ as of Jan~ 1, 1945. ~ ~r~ns, many of whom lived in ~s ~un~, ~ve ~en followed in~vid~ily so ~eir rent viral smms could be desexing, All ha~ positively ~ses~ as ~ing either alive or dead as of Jan~ 1, 1975. ~e follo~ng pr~dur~ were us~ w ident~y 442 dec~ m~rs: Fore ~mm ~h~ M the de- c~ ~mb~hip file ~in~in~ since 19~ Moron ~urch in Salt ~ke City l~t~ 4~ d~ths;. pe~o~! contact wi~ mem~rs of ~e coho~ ~milies and other ~mch loden in Lm ~#es ~unty and a thorough s~rch of the ~lifo~ Sm~ d~ indi~, for MI ~m~s of unc~min status l~t~ ~3 ad~- do~ d~s. V~fi~don of d~ s~tu~ was ~ne by a ch~ of ~l 442 d~ c~tifi~t~ ob~i~; 95 ~ outs~e of ~omia inHu~ng 2 in fo~Ign ~es. Hemifi~:don of ~91 living m~ included fo~ ~te ~h~ of ~e ~les ~ ~e living mem~ (~mputeri~ and ~ fil~) o[ ~e Moron ~u~, ~rsonzl conga with ~hort mem~n and ~urch l~ in L~ ~l~ ~unty, a ch~k ~~t of Mowr Vehicles [or ~c ~s~n~ of a ~li~orn~ ~v~'s l~nse ~ ~ l~O or ~, and ~lifo~ia State dea~ ~ to ~in d~s sin~ among ~hor~ mem~rs ~liev~ w ~ alive a~ng w ~fo~z~on f~m ~e o~r so~. V~ffi~n of li~ng sta~ ~s ~ on ~e J'NCL VOL ~. .MO. ~. NOVEMltEIt T109371667

1075 that these members had a specific vesidc-~=c address as of J~nuary I, 1975, ~ h~, ~at a~ ~h in[o~a- tion o~ined f~m at l~st t~ of ~e following i~m sou~s: [elepho~ di~es, ~urch mem~rship r~s, ~r~ent of M~r Ve~cle ~o~s since Janua~ 1, 1~5, or ~r~t ~rsonai con- mcL ~u~ of ~ ~ns~n~ and ov~lapping com- p~miven~ of the ~om so~s, d~t ~n~ct ~th o~y th~ in~viduals with w~jor un~rmin~ in their status ~s ~nsi~ n~e~. RESULTS Mormon age-specific death rates in C.ali[ornia from 1968 to 1975 and in Utah during 1970 were calculated for all cancer and all causes by our combining the deaths and population-at-risk as determined above. The rotes and number of deaths on which they were based are given in table I for all causes of death and in table 2 for total cancer and cancers of the colon, lung, and breast. The populations-at-risk in 5-year age inter- vals up to 85 years are presented in table $. For comparison with death ~ates for Mormons, those for white males in the United States, California, and Utah (6, 8) and Utah-born Californians are given in table 1 for 1970 (6, 9). To obtain the expected deaths (tables 4 and 5), I multiplied the population-at-risk given in table 3 by the age-s~ 1970 U.S. cancer death rates (6, 8) and then added them for all the 5-year age intervals. The data in tables 4 and 5 ~show that, compared with 1970 U.S. whites, the SMK for all cancer (total malig- nant neoplasms) were ,50 and 49 for active Mormon males in California and Utah, respectively, and 68 [or all Mormon males and 83 for all Mormon females in Cali[omia. The SMR for the smoking-related sites (defined here as lip, oral cavity, pharynx, esophagus, lung, and bladder) and non-smoking-related sites (arbi- trarily defined as all those except the smoking-related ones) are as follows: Can~er sites California Mormons Utah Mor- mon Active All Females males males male~ Smoking re- 26 58 80 19 Non-smoking 67 74 85 volant These SMR value~ were statistically significantly less than I00 (at the level of P<<0.01), if one assumes a Poisson variation (I, 2). I have chosen not to test annual a~-rpecifie total mortality rates fwr Mormons in Cal~'ornia and Utah and set.rat comparison populah'ons Ararat? annual aze-sl~:ific total mortality rates" 1946-74: California 1968-75: 1970 and 1970:. Cali-1970: 1970:. Ictive California 1975: Utah 1968-75: 1970: Mormons active active California Utah Utahans'1970:" fornianSborn in forniaCali" whitesU'S" in the Muir Mormons Mormons Mormons Mormons Utah~ whitm, cohort (7) Males 85-44 L60 (1) 1.18 (~) 1.21 (,9,9) 1.98 ($12) 2.89 (99) 8.09 (16~) 2.09 (18) $.~ 8.44 4~-~4 B,~ (7) ~ (110) 8.12 (78) 4,88, (684) 7.~L~ (~) s,~ (41~ 6.44 (8~) 8.26 M 9.56 (28) $"20 (20~) 9,25 (928) 13.78 (1~) 15.76 (~96) I&12 (6~T5) 19,89 (1/8) ~0.64 22.0~ 65-74 80.18 (77) 92.66 (~51) 25.61 (505) 38.66 (1,4~2) 43.46 (602) 4~.25 (967) 46.15 (210) 45.60 48.10 75-84 77.95 (90) 61.66 (406) 68.8~ (68~) 72.65 (1.118) 70.71 (556) 88.46 (4~8) 126.92 (165) 97.25 100.99 ~65 1~L81 (89) 151.20 (206) 1~9A~ (828) 139.58 (413) 140.19 (240) 15T.~4 (872) ~ 1~6.25 185.52 -->$8a 11.68 (242) 9.36 (1.815) 10.09 (I.849) 12.91 (5,242) 15.42 (2.120} 17.21 (3.455) 18A1 19A1 Total deaths (242} (1,fl27) (1,856) (6A60) (2.577) (4,104) (741) (al! ~.-44 0.0 {0) 1.65 (998) 1.5T (,~9) 1.84 (100) 2.61 (29) 2.il 1.93 45-f)4 2.54 (6) &,~ (f~8) 8.57 ('12.8) 8.57 (18~) 6.19 (74) ,L94 4.6~ 5~r-64 8.1T(24) 8.~6(908) 7.73(21"/) 8.84(846) 12.48(126) 10.43 10.15 6~'4 1&24 (47) 18,87 (1.19~) 21.96 (42~) 92.98 (601) 26.~4 (142) 23.15 24.71 7~-84 6L98(~) 51.~(1,48~) ~6`86(~5) 60.29(825) 70~8(138) 62.~ 66.99 ~ 189.97 (48) 128.41 (980) 122.62 (~4~ 140AT (~10) ~ 154.71 ~a 8.12,(900) 8.5T (&499) 8.99 (L~'4) 9.8T (2,~6) !0.71 10.91 Total deaths (200) (6,109) (2.015) (2,957) (687) (all sees) " Number of de~t~ upo~ which raze is ba~ed is given in • Tl~e pepuktiea was 7~ " Of t~ Califernia~ bor~ in Utah. --40% were Mormon. " Values were s~e ~ b~ the direct ~ m the 1940 TI09371668

19~--7& 19'/~ ~ 19/0: 1968-75: M~rmons in California C~i~ I~R U~ U.S. muir~ ~r e~e~" Mo~en~ Mo~o~ ~ ~ ~n~ all ~n~ cancer# Males [- 0-.24 8.'Z (53) 0.0 (0) 0.0 (0) -- -- 8~S H 1&8 ~9) 0.5 (1) 1.1 (2) 7.9 (1) 22.4 (2) 16.2 86-.14 29.8 (47) 2.5 (4) 7.0 ([1) 3,8,6 (10) 21.0 (4) 50.1 45-94 97.Z (Z$6) 8.S (12) 24.3 ($4) 8~,$ (27) 71.9 (Z$) 172.0 55-64 608.8 (278) 94.9 (82) 117,8 (108) I6~.6 (42) 129.8 (82) 498.1 65-74 670,0 (289) ,~3.8 (2~) 208.6 (90) 443,3 (69) 466.6 (92) 997.0 >75 1275.7 (283) [47.8 (27) Z97.1 (36) 1.[8&2 (94) 1,128A (139) 1.617.4 75-84 1~41.6 (190) 1.14L9 (75) 1,188,8 (1[8) 1.592.7 ~8~ 1A52o7 (45) 1,397.1 (19) 892.9 (21) 1.772.2 Total deaths (1,975) (99) (281) (248) (287) Femlles 0-24 5.7 (43) 0.0 (0) 0.0 (0) 0.0 (0) 6.0 2r~34 17.7 (~) 0.5 (1) 0.9 (2) 3.7 (8) 16.3 35-44 ~9.6 (108) 2.8 (5) 63 (10) 26.0 (47) 62.4 4~ 1~.2 (2~) 11.1 (1~ 12.8 (20) 44R (72) 1~.3 ~ 2~.5 (S~) ~.6 (~) ~.7 (41) 7¢6 (77) ~74 459.6 (~1) ~.3 (42) 26.9 (1~ 90.0 (5~ ~.7 ~75 784.5 (2~) 1~.0 (~) ~.9 (12) 11&1 (~) ~.3 7~ 6~.7 (1~) ~.5 ~ ~0.3 (71) 1.1~.6 T~I ~s (1,263) (14~) (1~) (30~) (~1"~) • Number of d~ths upon which rate Ls heed iz given in /xzrent~ee~. • ICD (Eighth Revision) c~e Nm. 140-209. • ICD (Eighth Revision) code No. 153. '~ ICD (Eighth Revision) code No. 162. • ICD (Eighth Revision) code No. 174. TXBL~ 3.--A~e d~tributio~ of Mormmt popu~ationz-o.t.risk 0-4 16,~5 16,716 0 0 5-9 18.725 19,500 0 0 10-14 21.594 ~,291 0 0 1~19 I9.~ 20.878 0 0 2~ 13,678 1~,5~ 40 142 2~ 1~15 14.~ 418 3~ 10.~ 12,~ 1.1~ 3.168 ~9 9.~44 ~,619 I.~7 4~! 4~ 9.~2 I1.015 1.~ 4~9 9.1~ 10.5~ 2.1~ 6.1~0 5~ 8,4~ 9.765 I,~8 ~ 8.~1 7,4T8 1.761 6~4 6~ 4.556 5.775 1.~ ~8 6~ 8,188 4A~ 1.~ 7~74 2.259 3.~ ~ 4A79 ~ 370 ~ 17o 1.176 To~l 170.~ I~.~0 15.~ individual cancer sites for statistically significant differ- ences because underlying uncertainties that invalidate the tests may still exist in the dam. These uncertainties will be discussed. DISCUSSION On the basis of all the data presented in tables I, 2, 4, and 5, Mormons have unusually low cancer and total death rates. These low rates are not entirely explained by thdr lack of smok/ng, because they also have low SM~ [or all non-smoking-related cancer sites combined. These latter sites account for 41 and 39% of the deficit in observed versus expected cancer deaths seen in active Mormon males in California and Utah. tespe~tively, and 49 and 86% of that seen in all Mormon males and [emales in California, respectively. In representative U.S. whi~e male nonsmokers in 19~-68 compared wkh U.S, white males in 1970, the SMR [or all non-smoking-related cancer sites was essentially 100 (~). More extensive dam on representa- tive U.S. nonsmokers are presented {n this monograph UO). In specia]I.v selected nonsmokers, such as non. smokers in the American Cancer Society and U.S. veteran cohorts, the SMR [or all non-smoking-related cancer sites w~s less than 100 (IlL but these non- JNCL VOL r~. :~0. ~,. NOVEMBER T109371669

Ob~ Exp SMR Oba Exp SMR I "fetal n'~l|gntat mopMms 1,075 1,577.2 2 ~m ~ ~ 1 10.3 ~,r ~ un~i~ pht~x 8 ~.9 S~I in~ne 1 ~mm. me~i~o~ ~unetion. ~d ~ ~ 49.4 G~lbltddnr 3 ~.1 ~t~n,~. ~neum. ~ un- 4 10.4 ~ di~ o~ ~x 7 ~.fi ~ng 281 4~.7 ~er ~pim~ o~ ~ 8.4 C~n~. su~u~u,, ~d o~er ~ 6 8.8 t~ S~n, ~nom. ~ 20.0 B~t 1 2.2 Female ~ni~l U~ ~I* ~ni~ ~ ~ I~.9 ~m~ g~d ~ 115.0 Tm~ 4 10.9 Penis 0 LT S~m ~d ~r ~1, ~ni~ ~m 0 0.3 B~der ~ 49.8 K~n~y ~d ~ ~lv~ 19 ~.7 O~,r ~ ~ 3 1.8 ~e Ne~ m~m 40 ~0 ~r ~ ~ i£ 18.8 gnd~n~e ~er ~H~ i~s 2 2.6 ~mphem~ 69 ~.0 ~ph~a a~ ~tieulum cell ~ 36 39.5 H~in~ d~ ~ ~ 16 Mul~p~ ~a 18 19.1 ~ukem~ Ch~n~ 13 1~4 ~r mon~J 88 I~3 1~2$.7 88 63 25 21.2 118 12~ 0 0,2 0 2~ 3 4.9 61 -- 182 6 23 222 10 7 5.4 130 154 8 1.8 167 42 6 6.3 95 ~ 887 4~0.7 78 45 9 13.1 69 70 51 54.9 30 2 3.1 65 72 145 182.1 80 65 30 42.2 71 41 I0 19.6 51 59 8 18.3 ,44 118 9 9.1 99 63 60 72.7 83 38 13 15.8 82 57 111 139,3 80 3O 8 3.4 88 58 102 131.1 78 60 6 4.8 125 flO 7 9.7 72 68 4 8.0 50 140 26 17.6 148 45 308 324,4 93 159 231.1 69 48 64.4 75 14 18,2 77 16 31.7 50 75 106.9 70 1 2.7 37 3 4.6 66 2 2.4 83 77 37 0 0 65 3~ 48.6 66 72 16 2~.9 67 52 16 23.0 70 167 0 13 0 118 4 1.9 211 74 88 48.3 88 f~ ~9 ~8.0 88 90 9 10,8 87 66 8 9,2 87 57 4 &7 60 • ?T 4 2.5 160 90 63 84.7 82 91 28 33,7 8,3 72 16 I8.0 89 116 9 12.9 70 94 18 l&9 95 -- 106 Tt 72.3 98 110 9 11.? 77 125 5 7.7 71 4 ~-5 114 98 34 30.6 Ili 156 7 9.1 77 89 12 9.7 124 78 67 84.8 79 j,,v(~ VOL i~. ~ ~. .qOVF2d~F,t I~ T109371670

Utah O~ Exp SMR Obs Exp SMR To~a/m~li~nt ~eol~ssms ~4~ 481.1 ~ ~ 578.6 49 ~p 1 O.S ~ 0 0.7 0 T~e 0 ~5 0 0 4~ 0 S~l~ Zla~s 0 1~ 0 1 1.~ 77 Gu~ ~ ~ 0 ~ 0 0 ~T 0 H~h~x 0 1.0 0 0 1.I 0 ~er ~d u~i~ pha~x 1 5.7 18 " 0 8.5 0 ~a~s 0 10,6 0 5 19,5 40 S~h 19 ~ 75 ~ 31.6 ~]I in~tine 1 1.0 I~ 0 I~ 0 Col~ ~ 4~1 74 ~ ~.1 ~m. ~si~oid ~unc~on, ~d ~l 6 16.3 37 10 20.S 49 c~al Liar 2 7.6 28 2 9.2 O~lblsdd~ 1 1.8 ~ ~ ~3 1~0 ~er bil~ ~a~ 3 2.7 111 0 ~.S 0 Pa~ 13 ~.0 46 24 ~.7 ~t~H~ne~. ~neum, ~d u~ 1 3~ 31 1 4.0 ~fi~ di~ve or~ R~pim~ ~m ~7 157,6 ~ ~ I~,4 ~ I ?,a 14 0 8,5 0 ~n~ ~ 147.8 ~ ~ 169.1 19 Other ~plmw~ o~ns 9 2.~ ~ I ~8 Bon~ ~d jo(n~ 0 ~ 0 I ~.5 ~nn~, ~ubcu~n~u~, ~d other ~ 0 1.9 0 g 9.1 S~n, me~no~ 9 4.9 I~ 4 5.0 B~t 0 0.7 0 2 0.9 Male ~ni~l s~m ~ ~.1 76 70 ~.4 109 P~ ~]~nd ~ ~.S ~ ~ 6~,6 T~ 0 1~ 0 4 1.0 Penis 0 0.6 0 0 0.7 0 S~m ~d o~er ~le ~i~l ~m 0 0.1 0 0 OA 0 U~n~ s~m 15 ~,1 ~ 12 ~.6 B[~der 13 17.7 ~ 5 ~.4 ~d~y ~d ~nal pei~ 2 10.8 19 6 ~.5 48 0~er uHnsw o~ 0 0,6 0 1 0.7 ~e 0 0~ 0 1 0.5 Ne~s ~s~m 7 11.0 ~ 7 10.9 B~in 4 8.7 ~ 8 8.6 Other ~o~ ~m ~ 2.3 I~ 4 ~A 174 End~rine sys~m 2 L6 I~ 2 1.8 111 Th~id ~and 2 1.0 ~ I 1A Other ~d~Hne gk~ 0 0.6 0 I 0.6 167 ~pho~ 17 18,5 ~ 14 20.5 ~mph~a ~d retieulum ~I ~ 6 102 59 7 11.4 H~gkin~ dim 4 ~5 ~ ~ 4.5 O~er ~om~ 7 ~S ~ 4 4.5 89 ~ple m~s 10 6.0 167 2 7.3 ~ukemia 19 18.8 101 18 ~ T9 ~mphztic ~u~ 1 1~ ~ 0 1.4 0 Ch~nie 2 ~.6 ~ 2 4.7 O~er 0 1.3 0 1 1.7 69 ~h~ [eukemi~ (~., ~nul~c or mon~c) Acu~ 7 7.~ ~ 11 ~4 1~1 ~nie 3 ~5 1~ 2 ~0 67 ~er 6 2.9 ~ 2 3.7 Oth~ ~d ~known p~ 9 ~,6 ~ 14 ~.4 " 'l~ne ICD (Eighth Revision) c~xies were used. Appendix. p. 1191. ~NC]. VOL ~. NO. 5. NOVEt~LK 19S$ T109371671

smoke~ di~-d from average Americans in manT ways A ~j~ ~u~n o~ any ~emi~c ~udy shou~ be q~lity o~ ~ dam on which k is ~. ~is ~ns~r- a~on h ~ulvrty ~ue of ~e p~nt inv~d~fion, which is ~ on s~eml s~on~ ~our~s no~ ori~- n~ly intended ~or u~ in epidemiologic sm~es. Some of ~e ~o~ with ~c ~mpletene~ and a~u~ of ~e ~puladons-at-~sk and ~he d~ths ~ve al~dy ~n memioned, and addhional ~nsidemzions will now ~ made. The validity o[ the msuhs, as opted to • e ~suits &em~lves, will ~ ~e pri~ f~us of ~his ~s~ssion. ~e viral status og all the mem~rs of the Muir bio~phi~l cohort (7) was ascertained and vefifi~ over a ~-year ~ri~, and &ere was no loss m follow- up. ~alysis was done by ~he accumulation of d~ths and ~rson-y~rs og obse~tion by attained age with the use of ~e smnda~ pr~edu~ (12, 13). Age-s~dfic death ~tes we~ &en ~lculat~ and are pre~nt~ in mble 1. The I~0 U.S. population was select~ as the smn~ for a~ adjacent ~u~ this ~puhtion ~s used in othdr maior NCHS publi~ti0ns ~e 1~5-74 Muir cohort has an age-adjust~ mortality ~te of 11.~ d~th~l,~ for males at least age, which was 24% higher th~ &e rite of 9.~6 for active Moron males in ~lifornh during 19~-75 and 15% higher than the ~te of 10.~ for active Moron males in Utah during 1970. For females o~ [he Muir cohort, ~e a~dj~ted mo~lity ~te was 8.12 d~th~ 1,~, which was 5% lower t~n the rate of 8~7 for all Mormon females in ~ligomia during 1968-75 and 10% lower &an ~e ~ of 8.~ for all Moron females in Utah during 1970. Within the 95% ~nfidence ime~al o[ Poisson smtisfi~l variation (1~), the age- s~ific d~th ~tes for the Muir cohort we~ ~nsistem with all ~e co~spon~ng Utah and ~lifornia ~tes ex~pt those ~or active ~le Mormons ~tw~n 65 and 84 y~n of age. The rotes for &e Muir cohort could ex~t~ to ~ higher &an co~n~ng rotes centered ~ound 1970 ~ause ~hey ~cu~ from 1945 through 1~4, a ~ during which, the a~dj.mted U.S. d~lin~ by 18 and 38% for whRe ~les ~d females, ~tively (6, 14). The males o£ the Muir ~hort initially ~lec[~ wi~out knowl~ og their pfiesth~ storm, but &ey have sin~ ~m det~n~ m ~ active Morons ~co~ to Chur~ d~th ~s (a~ut ~ og the decea~d ~les we~ ~h ~es~ and 5% we~ ~venfies). ~e females of ~e Muir cohor~ had no pd~ status, but ~ndally all of ~ w~ wives of active memMrs and hen~ could ~ ~nsi~r~ ac~ ~Mrs. UnfoRu~mly, no Ch~h.~ dam ~ ~ ~l~la~ for active Moron female, who might ~ ex~d m ~ h~l&i~ ~n women as a whole, if ~e ~Htn~ o[ acd~ O~er inactions about ~e q~Iity of the Mormon in~t~n of the ~u~ d t~ d~ths iis~d ~u~h ~ ~pom w~ the h~ ~t ~ o~ all d~s of Mormo~.s in California and ~ of those o~ active Mormons in California and Utah were verified with official State records. As summarized in table 6, an indication of the completeness of the lists in the 1958-75 California ward reports was the fact that 94% (103/109) of the California deaths that occurred in the Muir cohort during 1968-74 were included in these reports. Totals of 88% (96/109) were recorded in the deceased membership file and 96~ (I05/I09) in either the ward reports or deceased file. The 4 deaths entirely unknown m the Church may have occurred among persons who were lost or removed from the Church membership files between 1945 and 1968 and hence not reported in the death records because of unknown membership status A more direct check of the com- pleteness of the ward reports of deaths was obtained by following 277 males and 487 females who were living members of four Los Angeles wards and at least 50 years of age as of January 1, 1974. Of the 33 California deaths among ward members confirmed by State death certificate records for 1974 and 1975, 29 (88~) were reported in the 1974 and 1975 California ward reports (ruble 7). The 4 deaths not listed in the ward reports included deaths of 2 inactive males and 2 elderly widows, according to the membership lists used to identify the members who have been followed. On the basis of the information above, the California ward reports may undcrreport deaths up to 5% for active Mormons and by about 10% for all members of the Church. The deceased membership file appears to underreport deaths by at least 10% for active Mormons and possibly even more for Mormons as a whole (1). These error estimates are only approximate, but they are the best available until more extensive checks can be made. The 1970 Utah ward reports listed 4,591 deaths of a total of 7,061 deaths in the State during 1970 (table 8). Thus Mormon deaths comprise 65~ of all Utah deaths, and, if the ward reports are 10% incomplete, then Mormon deaths would actually comprise at least 70% Deaths Deaths Deaths r~.orded in recorded recorded Total California in deceased in wad California wa~l member- rt~rta or deaths reports, ship d~mased eonfirm~d % of file. % of file, % o~ by death detths deaths d~tha Male ~6 (52) 89 (4,8) 96 (~) Fem=.le 93 (51) 87 (48) 96 (~) Both sexes 94 ~I0~) 88 (9~) 96 (10~) 100 (109~ • .Vam~r~ in pare~cb~e~e~ re~present the actual No. ef degtbs upon which ~e eemp|ete~ess percentage ('% of total deaths) is T109371672

Sex Deaths Total Total recorded in No. of No. ~f SMR of ob- California observed deaths served to ward Califotmia expoeted expoet~d reports, deaths ba~i on deaths. % of confirmed 1970 (95% eonfi- o.bz~_~rv~i~by death U.S. wbi~e denee limit~} deaths eer~flcat~ d~th ra~es Male 88 (15) 17 26.6 64 (87-102) Female 88 (14) 16 31,5 51 (29-88) Both sexes 88 (29) 33 ~8.1 57 (89-80) • Members were at ]ea~t 50 yr old as of January 1, 1974. ~ At'~mben {n partnt]~e~i represent the actual No. of deaths upon which the comp|e~eness percent*ge (% of observed deaths) is based. of all in the State. Using the Mormon liv/ng and deceased membership files, but not the Church ward reports, Lyon et al. (4) found only 60~ of the patients with cancer in Utah during 1966-70 reported and only 55~ of the cardiovascular deaths among whites in Utah during 1969-7l (5). These varying percentages, as summarized in table ,t, point out ~e difficulty of proper identification of all Mormon deaths in Utah with estimate ~ t~ ~ ~ ~ a~ ~ ~ ~t at ~ ~ ~ ~] U~ ~ ~ a~ U~ Mo~om. ~e hdi~tion of ~ ge~l a~m~ ~ the ~li- ~r~a Moron a~ ~s~n ~s obmi~ bg a ~m~fison of &e a~ ~s~bufi~ of ~lifomia Mor- mons with ~e age ~s~ibution of ~Iifomia ~idents ~rn in Utah, as re~n~ in ~e public ~ samples of ~fic r~ f~m &e 1970 cents (9). Of all l~0 d~ ~s~ ~lifomians ~m in Utah, 40~ we~ de~ined m ~ Morons by a ~m~son of ~r~ of deaths in ~lifornia with ~ose in ~e ward reports. ~is o~- va~on suggests t~t a~ut ~ of ~lifornians born in Utah a~ Mormons, and hen~ the a~ dis~ihution this migrant ~oup m~ht ~ similar to ~at of ~li- fornia Morons, ~ny of whom immi~ted from Utah. Table 9 shows that ~e a~ ~s~ibution ~nons at l~s~ 35 y~rs old was in hct similar for Morons in ~lifomia during 19697~ and the ~li- fornians in 1970 who were ~rn in Utah. However, similafi~ did not hold for a~-s~ific dea~ rotes. ~n in ruble 1, ~e d~ rotes in 1970 for ~lifor~ans born in Utah were about 50~ higher ~n th~ for ~lifomh Morons and higher ~n ~ose for ~li- fomia whites as a whole. ~e dir~tion of ~is ~ffer- ence was not s~pfising, if one considers ~at the of Umhans likely to mi~te to ~iifomh are non- TABLE 8.---Per~tnta~ of Mormons among total Lrtah population and oariouz t~tal ,tatistics oa determined by Mormon Church files of li~n~ a~d d~ased membtr~ and annual ~ ~s No, of To~ ~ Pereont Cohort e~min~ Sour¢e 0f Moron da~ Morons ulafion Referenc~ in U~ in U~ Moron July 1, 1970. U~h .~pula- ~nual wa~ re~ 770,~ 1,~6,~ 72 (I) tion 1970 ff~h d~s 4.~91 7.~1 65 (I. 1~70 ~h cancer ~s 3.376 6,1~ ~ (5) 1969-71 Utah white ear- diovaseular deaths among l~rsons at 30 yr old Annual ward reports Living and deceased membership files L/vin~ and deee~ed membership TAeLE 9.---Compa~son of a~ di~tributio~ fo~ .~o~moftt~ i~ CalfforMa durf~ 198&'.75 a~I ~?taJ~-bo~ ~'[dent~ of Cal~rt~a ~n 1970" Mo~no~ Mormon males in Utah-born Utah-born females in Califo~ia ~l~ in Calffo~ia f~al~ in d~ng # C~ifo~i~ C~ifo~ia &5-44 85.0 (19.716) 2~.4 (~600) ~.0 (~,f~l) 25.7 (11.I~) ~ 81.0 (17.513) ~.9 (I~) ~.6 (~) 80.9 (1~) ~ ~.8 (II.~ ~ (8,9~) 19.3 (I~) ~A (I0,I~) ~74 9.6 (5,~) ~.4 (4,~) II.5 (7,915~ ~.4 (~) 7~ 8.4 (1.913) 3.$ (1~} &2 (8,~) 6~ (2.7~) ~85 0.7 (370) 0.5 (2~) 1.4 (~) 1.4 (~) R~ 1~.0 (~6~1) 1~.0 (8~0) 1~0 (~) 1~.0 (~) Both eohor~ were at le~t 85 yr old. .Vumber~ in par~nthesel are No. of p~rsona upon whom ~rcentage distribution is based. Values ~ ba.~d on (1/5 x ,~ + 1/15 x 1,5~/) public use ~amples x 50. jam. vo- ~ xo. 5. NOVEMBER T109371673

Movmous who ~ to avvid ~ne s~mg inil~uer~ce of ~ Mo~ ~u~h in U~. ~ ~it~ o~ • ~n~ ~s ~g and ~s ~fi~ty n~ a ~~n ~ ~e ~a M~ ~e ~t av~lable ~n~a~on of ~e m~s of ~l~om~ Mo~om is ~e [act ~at they were simil~ m &~ o~ Utah Morons and U~ns as a whole. ~o&er ~adon ~e ~om &e 1~7~ mo~lity follow-up of ~e ~7 ~les and 4~ ~e~l~ in the ~* An~les ~s (ruble 6). For both ~es ~mHned, M dea~ we~ o~ com~ wi~ 58.1 ex~ on ~e b~is of 1970 d~th m~ o~ whi~s in &e Unit~ Smt~. The SMR ~s 57~ ~ 95~ confi- den~ limim ~mm $9 m 80~, i[ one am~ a Poi~n ~tion (1~). ~i~ ~nfidence inte~l, though lar~, ~ si~fi~nOy les~ t~n 1~ and consistent with • e total mor~litF ra~s oi Morons in ~llfomia (ruble 1). Another ~ni~lar ~n~n is the ~ssible e~or in • e Moron ~pulation~t-~sk induced by &e Mo~ morn am~n~ m &e "l~t and unknown" ~tego~. of D~m~ M, 1970, 287,8~9 ~rsons we~ l~t and u~nown out of a mml of 2,926,475 Mo~om, and as of D~r $1, 1975, the~ numbs incr~ to 478,1~ and ~,572,~2, r~dv~y. ~e ~li~orn~ and Utah Moron populafion-at-~sk deified a~ve po~y d~s not inclu~ the lost and unknown ~rsons, but i[ ~y o[ ~e~ resid~ in ~li~omia or Utah, • en a substantial ~or ~uld be indued i[ ~ey act~lly cont~bu~ m ei&er ~e ~pulation~t-~sk or &e ~ported d~&s. Presumabl~, mint of &~ pe~ons o~ un~in status a~ i~ctive mem~rs and should in no ~y ~[~t &e sm~s~cs lot active Mo~on~. ~em va~om ty~s o[ e~ors unsure ~e de$imbiliw that epidemiolo~s~ u~ cohom o~ well~efin~ in~viduals who ~n ~ followed inde~ndently. ~though &e ~ch ~ords ~n ~e as effecti~ m~ns o~ obtain- ing in~o~fion a~ut Morons, ~ey should ~ for epidemiolo~c pur~s only with a [ull under- ~nding of the~ limitations. CONCLUSION The major conclusion m be drawn from the results so far is that California Mormons, part/cularly active Mormon males, are a low-risk population, with cancer mortality patterns not clearly explained by their smok- ing habits. This population is excellent [or epidemio- log/c investigation because it appears to be derao- graphically and ~c~ctmemi~lly similar to the ben- ml white poptflation and yet has several distinct health-related characteristics flint could he related to its low risk o[ cancer. The existence of extensive Church membership and death records makes the conduct o[ an epidemiologic study easier than would be possible among the general population. A prospective cohort study of active California Mormons, currently uraler- way (~)0 will relate in derail the health characteristics of individual Mormons to their subsequent cancer mortality. The goal /s to gain new understanding with regard to reducing cancer risk. REFERENCES (1) ENS'fgOM JE. C~ncer mormliw among Mormom. C~ncer 1975; ~:8~1. (2) ~. ~n~ a~ mini mormliw among actiw Mo~. ~ 1~8;42:i~*!951. ~) ~. H~I~ a~ die~ p~tic~ and ~n~ mo~iity ~ong ~lifom~ Mo~ Im ~im$ J, Lyon J~ Skolnick M, ~n~ re~ 4: ~er in~de~ in ~iin~ ~pu~do~. ~ld Sp~ng Hater, N.Y.: ~ld Sp~ng ~r~r 19~. (4) L~N J~ ~UIER MR, GArDNer ~, ~AgT C~ ~nccr d~ in Morons ~ non-blo~on~ in U~h, 1~1970. N Engl J M~ ($) ~os J~ W~gt HP. O~vstt JW, ~ve~t MR. WI~ R~ ~io~uhr mortality in Morons and non-Mo~om in Utah. 1~1~1. Am J Epldemiol 1~8;108:~57-3~. (6) Nafio~l ~nter f~ H~i~ S~titti~. Vi~I t~fisti~ Uni~ S~m. annul ream for I~0 a~ 1975, i~luding momlhy ~ fiks a~ outline of i~$ and c~. Md.: Na6o~l ~n~r for H~t~ S~$t~$, 1979. (7) ML~g LJ. A ~ntuW oi Moron a~ivitie$ in ~lifom~. Vol 2. Bi~i~l. Sah ~ Ci~ ~ N~$ P~$. 1952. ($) U.S. Bu~u o[ ~e ~m$. Unit~ S~te, censu~ of ~e ~pu~- tion--1970. W~n~on. D.~: U.S. ~vt Pdnt Of L 1974. (9) ~. Public u~ ~mp~ d ~$1c ~o~ f~ ~ t970 ~$~ ~p~n a~ *~hn~ d~mtlon. Wa~hlngton, D.C.: U.S. ~ Print Of L (10) ~ou JE, ~D~g ~. ~n~r mo~iity ~oag a ~. ~ti~ ~mple d ~n~oke~ in ~e Unit~ S~t~ during (I1) G~s~L~ ~n~r ~rmliw in n~$mok~: ~ive ~t~ by the Am~n ~ S~iety. JNCI (12) ~ID. ~mp~ng ~n y~n at ~$~ Br J ~v ~ M~ 19~2~1~-1~4. (13) M~N RR. ~lyth of ~five ~u~l a~ mo~lity. ~put Bt~ R~ (14) ~o~ ~, H~ ~. Vi~i $~ti~i~ ~ in ~ Uni~ S~m, l~t~ W~hin~on. D,~ U~. ~ ~int Off, 1~ [DHEW publi~t~n No. (I$) ~t JC III, ~t F. Signifi~ faa~ for a Poi~n ~ble m its ~ntion. Biomet~ 1~;20:~ .J~CL VOL. ~. NO. ~. NO~'r.N,~F_~ I~ T109371674

Cancer Risk Facto An Analys s of Utah Mormons and Non-Morrnons, D. W. We~t, Ph.D., ~ J. L. Lyon, M.D., M.P.H., z and J. W. Gardner, M.D., M.P.H. .ABSTRACT-.-4n a coml~r~m of Mormons and non-Mormons in Utah) rn~re Mormon rr~n ~nd w~men rrmrfl~d spouses of the seine MP.t~ ~m reli~lou~ly I¢~ve. were ~f Northern Europe ancestry, lived in rural Mess, h~d fewer exposures to occupa- tional h~zerd~, were less likely to smoke cigarettes or drink coffee, t~. and alcohol, u~ed fat~ In ca~king, and w~re more ot~an mm~Jed that was the cohort of other religions. No dlffeP chose existed in occupation, but Mormon mon had c~mpleted m~re y~em o¢ s~hooling. Mormon w~rnefl ware les~ likely to be cotiege.greduete~, had fewer sexual pmlnere, had more preg- nancies, were older It first pregnancy, ware less likely to birth ao~rol pills, had fewer mlscerfleges and hysterectomfes. examined their br~ast~ more often, and h~d more breast X-rey~. For women, there was only s small difference by religion for age It flret Intercouree and no difference for age at which they began u~ing birth control pills. Religious ~tlvlty was examined for Morm~n~. ~nd In most Instance~ Inactive Mormons ware more like the non-Mormon popullUon in re~pect to the vsdlbles mea~ured.---JNCI ~5: 1083-10~5, 1980. The age-adjusted incidence rates for cancer are sir- nificandy lower in Utah than in other areas of the United States. According to a recent SEER publication (1), the age-adjusred incidence rate for 1973-76 for all cancers in the SEEK areas was 361.1 and 297A/100,000 for men and women, resl:m:tively. The corresponding rates in Utah were 300.5 and 2~0.5/100,000. The rates for specific cancer sites in Utah were all lower, with the exceptions of the lip, the prostate gland, and malignant melanoma of the skin. Within the State of Utah, the incidence rates vary by religious denomination when members of the LDS Church (Mormons) are compared with non-Mormons. The annual age-adjusted incidence rates for cancer among Mormon men and women for 1967-76 were 252.9 and 205.2/100,000, respectively. The correspond- ing rates for non-Mormons were ~44.6 and 303.8/100~000 (2). The rates for all sites of cancer were lower among the Mormons, with the exceptions of the testis and prostate gland in men and the b~in, central nervous system, and stomach in women. The most likely reason for lower cancer incidence rates among Mormons is their religious proscription on the use of tobacco, alcohol, coffee, and tea and the teaching of the LOS C~urch of strict moral standards regarding premarital and extramarital sexual relation- ships. Although adherence to these mores may explain their lower rates for cancers at some sites (e.g., lung, bladder; esophagus, larynx, mouth, and uterine cervix), how much of the lower incidence is due to the a~ence oi these factors is not clear because the degree of their adherence to these proscriptions has never been curately measured. Furthermore, these factors have not been associated with other cancers for which incidence ~ rates are also lower among the Mormons (e.g., colon and breast cancers) and, therefore, they would not account for the reduced incidence for these sites in the Mormon population. We are unable to calculate age-adjusted incidence rates of cancer by religious activity in Utah because we have no information on age-specific levels of religious activity. However, acdve participation may be a good indicator of adherence to Church proscriptions and, if so, the incidence of cancer for some sites may be lower for Mormons involved in Church activities. .In this paper, we present data comparing by sex the Mormons with the non-Mormons in the general popu- lation of Utah with regard to m~ny potential risk factors in cancer etioloay. In addition, we analyzed those same risk factors in terms of religious activity. Our purpose is to describe differences between the Mormons and persons not of that faith and from these differences to identify factors that may explain the lower cancer incidence rates among the Mormons in Utah. Also, we studied differences benveen the active and inactive Mormons to determine if activity is related to adherence to Church proscriptions and to see if these findings help to explain the lower incidence of cancer in this religious group in Utah. METHODS Data for each demographic variable and possible risk factors presented were measured as par~ of ongoing case-control studies of cancers of the colon, cervix, and ovary. Information from a bladder cancer study will be added later. The information reported is based on the control group of each study. These controls were selected by an RDD technique that assures every resident of the study area who has a telephone an equal probability of inclusion. Controls were always AssttXVtATtO~s t.'SgD: LDS Church ='Churrh of Jesus christ of Latter,- Saints: RDD==random-digit dialing; SEEK=Surveillance, Epi- demiology, and End R~ult~ Program: SDA=Seventh.Day Advendsu: VD m venereal disesse. s Ps~-sented at the Wo)kshop on Populations at Low R/sk of C.~n- car ~ a~ Sn~bi~. Utah. Au~ = Divisi~ of Epid~iolo~. ~r~t of Fami~ and~mmuni~ M~cine, Uniw~ity of U~h ~H~ of M~ine. ~lt ~ke City. j.N~. VOL 65. .~O. 5. NOVEMBER T109371675

10S4 Wes~ Lym~ ~d Ga~er qttestioned by w~ined i .ntfrviewe~ who used a stan- d~zed schedule, and the/r completion rates ranged between 85 and 92Y,. A relatively new te~-hnique for drawing a random sample of controls for case-control studies, RDD is an inexpensive method because telephones have become so common in the United States. Approximately 95~ of the households in Utah have tdephone$.J This tech- nique beg/ns by a random generation of telephone numbers that are then called by interv/ewers, who are instructed to determine if the number is a working one and, if so, if it is in a private household. An age and .sex census is taken and an address is recorded for each household. (We obtained the census and address from about 90% of the households.) Once the census infor- mat/on is recorded, a sample is drawn randomly within each age and sex stratum. People selected are sent letters informing them of their selection, and these lette~ are followed by telephone calls to make appoint- ments for the interviews. The major shortcoming of this method, superior to other telephone methods because unlisted numbers have an equal chance of being called, is that people who do not have a telephone or who are not considered part of a household in which a telephone is located are not included. However, results obtained by RDD in an Ohio study (3) were comparable to results obtained by other nontelephone methods. The colon cancer study was a State-wide one con- ducted between November 1977 and February 1979. The controls were age and sex stratified, with ages dsawn in almost equal proportions for each 5-year age category and sex equaling the portion of cancer pa- tients who were male and female. A to~al of 61~ controls was interviewed. The cervical cancer study included controls drawn only from the urban areas of the State. The three standard metropolitan statistical areas in Utah, which have been defined as urban, comprise about 80% of the State's population. Between September 1977 and Febru- ~ E~dmate war made by the Motmtain Bell Telephone Company. ary 1979, 505 control respondents ~ interviewed, and the sample was m-adfied by ages of the padems. lk-~ween Jant~ry I976 and December 1977, 109 women were interviewed and their ages were stratified to those o£ ovarian cancer patients. These controls were also drawn from the urban areas of the State. In this paper, if more than one study is reported f~r each variable, the percentages or means are averaged with the use of the weighted rumple size for each study, the re~uhs of which are presemed separately. Religion was always measured by each respondent's indication of his or her religious preference. These responses are accepted at face value because confidemiality require- ments have not allowed us to check names against LDS Church records. We measured religious activity by asking each respondent which of the following state- ments best described his or her attendance at Church services or meetings: 1) never attend, 2) attend less than once a month, 3) attend once or twice a month, 4) attend once a week, and 5) attend two or more times a week. Persons are considered "not active" if they choose category l or 2, "partially active" if they choose category 3 or 4, and "active" if they choose category 5. Data on religious act/vity were collected only in the colon and cervix studies. We discuss religious activity here only for the Mormons because the number of peuple of other faiths was too small to consider them by frequency of participation. Inasmuch as the ages in each study were stratified differently, the age categories had to be adjusted to reflect those of the State. This adjustment was done by a weighting procedure summarized in table 1. The weighting facwrs were lowest for the colon control group and highest for the sample of ovarian controls. RESULTS AND DISCUSSION Each of the tables summarizes risk factors considered by sex, religion, and degree of religious activity among the Mormons. Sample size.~Table 2 includes a breakdown of the weighted samples in table 1 by religion and religious activity. The sample sizes for the partially active Utah Colon study Cervix study Ovary study m, n. ~*' Weight ~' ~t, Weight Age. yr Population Percent aeh~l Weight weight~i aetual weighted tetu~i weighted rumple factor ~mple ~ample factor sample ~arnple faetor ~ample ~ 1~9.961 27.2 84 &86 492 1~ 1.2S 1~ 6 1~17 ~ ~ ~0 11.~ ~ ~ ~ ~ 1.76 ~ 8 11.6~ ~ ~9 ~1 1L0 ~ 4.15 1~ 17 8,~ ~ 18 1.~ ~ ~ ~4~2 I0~ 51 &59 1~ 19 3.11 59 9 &~ ~ ~ 4~6 8.6 ~ Z~ 1~ 9 5.~ ~ 28 1.17 ~ ~ 84,~9 7.3 ~ 1.~ 1~ ~ ~ ~ ~ 1.~ ~ ~ 27.~0 &7 ~ L48 1~ 5 6.~ ~ 8 ~ 18 7~ ~1 7~ I3~ L~ 1~ ~ 3~1 ~ 16 1.~ ~ Ceml 4~ 100.0 61S 1~6 ~ 579 109 314 JNCI. YOU ~. NO. T109371676

108S No.. wei~hr.zd ~mple Colon ~'~6 616 ~s~n Pe~en~ ~i~h~ sample Colon ~ 72 49 30 21 ~arizn Ave~ 28 72 49 ~ 21 Age. ~ 2~ CoI~ ~ 26 27 ~ 14 Cereal Avenge 32 26 27 32 14 ~lon ~ ~ 24 12 ~ Ce~l ~a~an Ave~ ~ ~ ~ 12 30 ~lon 19 ~ 17 Ce~l ~arian Avem~ 19 21 17 25 Colon 14 19 21 14 Ce~l Avenge 14 19 21 14 ~lon 12 12 11 17 7 ~cal ~n A~n~ 12 12 11 17 ~o~o~. %. ~lon 21 99 ~unt~ ~ ane~t~ Uni~ Kin~. %. ~ ~ 67 ~ 70 col~ S~ ~ ~on 9 16 G~, %, ~ion 13 11 12 I3 5 O~er, ~, ~lon 21 4 7 3 Colon ~ ~ 36 39 Ave~ ~ ~ ~ ~ 41 PI~ li~ p~or ~ ~ Colon ~6 ~ Cereal~ A~ 16 ~ ~ 41 ~Io~ ~ 4~ 40 Ce~] ' Ave~ 44 ~ 40 ~e cir. ~ ~lon ~ ~ 32 16 • ~ 40 ~ 32 215 7~9 392 174 173 115 477 291 67 119 80 234 -- -- -- 23 77 53 24" 23 19 81 61 14 25 22 78 56 21 24 18 27 3O ~2 24 28 28 14 41 24 24 35 18 80 27 26 20 32 18 3O 18 19 27 28 18 29 22 22 26 18 23 14~ 24 16 18 13 19 19 17 17 18 15 14 16 I3 13 1~ 5~ I4 18 -- ~ 14 1~ 11 12 93 99 98 78 62 62 70 19 18 9"2 18 15 15 7 20 4 4 0 7 41 41 38 48 89 38 38 27 21 22 44 32 21 2O 22 18 22 24 21 20 22 10 16 8 19 12 17 10 17 14 12 38 14 31 17 22 22 29 32 28 23 15 19 21 ~0 13 22 20 2S 27 26 19 45 36 43 42 52 36 26 47 49 85 48 45 83 4~. 45 47 41 38 42 29 27 21 41 59 ~ 31 45 39 83 48 31 29 28 40 • D,~ = no data ~ere available frvm the study. s Number h <10. ~ The cervix sample was drawn only from urban areas. .~CL VOL ~. .~0. 3. NOVEMBEK T109371677

~ Dazkae=no data were available from the study. Number is Mormons in the cervical cancer study and the non- Mormons in the ovarian cancer study were smaller than desired; therefore, when breaking down risk factors into several strata, one must be careful not to overinterpret data based on small numbers. Religfon.~About 72~ of the men and 77% of t~e women reported that they were members of the LDS Church. According to calculations made from census and LDS Church data, 72~ of the State's population are Mormons, but only 6~Pg are Mormons when the proportion of adults is considered. Therefore, religious affiliation may be overreported in the three studies presented here. Inasmuch as many who overreport may be non- Mormon spouses of Mormons, overrepordng may result in conservative measurements of Mormon-non-Mor- mon differences in risk factors and the incidence of the disease. This needs further study. Religious act/~ity.~Litde difference exists between Mormon men and women with regard to religious activity. Ahhou~h the rates for members of other denominations are not reported in table 2, 89% of these men and 55% of the women were not active (9 and $3~g partly active and 2 and 12% active, respectively). ~/ge.--Ahhough age was weighted to the Utah population, it was not done so by religion or religious activity. As seen in table 2, there were a few more non- Mormons in the youngest age categotT, and the other JNCL YOL 65. NO, 5. NOVEM~F-R categories were about equal. No real differences were apparent when groups were considered for their re- ligious activity. However, in the cervical cancer study, the part/ally active were much younger than were the other groups. This finding is important to bear in mind because some of the variables reported were measured only in this one case-control study. Spot~es's religion.~Only 1 and 7~ of the Mormon men and women, respectively, have spouses who he- long to other denominations, whereas a corresponding 21 and IRg of non-Mormon men and women chose to marry Mormons. The high percentage of Mormons marrying within the Church is expected because this is a basic teaching of their faith, and because over 70~; o| potential marriage partners are of the same religion. The pre- dominance of Mormons in the area also.-explains why many non-Mormons marr~ re.embers of the Mormon religion. Although those Mormons may not be ac- Live in the Church, at least the potential exists that some of the Mormon way of life is adopted by these non-Mormon spouses. Thus non-Mormons in Umt~ may be more like Mormons in some social customs than non-Mormons outside oI Utah. This situa.don may result in conservative measurements of Mormon- non-Mormon differences in many risk factors and the incidences of diseases. Finally, because inacdve Mormons are more Hkely to TI09371678

conw-,~-t a m/x~l nmrsiafe than those who are active, more non-Mormo¢~ life-style ir~quenc~ would be noted Coumry o/an¢estry.mMormon men are more likely than non-Mormon men m come from English and Scandinavian backgrounds, This is particularly true for inactive Mon-nons" though the differences by activity are riot great. Mormon women are also more likely to be from Northern European backgrounds, and those who are pare'ally active contribute the highest per- centage. The data here suggest a more homogeneous Mormon population than is found among the non-Mormons. The large percentage of Northern Europeans among the Mormons may partially explain their high inci- dence of skin and lip cancers. The 21~ of non-Mormon males in the "other" category consists of 95 with Spanish ancestry, 6% from Slavic countries, and 6~ from France or Italy. For non- Mormon women in the other category, 5% are of Spanish descent, 59g French. and 75 Italian. Inasmuch as 9~t of the male and 5~ of the female non-Mormon population in Utah is Spanish, one would expect the non-Mormon rates to be lower than in other areas of the United States because findings from New Mexico show that this group has lower cancer rates for most sites (4). Adjusting for ancesLry may widen the Mor- mononon-Mormon differences in incidence for some sites because the Spanish-Amerlcan group constitutes a substantial percentage of the non-Mormons. Residence.~Mormons have lived in the State of Utah longer than non-Mormons. Until recently, Utah has been relatively free of environmental pollutants and contaminants from large industries. Having lived in the State longer, Mormons may have had less exposure to certain environmental risk factors for cancer. Little difference in their length of residence by religious activity has been observed. Prior to 18 years of age and afterward, Mormon men are more likely to live on farms and Mormon women are more likely to live on farms or in small towns. The fact that cancer rates for most sites are higher in urban areas may explain some of the differences in cancer incidence between Mormons and non*Mormons. If one compares the question measuring residence before and after the age of !8 years, a migration pattern from farms and small communities into lar~er cities is observed; this is true for both sexes and religious groups, though the shift seems greate~ for the Mor- mons. Such a shift suggests that cancer rates for some sites, particularly among the Mormons, should increase over time. From another study (5), however, we have observed that the urban-rural rates among the Mor- mons are similar when compared with those of the non-Mormons who demonstrate the expected urban- rural gradient. In other words~ if migration to urban areas has occurred, it has not altered the Mormon rams as much as one might expect. Many urhan Mormons may live rural life-styles which they brought with them. For example, they may maintain cohesive rela- tiomhips, extended kimlxip netw~, and rural dietary and exercise cus¢oms. With regard to religious activity, foc residence beCaTre and after age 18 .years" the least active Mormon men and women were least likely to live on farms, whereas the partially active group was least likely to live in large dries, When the place of residence was compared before and after ag~ 18, the greatest shift to large dries occurred among the partially active. Edueation.--M~men men attain highe~ educational levels than do non-Mormon men, whereas the greatest educational differences for women are noted among the college, educated. Non-Mormon women are more likely m graduate and/or receive professional training, which may be attributed to the fact that Mormon women are younger than their counterparts when they marry. Active Mormons receive more schooling than do in- active members. Exactly what dfects education may have on cancer etiology is not dear, but it may rdlect life-styles, health habits and/or awareness, and other behavior that needs to be identified. Occupation.--In the colon cancer study, occupations were coded according to the Alphabetical Index of Occupation and Industries from the 1970 census (6) on a scale from 1 to 12, with the most prestigious occupations and those requiring the most training assigned a "1" and the unskilled occupations assigned a "12." Mormon men and women have only slightly higher ratings than non-Mormons (table 2). Inasmuch as Mormon men have spent more years in school than have other males" one would have expected the occupa- tional differences to have been greater, why they are not is unclear. With respect to religious activity, the occupations of active Mormons are in the higher ranked range, and the other two groups are similar. Thus active Mormons usually have professional and white collar jobs, which would give them less chance to be exposed to many occupationally related risk factors for cancer. Occupational exposures.tThe above conclusion is substantiated in table 3: Active Mormon men had less exposure to every occupational hazard measured except excessive heal Active Mormon women had a similar pattern, but the numbers were so small interpretation was difficult. In the Mormon-non-Mormon comparison in table 3, the non-Mormon men had greater exposure to every hazard except excessive heat. There was lfittle difference among the women except that the non-Mormons were exposed to more dust and heal The conclusion one draws from table ~ is that non- Mormons and inactive members of the ~ Church are exposed to occupational hazards that may he associated with certain types of cancer more frequently than are Mormons and active partidpants, This may partly explain a few of the differences in inddenoe for cancers of some sites. Smoking.--The non-Mormons reported more ever- smoked and current smoking behavior (table 4). Their 35% current smoking rate is similar to a 38~ rate JNCL VOL. ~, NO. 5 NOVEM~tgR I~ T109371679

10e~ • Number is <10. reported by the National Clearing House for males in the United States over age 21 (7). We were surprised to find that 41~ of non-Mormon women were current smokers because the National Clearing House reported about 28~. Preliminary analysis from our bladder cancer study shows lower smoking rates for women. Although more non-Mormon women smoked than did their male cohorts, they started smoking at a later age, smoked for fewer years, and smoked fewer cigarettes per day. Not only do fewer Mormon men smoke, but they also start at a later age, have smoked for fewer years, and smoke fewer cigarettes per day than do non- Mormon males. Mormon women have the same pattern except that the number of cigarettes smoked is about the ~ame as that of other women. Thus the Mormons who do smoke have different patterns, and any analysis of data that only involves current smokers may over- state the contribution of smoking among Mormons to smoking-related cancer sites. The number of Mormon women who smoke is too small [or any analysis of smoking by activity, but, among males, Church activity is directly related to smoking behavior. Active Mormons are less likely to smoke, they begin smoking later in life, and smoke for fewer years and fewer cigarettes per day. Finally, fewer of the inactive members smoke than do. the non- Mormons, but those who are inactive and smoke have smoked more years and smoke more cigarettes per day. Coffee and tea.~The only data we currently have on coffee and tea consumption are from the cervical cancer study and have little detail concerning the drinking of these beverages. From these data, however, we deter- mined that Mormon women consume much less of both drinks than do non-Mormons, and the active members drink less than the inactive. Wide use of coffee, in particular is surprising bozause it* consumption is against Church teachings. If coffee and tea are etiologic factors for some cancers, less use among the Mormons may explain some of the Mormon-non-Mormon dif- ferences in cancer incidence rates for those neoplasms. A lcohol.--The difference in the number of alcohol drinkers between Mormon and non-Mormon women is about threefold less for the former group for each type of alcoholic beverage. For men, only those who drank liquor were identified, but Mormons had about one-. fourth as many drinkers as non-Mormons. The lower number of alcohol users among Mormons should explain most Mormon-non-Mormon differences for alcohol-related cancer sites, such as the esophagus and larynx. Even though the numbers are small, drinking of .alcohol is also related to amount of religiom participation; active Mormons consume less than do the inactive. Dietary patterm.~Most of the dietary data from the colon cancer study have not been analyzed, but a few Mormon-non-Mormon comparisons are reported in table 5. Respondent* were asked which types of fat they used for cooking at least once a week. Mormons used nearlx all tX.oes of fat more often in their cQokin~. measured only the types, not the amount, of fat consumed. However, those who used more types sibly consumed larger amounts as well. This pattern in cooking may reflect the rural background of many Mormons in that they continue to eat more home- prepared meals than do other groups. One clear pattern revealed that religiously active Mormon men used a great variety of fat. Mormon men and women eat more natural and fewer "ethnic" foods, and few will skip meals. Active Mormons follow this same pattern when they are compared with the less active. These data suggest a hypothesis in need of testing: Fat may be a risk factor for cancer (e.g., colon and breast) only in the absence of other factors tl~-~t are protective, such as natural or fibrous foods~ This phenomenon, may explain the low rates of colon and breast cancers among the Mormons. Marital hi~ten~y.--Mormon men and women were more likely .to be married than single, divorced, or ~parated than were other men and women (table 6). Rates [or the widowed did not differ between Mormons and non-Mormons. Those members of the LDS Church who were inactive were less likely to be married and. more f~uently divorced or separated than were active members. Men who were not active also tended to be single and widowed. Cancer incidence for many sites is lower among married people {8). Why this occurs is not c|ear, but it may be related to support networks, lowered stress, and T109371680

Fe.~e Speeif~t~on N~,- Mermo~ 1~- Mormo~ Wta[ T(~al Ac~ve P,.rtly Insetive totaJ To~al Active Pa~ly Inactive ~tive ~tive Tob~eo Current smokers. % ~a~n ~ 7 Av~ ~ 10 1 1~ ~ 4I 7 1 4 S~ at I~t 1 ~, ~ 7S 40 ~ 44 71 colon Colon ~ z~ smoking, ~ ~16 ~ 11 i0" 1" 19 10 6 >81 O" 4 O" 5" 6" 8 8 O" 10 10" ~ a~ ~ ~ 19 21 19 ~9 ~ 17 2" ~ ~ 11 11 • yem 21 18 9 1~ ~ 17 13 10" ? 16 Ci;a~t~s smok~d~ (I0 3 4 II" 2" 0" I~19 ~ 15 14" 24 I0~ ~59 ~ ~ 13" 16 40" ~ 8 4 10" 0" 1" 4 ~ No, ~ ci~t~ 26 ~ ~ 20 ~ ~ ~ ~* ~ 19 Coff~ Dnnk befo~ a~ 30 78 ~ 13 39 69 Dnnk ~Mr ~e 30 ~ 41 ~ 66 79 yr, % Tea yr. ~ ~ol ~n~ ~nk ~ 16 6 12 ~ ~ 17 1 ~ 48 liquor. % ~nk Hquor ~ ~ 7 4 3" 16 a~,~ D~nk Hq~r sin~ Drank ~r ~fo~ ~ ~0~.~ ~k ~r since a~ D~nk wine ~fo~ Drank ~ne sinee ~ 47 15 10 Yumber is <10. Values in this horizontal line on~ are ~or colon. T109371681

10~0 ~ of cooking fat Oli~ oil 8" 1" 2" ~ O" 6 2 ~ 3" 1" Ma~ 70 79 S1 81 69 70 78 ~ 73 ~ ~ le oil ~ ~ 76 59 ~ ~ 68 69 75 ~ ~ble aborning ~ 70 69 ~ 67 ~ 68 75 57 ~ * ~s~l ~s 8 16 ~ 17 B 13 24 ~ 16 10 Me~ ~k. ~d 62 ~ 53 28 40 71 47 ~ ~ 52 Imlhn f~s Pol~ian, ~Ine~. ~d ~ ~ 81 19 ~ 86 ~ ~ 21 47 ~#r O~en~ f~ He~8 ~ 47 ~ .~ ~ 70 ~ ~ ~ 61 S~p ~a~ ~ 46 ~ 4~ 68 47 35 ~ 45 ~ • Iguml~r is <10. other life-style differences among the married. The higher marriage rate among the Mormons may account for some of their lower cancer incidence also. Cancer sites related to rn~rital status need to be determined, and behaviors asso<iated with marriage that reduce the incidence o! cancer should be identified. Mormon women marry at a slightly younger age than do non-Mormon women (table 6), but no di[. ference occurs among men. Mormons who are not active also marry at earlier ages. Although early age at man'/age has been identified as a risk factor for cervical cancer (9-13), our data would not support this hy- pothesis. Finally, table 6 shows that more Mormons have been married only once compared with other men and women. Inactive Mormons are also likely to have been married more than once. The literature has shown that the number of marriages is related to cervical cancer (9-13). Our data support this hypothesis in that the incidence of cervical cancer is lower among Mormons and so is the number of second mm~iages. Sexual activity.mAt present, we only have data on sexual activity for females; age at first intercourse and number of sexual parmers are shown in table 7. Slight differences in age at first intercourse occur between both groups of women. Non-Mormon women are a liRIe more likely to begin sexual activity when they are 17 or 18 years old, but all other categories show little .difference. Although this finding has been considered an important risk factor in some cervical cancer stud/es (9, I0, 12, I3), our data do not corroborate it. There are some differences among Mormon women by religious activity, the nonactive begin sexual activit.v earliest, and the active women begin the latest of the $ groups. A large difference was observed between Mormon and ]NO. VOL 65. .%'0. 5 NOYF-.MIEK 19~ non-Mormon women when we considered the number of sexual partners. We found that people who are not members of the LDS Church have much higher rates of multiple partners, and nonactive Mormon women also have many more sexual partners than the active or part/ally active. Inasmuch as the number of sexual partners has also been identified as a risk factor (9, 13) in cervical cancer, part of the lower incidence of this type of neoplasm among Mormon women may be attributed to their sexual behavior. Pregnancy h/story.~Table 7 shows that Mormon women had more pregnancies as measured by our studies on cancer of the cervix and ovary. The sig- nificance of this, if any, for female cancers needs to be determined. With regard to Church activity, the par- tially active Mormons report the fewest number of pregnancies. Age at first pregnancy was also measured, with the non-Mormons showing a slightly higher percentage in the 19~ to 20-year-old ~up and the Mormons showing a slightly higher percentage in most of the older age groups. The inactive Mormons are youngest at first pregnancy and the active are the oldest. Overall then, women of other ~aiths and inactive Mormons may start their reproductive history slightly earlier than practic- ing Mormons. A/though age at first pregnancy has been identified as a risk factor for breast cancer, with older women having the highest risk (I¢), this finding is not supported by our data nor does this variable explain any of the lower incidence of breast cancer among Mormon women. Birth control pills.~Non.Mormon women are more Likely to have used birth control pills and to have used them longer than do women who are members of this religious group. However. both began using them at T109371682

~ Rl~k Factor~ Momton~ a~d ~mm~ 1091 abo~ the same age (table 7). Inactive ,'~orm~m k, ma~ are more likely to use birth cow~ol pills and to have reed flaem long, r, whereas Lhe partially active began using them at an earlier age ~han did the other groups. Although we do no~ know exactly what effects birth control piIh have on the devdopraem of cancer, we do recognize tha~ Mormon women should be less suscepti- ble to them. Some invesdgawrs (1~, I5) have reported uhat birth comroI pills may be pro~:tive gor breast cancer, but dais hypothesis would not be supported by our data, When one considers the Mormon-non-Mormon rates depicted in table 7, one notes that the rates are less for effect may be operating ~uch as t, bose women okier at h~ d~ ~ ~ int~ ~& the~ore, less likely to have m~ bir~ ~n~! pillg Oth~ s~l f~t~s.~M~t of ~ ~ ~me ~m • e stay on ~n~ of ~e o~, and ~li~ a~fivity was n~ ~s~. The n~ ~e of~n small, so ~mion ~ wa~nt~ h thdr inte~emdon. Howler, th~ fin~ ~e s~sdve o[ variables n~ng ~- talnment in Iar~r studieg A high ~te o[ ~o~a~s is not~ among non- Moron women. The eff~ o~ ~ff~, alcohol, &u~, and other hctors on mls~ges should TA,IZ 6.---Ma~tal ~ amon~ control groups in Uto~ ~ ~¢ro rel~ion, and rel~io~z actlvitll Marital atatua Male Female Non- Mormons Non- Mormons Mo~rno~ Moemons= total Total Active Partly Inactive total Total A~tive Partly Inactive Single Colon I0 2 Cor'v'le~*l 4 0 Ovarian 18 6 Av,m~ 10 2 2 0 4 8 2 1 2 2 Colon 81 ~ 76 92 ~ ~ 9I ~a~an 69 82 A~ 81 ~ ~ 99 76 73 ~ 86 77 80 Di~~ Co[~ 5 2 0" 0* 6" 14 9 8 5" 15 ~a~an 4 Avem~ ~idow~ Col~ 4 4 1" 1' 13 7 12 11 17 9 Ce~i~! 6 4 4 14 0~an 9 11 Ave~ 4 4 1 1 13 . 7 9 8 16 6 ~ tint ~. <18 ~Ion 0 1 0* 1" ~" 9 13 3 19 29 Ce~i~l 9 Aven~ 0 1 1~19 Colon ~ 12 6 14 18 26 ~ ~ ~ 41 Ce~i~l Avenge Colon Ce~! 49 ~ ~ ~ 31 Avem~ ~ ~ ~0 ~ ~ 50 41 ~ 46 26 Colon Ce~i~i 11 l0 12 6 1 Av~ 21 21 ~ 17 ~ 13 7 9 9 3 • ~ a~ tint ma~a~ Colon 28 ~ ~ ~ ~ 21 ~ ~ ~ 19 Ce~i~ 20 21 21 20 20 Avem~ ~ ~ ~ ~ ~ 21 ~ 20 21 19 Ma~ me~ than once. Colon Ce~I 2~ Ave~ 87 I~ 1S 8 21 25 ~ ~ 21 31 • ~umber is <I0. ,~NCI. I,T~L ~. NO. 5. NOVE~BE~. te~O T109371683

West, Lyon, and Gardner So~o Reproducti~h~tory ~Iarmor~ total Total Active Partly active Inactive Age at first intercourse, yr <17 Cervical 11 11 5 9 25 17-18 Cervical 39 29 23 42 38 Cervical 25 27 27 27 26 21-22 Cervical 8" 9 12 8 3 23-24 Cervical $" 13 18 6 >~-5 Cervical 9 9 14 8 0 ~ age at first intercouree Cer,dcal 20 20 21 19 18 Percent with intercourse with more than one partner Cervical 60 24 16 14 51 No. of pregnancies None Cervical 10 5 5 6 Ovarian 2" I" Average 7 4 5 6 1-2 Cervical 47 17 12 37 19 Ovarian 40 21 Average 44 18 12 37 19 3-4 Cervical 34 37 38 28 40 Ovarian 26 36 Average 31 37 38 28 40 Cer~ica| 4" 22 28 16 13 Ovarian 23 31 Average 12 23 28 16 13 ;>7 Cervical 5" 18 18 12 22 Ovarian 8" 10 Average 6 15 18 12 22 • No. of pregnancies Cervical 2.5 4.2 4.5 3.4 Ovarian 3.4 4.0 Average 2.9 4.1 4.5 3A 4.3 Age at tint pregnancy, yr 19 Cervical 13 16 6 19 38 Ovarian 18 12 Average 15 15 6 19 38 19-20 Cervical 24 24 23 38 21 Ovarian 61 32 Average 39 27 23 38 21 21-22 Cervical 16 2]~ 25 18 13 Ovarian 9 Average 13 24 25 18 13 23-24 Cervical 13 16 19 7 15 Ovarian 4 11 Average 9 .14 19 7 15 25-29 Cervical 34 19 22 18 12 Ovarian 4 13 Average 22 17 22 18 12 Number is ~10. j.xo. vo~_ ~. XO. 3. .XOV~-XmER IS~ TI09371

Cancer Risk Factorz: Mormons and Non-Mormorm 1(~93 total Tc~a] Active Partly In,rive active Cer~I 1" 4 Ovarian 4" ,4.ve~a~ 2 4 ~a~e a~ flint p~anoy. ~ ~1 ~.5 ~.0 ~;an 20.3 21.6 Avenge 21.6 Ever ~ bi~h ~ntrol pills, % ~ezl 72 ~ar~an ~ Aven~ ~ A~ ~an using bi~h con~ol pills. ~ Ce~,! 25 Ovarian 0 Ave~ 15 ~21 ~I 17 ~an 24 Ave~ ~ 18 ~-24 Ovar~n ~ 14 Ave~ 16 Ce~l 14 19 O~an ~ Ave~ ~ 15 ~I 16 18 ~an 9 ~an 5 Ave~ 11 ~a~. ~ Ce~l ~ Ovarian ~ Ave~ ~ Monks ~ bi~ cont~l pills. tint ~ Ce~cal 5" 7-12 Ce~l 8" 16 1~24 Ce~ical 24 19 ~a~nz mima~ag~ or s~illb~s. %, o~an ~ 13 Frequency of b~ut e~mination. % ~one Eve~ monzh ~rian 15 48 Morn than once a y~r. but not eve~ m~ ~aria~ 61 ~" ~ ~an once a year Ova~n I~ 6 Ha~ng a brmt X-~y, %, ov~an 6 21 6 0" 6 0 1 23 21 21 23 21 21 31 24 32 36 ~ 43 14 24 13 14 24 13 22 29 14 22 29 14 13 16 17 13 15 17 18 15 0.3 I8 15 23 15 9 28 15 9 28 17 9 6 17 9 6 26 21 26 26 21 26 43 33 33 21 15 6 19 21 17 17 30 44 Number is <I0. ,~.NC]. VOL. 63. .NO. 5. NOVKM~ER Tl09371685

Lyon, and Gardner Ag~ at rnenop~use, yr Ovarian 64 18 41-48 Ovsrisn $ 16 46-~0 Ovarian 19 26 O~ri~n i0 41 Ovarian 38 47 Treated for VD, %, cerdca] 3 5 Htmbgnd~ tr~a~d for VD. %, cerv/- ezl 4 4 2" 8" 11 • Number is <10. studied as well as the high hysterectomy rate of this same group. One may speculate from these data that Mormon babies may be healthier at birth. This specu- lation has some support because the proportion of Mormon babies who weigh less than 2,500 g is less than expected (17). Reasons for this phenomenon, whether they be genetic or life-style, need to be investigated. With the high rate of hysterectomies shown for the non-Mormons, the fact that their age at menopause is also earlier is not surprising. When asked about breast sel~-examination, about 7?% of all the women indicated that they examine their breasts. Mormon women, however, do so more fre- quently; three times as many of them as non-Mormon women examine their breasts at least monthly. This examination may result in earlier detection of breast cancer and may reflect a greater awareness not only of cancer symptoms but also of health care in general. Mormon women were also about three times more likely to have had a breast X-ray. This, too, may lead to earlier detection but may at the same time put some women at greater cancer risk due to X-ray exposure. Finally, although the numbers are small, both groups showed little difference in the amount of VD reported. However, the inactive Mormons reported more VD for themselves and their spouses. CONCLUSIONS These data were presented as suggestions for the Mormon-non-Mormon differences in cancer incidence in Utah. What some of the data mean is not entirely clear, but some conclusions can be drawn: 1) Religious activity has a strong relationship with many risk variables. Although the patterns are not always consistent, when Mormon-non-Mormon risk [actor differences exist, inactive Mormons have more in common with the non-Mormons than they do with the .~NCI. VOL. ~. NO. ~. NOVEMBER partially active and active groups of their faith. In any comparison of incidence rates of Mormons with those of other religious groups, such as SDA, it would be accurate to exclude the inactive Mormons because the SDA Church eliminates inactive members from their rosters. 2) Differences between Mormons and non-Mormons in urban-rural residence should be controlled in any Mormon-non-Mormon analyses. Also, more investiga- tions regarding urban-rural life-styles and relation- ships to cancer are to be promoted. ~) Non-Mormons in Utah may have more exposure than the Mormons to environmental and occupational hazards that may predispose them to cancers of some sites with greater frequency. 'D Smoking of tobacco and alcohol, coffee, and tea consumption are lower among the Mormons. This no doubt explains much of the ~ormon-non-Mormon difference in incidence for cancer of some sites, such as oral pharynx, larynx, lung, esophagus, and bladder, but additional research is required on the effects these agents have on cancer of other sites as well, 5) Not only do the Mormons smoke less than others, but also their patterns of smoking are different. The effect that these different patterns have on cancer incidence needs to be studied. 6) Although only a cursory look has been given to dietary customs, major differences may exist between Mormons and non-Mormons. A study of eating habits may be fruitful in providing an explanation of the lower incidence rates for Mormons for cancers of the breast and colon. 7) Marital status differs by religion. This is another area in which more study is needed for us to determine how marital status may affect cancer incidence. 8) Differences in sexual behavior for each religious group have been found, some of which support current hypotheses (e.g., number of sexual partners and mar- riages as related to cervical cancer) but not others (e.g., age at first marriag~ and ag~ at first intercourse T[09371686

related to cen~cal canc¢~ and age at first pregnancy related m breast 9) Some o[ our ~'iden~ in~es hyst~cto~ and mis~a~s ~e higher among non- Moron women, but this n~ s'~i~on. fining is v~fied, more work is n~ssa~ to det~in¢ why thee high in~denc~ ~r. 10) Morons ~ be more aware o[ g~d h~lth ~c~, but this possibility n~ to ~ d~ment~ by additional stu~es. Some o[ the m~ ~fferen~s ~at ~t be~n Moron popula~on in Utah and ~ople no~ reli~on explain ~he low~ ~n~r in~den~ ra~ • e Morons. Accor~ng to cu~ent hypoth~s, some • fferences would sugg~t that the~ higher [or some sites, bm the~e assumptions call [or r~mi~ion. Finally, al~ough what some o[ the • fferences m~n [or ~r in~n~ are not dear, they ~ provide ar~s for f~itful re~rch in Utah. REFERENCES (I) YOb.'~G JL. Aslx~r AJ. POLLA(:X ES. ~. S~R Pr~: ~n~r in~e a~ mormliw in ~¢ Uni~ Sm~ 197~1~6. Wash- inston, D.~: U.S. ~ o[ H~I~. ~u~tion. ~d Welh~. 1977. (2) LVO~ ~, Gs~ JW, W~ DW. ~x~ induce in Mor- mons and non-Morons in Utah d~ng 1~7-75. JNCI 19~ ~:1055-1~1. (~) TU~rA~a AJ, ~ WR. Ra~t ~iin~ ~we~ng • e cost ~ vic~m~tion su~ Wmhin~n. D.~: Police Foun~on. 1976. (4) New Me~dco Tumor Regis,.~. Cance~ in New ~',.*exi~ 1972..~b:qu~=m New M~o Tum~ Re~s~ 1~. (5) L¢os JL G~n~zt JW, W~ DW. ~ in Uta~ ~k ~i~n an~ place of resinate. JN~ 1~0~:1C~-1071. (6) ~.5. ~ur~u ~[ ~e ~ns~. Alpha~i in~ of ir~u,~, and ~ons. 1970 c~su~ o[ the ~pnladon. W~hin~on. D.~: ff~. ~va P~nt Off, 1971. (7) U~. De~r~ent o[ H~hh. Ed~6~n, ~d WeIf~. ~ult u~ ~ tobacc~1975. W~hington, D.C: U.S. ~ Print 1976. (8) Natio~l Offi~ o[ Viral Statistic. Mo~lity [rom ~l~ ~u~ by ~mi s~tus. Unit~ Smt~ 1~9-1~L Viral H~lth Star [7] (9) T~a~ M. O~a~ MC. ~dn~a of the c~ ~ epid~ miol~c study. JA~ !~174:1847-1~!. (10) T~tls M, Winos F, S~ H, S~USG ~, ~s JH. d~iolo~ of ~n~r o[ ~e c~ix. V. The rela¢ion~ip of coitus to c~n~a ~ &e ce~x. Am J Public H~I~ 57:~47. (11) ~ ~. ~ R. A study o[ the aetioi~ of ~dnoma of • e c~x urea. Br J ~nc~ 1~;13:419~4. ¢~efis~ of a ~i ca~ ~puhtion. Am J Public Hml~ 1~7;57:815~. ~u~ BJ. ~i~! ~n~ in Yu~hvia. If. Epid~iol~c hao~ of ~ible etiol~ si~i~e. J Nad ~nc~ ~t 197~53:51~. (14) ~o~ B. ~ P, ~ ~. ~ al. ~ at lint binh a~ b~t ~ ~k. Bull WHO hr~st n~pi~h: A retentive stay. Br M~ J 1~3: 719-724. {16) No~u~ ~ CoM~ GW. ~ni~ breast tumor and estrogenic ho~: A ~pulation-~ ~os~ve stay. Am J Epidemiol {17) ~c ~ S~ E. ~a~efistics of blr~s, Unit~ Smt~. 1~3-1975. W~hlngmn. D.~: Natio~l Ont~ for H~I~ Sta- tistics. 1~8 [DHEW p~li~fion No. (PHS)78-1~]. JNCI. '.'OL. ~. NO. 5. .'~Oi'ILM~L~ I~ Tl09371687

Mortality Among Selected Cancer California Seventh-Day Adventists for Sites ,., Roland L. Phillips, M.D., Dr. P.H., 3 Lawrence Garflnkel, M,A., * J. W. Kuzma, Ph.D., W. Lawrence Beeson, M.S.P.H., ~ Terry Lolz, M.S.P.H., a and Burton Brin, M.P.H. ~ ABSTRACT~In pmvloue reports concemlng cancer among S~enth-DW Adv~ntlsts (SDA), comparisons were msda only wRh the general population. This report cornered California SDA to s sample Of non-SDA who were demographically similar to SDA. The study consisted of 17 years of follow-up (1~0-76) on 22,940 white California SDA and 13 years of fi:llow-up (1960-72) on 112,725 white California non-SDA. Both groups completed the same base-line questionnaire In 1960. Deaths were ascertained" by annual contacts with each study member end by computar-s, isted record linkage with the California State death certificate file. Results indicated that, with the exception of colon-rectal cancer and smoking-rel•tad cancer,J, the difference In risk of fatal cancer between SDA and non-SOA w~ substan- tially reduced when SDA were compared with • more soclo- economically similar population. The persistence of the low risk for co~on-rectal can~er can probably be attributed to some Mpect of the diet or Ilfe-/tyle of the SDA.~JNCI 65: 1097-1107, 19~0. The SDA are a conservative religious denomination that was founded in 186.~ and currently has about ~ million members worldwide, ~00,000 in North America and 120,000 in California. By Church proscription, virtually all SDA abstain from the use of tobacco and alcohol, and a large majority adhere to one or more of the recommendations of the Church regarding other health habits and practices that are advocated primarily for their established or supposed health-promoting effects. Presently, about 54% of SDA follow a vegetarian diet that includes milk" and eggs, and ,t1% rarely or never use caffeine-containing beverages. They also tend to abstain from sweets, other highly refined foods, hot condiments, and "spices. Regularity in vigorous exer- cise, adequate rest, and conservative sodal mores are strongly encouraged. Approximately one-half of SDA in the cancer age range (_~40 yr) are adult converts to the Church; others were either born into an SDA home or joined the Church prior to age 20 (usually concurrently with other immediate family members). Although most of these converts undergo a significant change in life-style at the time they join the Church, little is known about the characteristics of the typical person who chooses to convert to the religion. However, compared with the general population, a markedly higher proportion of SDA have a college education and/or a professional occupation. This strongl.v suggests that the prosely- tizing efforts of the Church attract persons of above- average socioeconomic status. These converts likely differ in many other ways from the average American prior to their conversion to the SDA Church. Table I summarizes published (I, 2) and unpub- lished data regarding risk of cancer death among California SDA. For most cancer sites, these data clearly indicate a significantly lower risk among SDA compared with that for the general California popula- r.ion. However. the selective factors that characterize persons in the general population who'convert to the SDA Church may also be associated with low risk of cancer. Thus the general population is probably not the most appropriate comparison group to use in the determination of whether SDA are a low-risk group for cancer. Furthermore, considerable caution is indi- cated before the inference can be drown that the typical SDA life-style adopted at conversion accounts for the lower risk of cancer in SDA compared with the risk in the general population. Indeed, measurements of can- cer risk in the segment of the general population who are similar m SDA converts prior to their conversion (except for life-style traits unique to SDA) might reveal a group with a low cancer risk equivalent to that of members. This ~eport will differ from all previous reports of cancer risk among SDA in that SDA are compared with a demographically similar ~egment of the general population. METHODS The SDA subjects for this study consisted of white respondents to the same four-page self-administered questionnaire collected by the ACS study from 1 million subjects throughout the United States (4). This questionnaire was disu'ibmed in 19~0 to the members of 198 of the ~51 SDA congregations in California. Volunteers who were identified in each congregation were responsible for obtaining a completed question- Aa~vg'no.~s user: ACS:= American Cancer Sociew; SDA=$evemh- Day Ad~mist(s); SMR~standardized mortality radons). ~ Presented at the Workshop on Populations at Low Risk of C~ncer conducted at Snowbird, Utah. August 23-25, 1978. s Supported in part b.v Public Health Ser~qce grants CA147~ and CAI8185 from the Nadonal Cancer Insdtu~ and by the American Cancer Society. s Department of Biosmistics and Epidemiology, School of Health, Loma Linda Unis~ersiw, L~ma Lin~a. Calif. • American Cancer Society. lac.. 777 Third Ave.. New York, t0017. 1097 .~'CI. voL. f.,5. NO. 5. .NOVEMBER 19~ T[09371688

Prima~ cancer~ si~e NO. of death~b S~[R#'c Male Female Hale Female Both sexes Lung, 7 9 Ii'" 46** 20** Facphagus 4 2 88" 30 34** Blzdder 4 6 20** 38** 28** ~dney 11 7 116 78 S~ma~ ~ ~ 69" 71 70" ~I~m 40 ~ 65** 70~ 68"* P~e~ 14 ~ 51~ 76 65 Gallbl~der 5 13 ~ 80 Liar 0 6 ~ 141 B~t ~ 91 ~ 72** Utedne cewix ~ 19 -- 51*~ Endome~ium -- 6 -- 68 Ov~ ~ 25 ~ 61" P~ gland 45 ~ 81 ~ ~n~l ne~ ~5 13 142 ~ 1~3 s~m (~ li~ant) ~ukemia 13 13 70 ~* 62* ~mphoma 19 30 ~ 1~ ~1 o~er ~ncer ~ 70 ~** 60** 47** All ~neer 242 ~ ~** 68** 61"* • SMRf(O:E)XI00, E=aceumulated No. of expected deaths obtained by application of the aft-sex cause-specific death rat~ for the total California population for each year to the corr~ spending 10-yr ~ group in the SDA population-at-risk at the beginning of each year and On.No. of observed deaths among SDA from 1958 to 1965. ' Dashe#fzero deaths or not applicable. ¢ *. P<0.0~; **, P<0.01. Both ar~ baaed on Bailar and Eder~r'~ method naire from a list of 10-20 individual adult (>30 yr) members of their congregation. These lists always consisted of all the adult SDA members of five spe- cified households. These volunteers also agreed to report annually on the living-dead status and the location of each person on their list who completed the ACS questionnaire. Every 2 years a brief follow-up questionnaire was also collected by these volunteers from each o[ the surviving questionnaire respondents. Although the procedure used to enroll and trace persons in the non-SDA group was similar, it differed in some respects. Volunteers recruited through local ACS officers were asked to enroll at least l0 families whom they knew well and expected to be in touch with over a period of years. These families included relatives, neighbors, friends, or others they knew from work or organizations to which they belonged. To be eligible, at least I member in the household had to be over 45 years, and all members over 30 years old were asked to complete the questionnaire. They were asked not to enroll itinerant workers, construction workers, or members of the armed forces. Some volunteers recruited only i or _9 families, whereas others recruited 20 or 30; the mean number of families was 16. The follow-up procedure described above served as the primary method of death ascertainment in the SDA JNCL VOL ~. NO. 5. NOVLMBER l-C~c3 populadon during 1960-65 and in the non-SDA popu- lation during lg~0-71. In both groups, special efforts were exerted to trace subjects whose living-tired status • /,'as not dearly determined by the volunteers. From 1966 through 1976, deaths in the SDA group were ascertained solely by a computer-assisted record- linkage procedure that linked the original question- naire record with the computerized file of all California death certificates for I~65.-76. A detailed description of the methods and resuhs of this record-linkage process will be reported elsewhere. This process involved a computer program that selected from the two files many pairs of records which were potent/ally matched. These potentially matched pairs had to match exactly on the first four letters of the last name, on sex, and on .any six of ten additional unchanging characteristics (six segments" of full name, three components of birth date, and state or country of birth). Pairs of records that matched on all ten variables were accepted as definhe matches without further processing. The nam.e, sex, birth date, ethnicity, and birthplace of all other pairs of potentially matched records were manually examined to identify additional matches. For pairs with remaining questions, the original questionnaire was compared with the death certificate to resolve the ambiguity. As a means of validating the record linkage, this procedure was also done for the 1960-65 period, which allowed a one-to-one comparison between deaths ascer- tained by the record-linkage procedure versus deaths that were previously ascertained by the more tradi- tional follow-up methods previously described. For 1960-65, record linkage ascertained 93.2% of the 2,011 California deaths in the SDA group that were previ- ously ascertained (table 2). Previously ascertained deaths that occurred outside of California (54) were obviously not reascertained by record linkage with the California death certificate file. Overall, record linkage reascer- rained 90.8% of all the previously verified deaths. However, record linkage ascertained 29 deaths that had been missed by the traditional follow-up methods. Only 7 subjects were falsely labeled as dead by the record-linkage procedure. These false linkages were detected by a meticulous comparison of the original questionnaire data with the death certificate for each death that was ascertained only by record linkage. The proportion of deaths missed by record linkage does not show substantial variation by sex, age, or interval. The pattern for cancer deaths is similar to that for all deaths. Furthermore, the previously ascertained deaths that occurred in California during 1960-65 did not differ substantially from the non-California deaths with regard to sex, age, education, marital status. smoking history, age at baptism, or cause of death of the decedent. Thus a reasonable assumption is that rlon-California deaths during 1965--76 that were missed by the record-linkage process do not differ substantially from the deaths ascertained by record iin'l'l~ge. To account for the underascenainment of deaths by record linkage, the counts of SDA deaths used to T[093716~9

F obtain cau~-six-dfic mop, alia- rotes were adjusted u~ ~rd for 19~76. This adjus~em ~k account of ~e 6.8~ o[ d~ths ~a~ are mi~ b~use of w~kncs~s in and ~he out~f-~li~o~ia ~ ~ co~d no~ ~ dct~t~ by the ~co~-lin~ p~s. ~ estimam o[ ~e pro~r~on of d~s ~cted to ~r among SDA study subj~s who mi~ted out of ~lifomia was obt~n~ by ~e ~mpfion that ~e pro~rtion of ~ oc~ng out of ~e Smm would ~ ~e ~me among SDA ~d non-SDA subj~ts who completed ~e A~ qu~tionnai~ while r~i~ng in ~Hfomia. ~n~r ~sion analysis and a ~phic #or of ~e proof- don of total non-SDA dea~s ~at ~c~ out of ~liforn~ for ~ch ye~ during 19~-71 r~ealed ~at an avera~ of ~.9% of the total d~s ocm~ else- wh~e. ~e pro~rdon of out~DSmte d~s showed • e exacted lin~ in~e ~ dme from 3.0~% in 1~61 to 5.58~ in 1971, which #v~ an avem~ in~se Fatal Cancer Among Seventh-Day A@¢ent~sts 1059 of 0.37%/year by linear re~ression. Thus the number of SDA d~aths was adjusted upward by a factor of 0.37,%/year during 1966--76. This report is limited solely to white subjects 35 years old or older in 1960. For each subject, l:mrson- years were calculated from January 1, 1960, until death, loss to follow-up, or termination of follow-up (June 30, 1972, for non-SDA; December 81, 1976, for SDA). Counts of deaths and person-yedrs accumulated over the follow-up period were stratified by age, sex, re- ligion, and education. Similarly accumulated counts of deaths and estimated total population for all U.S. whites during 1960-75 were stratified by age and sex. The Mantel-Haenszel chi-square procedure ($) was used to produce age-adjusted, sex-specific mortality ratios for SDA versus non-SDA and SDA versus U.S. whites from these stratified data, and the mortality ratios obtained for both sexes were also adjusted for a~Z old~r by method of ~a~ a~erfalnm~#2 SDA California ACS study ptrtieipant~. 1960-~ Primary calm No. of California deaths Percent of of death by Ascertained onlyLby record previously age in yr, Previously ascertained" linkage" ascertained year of death, deaths and sex Rettcertained ML~ed by True F~lee reaseertained by record record by record linkage linkage linkages ¢ linkages~ linkage All muses Male 78~ 40 9 4 95,I Female 1,093 • 96 20 3 91.9 Both sexes ~ 49 4 1 1 92.5 45-I~4 113 4 2 0 96.6 5,5-64 191 12 5 1 94.1 65-74 472 46 6 1 91.1 >--75 L030 70 15 4 93.8 All a~ 1,871~ 136 29 7 93.2 1960-61 494 39 2 2 92.7 1962-63 680 41 6 2 94.3 1964-65 701 56 21 3 92.6 All cancers Male 124 ' 5 2 0 96.1 Female 203 13 2 0 94.0 Both sexes 35-44 14 0 0 0 100.0 45-~4 43 1 0 0 97.7 ~5-64 57 2 2 0 96.6 65-74 103 4 1 0 96.3 >--75 112 II I 0 91.1 All ait~ 329 18 4 0 94.8 1960-61 103 3 0 0 97,2 1962-63 106 6 0 0 94.6 1964-65 120 9 4 0 93.0 • Deaths were identified by quarterly reporta from the ~lerk of each SDA Church congregation and by annual mail or person contact of each participant. Ti,~m clerks' reporte listed all deaths of which the clerk wm aware that occurred during the previous quarter, # Deaths were ascertained by a computer-misted linkage of the SDA study population to the 19~-65 death certificate file for the entire State of California. • True linkages are deaths on the Stat~ death certificate file that are perfect (or almost perfect) linkages on all linkage variables. The~e were confirmed by comparison of the subject's death certificate to the original questionnaire. The false linkages are deaths for which comparison of the death certificate to the original qu~ti~nnaire clearly indicated that the two records were for different let-sons. .INCl. YOL ¢5. NO. 5. NOVEMBER Ti09371650

PhlIllp~, GirflnkeI, Ku~m~, e! al. age and sex. For cancer sites that accounted for small numbers of deaths among SDA (stomach, lung, other smo'king-related sites, prostate gland, lymphoma, Icu- .kemia)~ 3 age groups (35-5,1, 55-74, --~75) were used for age adjustment, but for the more common sites, 10-year age groupings wcrc used. For each age-adjusted mortality ratio, separate values were obtained for the homogeneous ch/osquare and the associative chi-squarc. With the associative chi-square, the null hypothesis that the age-adjusted risk of dying of the specified cause is equal in the 2 comparison groups (i.e., mortality ratio--l) is tested. The homo- geneous chi-square tests the null hypothesis that the crude mortality ratio is equal for each of the age strata (5). A statistically significant homogeneous chi-square (P<0.05) indicates that the mortality ratio shows sub- stantial variation with age; therefore, the age-adjusted mortality ratio for all ages should be interpreted cautiously. The standard chi-square test was used to ascertain whether the age-specific mortality ratios were significandy different from 1.0. In the tables, appropri- ate symbols (defined in the footnotes) are used to indicate the P-values obtained from the various chi- square values. TA~L~ 3.---Number of ~chits Cal~fornia participants in th~ ACS study by a¢~ and reli~io~ during 1960 Age. yr SDA Non-SDA Male Female Male Female 35-44 ~.254 3.569 4,764 11,876 45-54 2,015 3,443 23.015 25,499 55-~4 L497 3.131 13.599 15.177 6~-74 1.369 2.848 6248 8.158 75-84 812 1.500 1.582 2.347 585 169 333 137 326 All ages 8,116 14,824 49.345 63,381 RESULTS The age-sex distribution of the SDA and non-SDA subjects is shown in table 3. Women predominated in both the SDA and non-SDA groups (65 and 56~, respectively) and a significantly higher proportion of persons 75 years and over was found among the SDA, particularly for females (12 vs. 4%). Inasmuch as all the mortality data in this report are either age-adjusted, sex-spedfic mortality ratios or age- and sex-adjusted mortality ratios, these differences in age-sex distribu- tions should not have a significant effect on the results. Table 4 provides convincing evidence that the SDA are much more comparable to non-SDA in the ACS study than m the general population in terms of educational attainment and marital status. Unfortunately, different occupation coding schemes were used for the SDA and non-SDA, which precludes meaningful comparisons on this variable. On the basis of the much greater degree of similarity in educational attainment for SDA versus non-SDA compared with SDA versus all U.S. whites, one can reasonably assume that broad occupational groupings would be similar between SDA and non- SDA who completed the ACS questionnaire. Thus the non-SDA in the ACS study appear to be a more appropriate comparison group for SDA. Table 5 shows SDA:United States and SDA:non.SDA mortality ratios for eight specific cancer sites as well as all cancer sites, all causes of death, and a composite of all other cancer sites not specifically listed in the table. Before examination of the mortality ratios for specific ~ncer sites, one should note that the SDA:United States mortality ratios for all causes of death were just as low as those [or all cancer (0.52 vs. 0.51 for males, 0.68 vs. 0.60 for females). This observation reflects the previously reported one that risk of death among SDA for most major nonmalignant causes of death is substantially below the risk in U.S. whites. However, TA~I~ 4~Prrerat of whits Californ~ SDA, non.~DA, and U.S. whites 85 yea~s of a~ and oldrr with various duriag 1960 Characteristic . California ACS study p~rtlcipants' SDA Non-SDA U.S. white popular/on for 1960 Male Female Male Female Male Female 8.116 14.824 49,345 63.381 33,316.901 36.102.755 Education Grade school or l~s 25.1 24.6 Some high school 20.5 23.7 High sehcoi graduate 9.9 12.9 Some college or trade school 20.3 28.7 Colleg~ gr~duat~ 24.2 10.1 Occupation Prcf~ssi0nal 29.8 19.9 Nonpr~l~ional 70.2 80.1 M~rit~l status Single 1.7 5.3 Married 94.4 70.0 Widowed 2.4 18.6 D[vorced-sel~rated 1.5 6.1 15.4 11.0 45.5 41.5 20.2 18.9 18.2 18.9 17.9 25.0 19.4 25.1 22.9 27.1 8.1 9.0 23.6 17.9 8.7 5.5 10.5 14.7 89.5 85.3 1.2 4.0 7.6 7.4 96.2 78.0 83.2 65.9 1.7 13.9 5.3 18.9 0.9 4.1 3.9 4.7 j.xc~ vOL ~3. No 5. NOVEMBER I~E3 T109371691

1101 No. of deaths ~ti°°'~ f~r ICD ~ma~ ~u~ S~ SDA: SD~ ~e ~f d~ SD~ U.$. U.S. No." SDA Non~DA n~n~DA ~i~ for whi~ 1~7~ for 1970 140°209 All cancer ~ 388 1.M 0.60** 0.52® 51"*** ~ 661 2,194 0.76"" 0.68"* 68**** ~, ~ 1,049 4,157 0.69** 0.61"" 61"** 151 Stomach can~er ~ 42 100 1.41 0.93 98 ~ 30 80 0.89 0.62"* 67*** ~, 2 72 180 1.14 0.77* 82 1~3-164 Colon-rectal ~ 59 277 0.62** 0.57"* 56***° cancer ~ 91 353 0.58** 0.51"* 51"*** 6, ~ 1~0 630 0.60** 0.53** 162 Lung cancer ~ 24 542 0.18"* 0.17.* 11"*** ~ 18 143 0.31"* 0.34** 23**** C~, ~ 42 685 021.* 0,21.* 15.*** Footnote Other smoking-re- ~ 47 265 0.f~9* * 0.40"* 97**** ~ l~ted cancer" ~ 57 183 0.74 0.58** 55**** ~. ~ 104 448 0.66** 0.49"* 45**** 174 Bre~t cancer ~ 160 553 0.85 0.90 89 165 Prmtate cancer 6 78 201 0.92 0.87 86 20~-203 Lymphomas ~ 37 122 1.13 1.10 107 ~ 51 144 0.99 1.08 104 ~, ~ 88 266 1.06 1.09 105 204-207 Leukemia ~ 18 93 0.54* 0.62* ~ 27 62 1.01 0.78 74 (~, ~ 45 155 0.75 0.71" 67**** Footnote All other cancerf ~ 83 363 0.74* 0.52** 72**** .f ~ 227 676 0.89 0.70** 75**** c~, ~ 310 1,039 0.&5*_ 0.64"* 74**** 001-999 All causes ~ 2.400 9.387 0.66@ 0.5I@ 58**** ~ 3.803 7,918 0.88@ 0.60~ 63**** 6, ~ 6.204 17,305 O.TS® 0.56~ 59**** • ICD-International Clmificatlon of Diseases (Eighth Revision). b Ratios were adju~i by the Mantol-Haen~zel procedure with the use of the age groups shown in table 6. " *, **-P<0.05 and P<0.01, respectively, h~sed on chi-square (a~ciative). Encircled aetrrf~k indicate~ that the hypothesis of uniform relative risk for the various strata was rejected. P<0.08 wa~ ba~ed on ehi-uluare (homogeneous). ***. ****-P<0.05 and P<0.01, respective|y, ba~ed on Bailar and Ederer's method (3). * SMR=(O:E)×100. where E~,the aceumu|at~l No. of expected deaths obtained by application of the ago-~ex cause-specific death rates for the total U.S. white population in 1970 to the' corresponding 5-yr age-sex group in the SDA population-at.risk at the beginning of each year, and O-No. of observed deaths among SDA from 1960 to 1976. • This group included mouth and pharynx (1400149). e~ophatm~ (150), larynx (161). bladder and other urinary organs (188. 189.2- 189.9L and pancrea.~ (157). /This group included all cancer sites not included in the specific categories above. especially for women, the cancer and all-cause mor- tality differentials between SDA and the general popu- lation were considerably reduced when the more appro- priam comparison group was substituted for U.$. whites. The same pattern (SDA:non-SDA ratios higher than SDA:United States ratios) was apparent for each of the specific cancer sites listed in table 5 except the lung and breast, and leukemia in males. As expected. the SDA:non-SDA mortality ratios remained signifi- candy below 1.0 for the smoking-related sites, which is obviously attributable to the abstinence from smoking among SDA. However, for male and female colon- rectal cancer and male leukemia, which are unrelated to smoking, the SDA:non-SDA mortality ratios were substantially below 1.0 (P<0.01 for colon-rectal and P<0.05 for male leukemia). The SMI% are included in table 5 for comparison to the data in this monograph from other populations. Tables 6A and 6B address the quesdon of whether the age-specific SDA:non-SDA or SDA:United States mortality ratios vary significantly. Considerable varia- tion was apparent in the age-specific SDA:non-SDA mortality ratios [or four of the eight causes of death shown in tables 6A and 63. Thi~ age variation was statistically significant only for all causes of death in both sexes. The statistical significance of the variation in the age-specific mortality ratios is indicated by the P-value below the SDA:non-SDA mortality ratios for each cause of death. For all causes of death in both sexes, the mortality ratios tended to get progressively closer to 1.0 with advancing age. A similar but statistically insignificant j.~cI. VOL. ~. XO. S. NOV'~MBEP. 1~3 T109371692

1102 Prir~ry eau~e Sex Age. yr of death No. cf deatY, s Mortality r-ati~s~ SDA: SDA: SDA Ncn-SDA n~n-SDA U.S. whit~ All ean~er ~ Colon-rectal Lung cancer Other smoking.re- lated cancer 35-44 4 16 -- -- 45-64 13 184 0,31°* 0.27"* 55-64 51 656 0.~0"* 0.3~*° 65-74 111 698 0.~5"* 0.51"* 75-84 123 345 0.65** 0.54** _>85 84 64 0.82 0.92 P=O~IO 35-44 11 4~ 0.72 1.00 45-~4 57 339 0.72* 0.68"* ~5-64 113 677 0.73** 0.~3"" 65-74 171 839 0.75** 75-84 211 ~8~ 0.85 0.71"* -->85 97 110 0.73* 0.75"* P-0.873 35-44 1 0 -- -- 45-64 2 31 -- 55-64 9 80 (0.68) (0.51)" 65-74 15 104 0.58" 0.51"* 75-84 19 46 0.76 0.54"" _>86 13 16 0.61 0.86 Pffi0,482 35-44 0 2 ~ -- 4(~'o4 2 42 (0.20)" ~ 14 85 0.72 0.56" 65-74 18 120 0.43"* 0.38** 75-84 38 87 0.68" 0,60** _>8~ 18 17 0.89 0.61" P=0.433 M 2 63 (0.12)'* (0.19)** ~74 15 407 0.21"* 0.17"* >75 6 72 (0.14)** (0.16)** P=0.319 3fr~4 2 37 (0.23)" ~ 55-74 3 76 (0.16)'" (0.10)** ->75 12 31 0.54 0.74 P=0.517 f~r'74 16 184 0.48** 0.28"* ->75 28 52 0.82 0.55"* " P=0279 35-$4 3 18 -- -- 5~-74 18 109 0.61 0.44"* >75 35 56 0.88 0.70" P=0.615 " D~hea indicate rarive baaed on <5 deaths where P>0.05. ", **=P<0.05 and P<0.01. respectively, ba~ed on chi-square. Ratio~ that were ~ an <10 deaths are in parenthtst$. pattern was evident for "other smoking-related cancer" in both sexes, for all cancer in males, and [or colon- rectal cancer and "all other cancers" in females only. There w~s no substantial age variation in the mortality mdos for all cancer or all causes of death among females nor for colon-rectal cancer, lung cancer, or all other cancers among males. Although the age pattern [or breast cancer was not statistically significant, the trend is particularly interesting. The mortality ratios for breast cancer tended to decrease with advancing age. Thus SDA women over 65 years o[ age may be a low-risk group compared with non-SDA women. JNcL VOL 85. NO. 3. NOI.'F.31BER Tl09371693

Table 7 shows that SDA:non-SDA mortality ratios are unchanged when the population is stratified by educational atta/nment. DISCUSSION The principal findings of this study are the signifi- candy lower risk of death from lung cancer, other smoking-related cancer,, colon-rectal cancer, and male leukemia among SDA compared with a similarly selected population of non-$DA who are nearly equiva- lent to SDA in terms of educational attainment. Inas- much as the reason for the low risk of smoking-related sites among SDA is obvious, the primary findings that deserve discussion are the low SDA:non-SDA mortality Fatal Cancer Among Seventh-Day AdventJzt,t 1103 rados for colon-rectal cancer (0X0) and male l_~ukcmia (0.H). Th, low SDA:non-$DA mortality rado for male leukemia is based on only 18 deaths in SDA males, whereas the mortality rado in SDA females {which is based on 27 deaths) is 1.01. Thus the apparent low risk of leukemia, which is confined to SDA males, is probably a chance observation with no etiologic sig- nificance. The small number of leukemia deaths clearly precludes any additional analyses to explain this obser- vation further. Although one is tempted to attribute the low risk of colon-rectal cancer among SDA to some aspect of the typical life-style unique to them, this would necessitate the assumption that the distribution of all other TASLE 6B.--Number of deaths and mortality ratios for ~elected ca~tcer sites among white California SDA (I960-76) and non.~DA (1960-71) age 35 years and older by age and sex Prirnsry mtt~ of d,~th S~x A~e, yr No. of d~ths Mortality ratio~~ SDA: SDA: SDA Non-SDA non-SDA U.S. whites Breast cancer ~ 3544 3 20 -- -- 45-54 31 122 1.07 1.22 55-64 41 189 0.97 1.00 65-74 36 134 0.76 0.81 ->75 47 88 0.7~ 0.78 P=0,425 Pr~stat~ cancer ~ 35-54 0 2 ~ ~ f~-74 25 114 1.18 0.99 75-84 28 68 0.75 0.69" >~ 25 17 0.98 1.11 All other ~ancer# c~ All muses P=0.464 3r~-44 3 7 ~ ~ 45-54 3 34 ~ (0~0)** fgr-64 16 140 0.74 0.43"* 6f>-74 22 113 0.81 0.46"* ->75 38 69 0.85 0.67" P=0,8~8 35-44 3 14 ~ ~ 45-54 15 9'2 0.73 0.~1" 5f~o4 40 230 0.77 0.64"* 65-74 72 190 1.07 0.79" ->75 94 150 0.88 0.72~ P=0.523 35-44 19 61 0f>6* 0.52'* 45-54 93 853 0.46"" 55-64 237 2,857 0.53°* 0.36"* 65-74 500 3.085 0.66"* 0.47** 75-34 830 1.981 0.76"* 0.~6.* -->86 722 550 0.82"" 0.74** P<0.001 35-44 32 83 1.09 0.98 45-54 136 709 0.81" 0.61" 55-64 280 1.591 0.77"" o.51"* 65-74 618 2.o47 0.34*" 0,51"* 75-84 1.337 2,306 o.9o** 0£9** _>85 1,401 1.182 0.98 0.76** P=0.0~ " D~shes indi~'~te ratios based on <5 deaths where P<0.05. ". *'=P<0.05 and P<0.0L respectively', based on chi-square. Ratios that were btted on <10 de~ths ~ i~ parentheses. v These are all cancer sims net specifically mentioned in this table or table j,xct, voL ~. No. ~. NOVF-~IBER 19~0 TI09371694

1104 Primary cause of death Sex California ACS study participants No. cf deaths Ag~-~ex-adjusted mortality High sch~l or le~ ~me college or mo~ SD~ non~DA SDA Non-SDA SDA Non-SDA High school Seine callege or I~ or more All cancer ~ 264 1,213 124 750 0.61"* 0.59"" @ 413 1,276 249 918 0.70"* 0.89 ~, ~ 677 2,489 373 1,665 0.55"" 0.78"* Stomach cancer ~ 31 69 11 31 1.39 1.35 g 21 50 9 30 0.86 (0.92) ~. ~ 52 119 19 61 1.12 1.12 Colon-r~tal <~ 39 170 19 107 0.55** 0.69 cancer g 61 215 25 138 0.54*" 0,55** 6~, ~2 100 385 44 245 0.55** 0.61"* Lung cancer <~ 18 355 7 187 0.19"" (0.14)** ~ 15 84 2 59 0.37"" (0.14)** ~, ~2 33 439 9 246 0.24** (0.14)** Other smokin~-re- ~ 35 160 12 105 0.65" 0.4,5"* lat~l cancer ~ 40 113 17 70 0.71 0.78 ~, g 75 273 29 175 0.68** 0.60* BreMt cancer @ 91 286 69 267" 0.84 0.89 Prostate ~ancer <~ 52 127 26 74 0.89 1.00 Lymphomas ~ 22 64 15 58 1.26 0.99 g 28 80 23 64 0.80 1.29 ~, g 50 144 38 122 0.97 1.15 Leukemia d' 8 50 9 43 (0.43)* (0.69) ~ 18 44 9 18 0,82 (1.51) ~, @ 27 64 18 61 0,63 0.95 All other cancer~ ~ 59 218 24 145 0.36 0.61" g 135 404 93 272 0.75* 1.14 d. ~ 194 622 117 417 0.78" 0,97 All causes ~ 1,685 5,728 715 3,659 0.68@ 0.61"" ~ 2.817 5.120 987 2,798 0.86" 0.90@ ~, ~ 4.503 10,848 1.702 6,457 0.78® 0.75@ • V-~ues were adjusted by the Mantel-Haenszei procedure with the use of the age group~ shown in tabis 6. ~ *. "*=P<0.05 and P<0.O1, respectively, based on chi-square (associative). Encircled ¢stcrisk indicates that the hypothesis of uniform relative risk for the variotm strata was rejected. P<O.05 was based on chl-square (homogeneous). Ratios based on <10 deathe are in ~renthe~ez. ~ This group included mouth and pharynx (ICD Cede So. 140-149)o esophagus (150), larynx (161), bladder and other urinary org-a~ (1~8, 189.2-189.9). and p~ncreas (157). " This group included all cancer sites not included in the specific categories above. relevant factors (except age) is comparable among ~he SDA and non-SDA who completed the same base-line questionnaire. This assumption cannot be substanti- ated with the present data. primarily because no known relevant factors are measurable by question- naire. Whereas considerable evidence suggests that dietary habits may be important in the etioloT/ of colon-rectal cancer (7), there are no clearly defined risk factors for colon-rectal cancer except for age and several rare syndromes involving colon polyps (which account for only a minor proportion of all colon-rectal cancer c~5). However, evidence to support tl~e comparability of the SDA and non-SDA in this study is substantial. The fact that the mortality ratio for colon-rectal cancer is ,Srtually the same when calculated in reference to all U.S. whites or non-SDA in the ACS study indicates that risk of dying from colon-rectal cancer among ACS J.~cL VOL e~. No ~. NOVEMBER non-SDA is comparable to that for all U.$. whites. Thus factors that are significantly different betw.een ACS non-SDA and U.S. whites (such as education and socioeconomic status) are not likely to be significantly related to risk of dying from colon-rectal cancer. In other words, the selective factors that characterize the subgroup o~ all U.S. whites who participated in the California portion of the ACS study are unrelated to risk of dying from colon-rectal cancer. Similarly, a set of selective factors exists that characterizes the subgroup of non-SDA who choose to convert to the SDA Church. These include high educational attainment, high socio- economic status, and probably many other unknown factors. The SDA and ACS non-SDA in this study are all white, noninstitutionalized persons who are similar in terms of education and marital status (table 4) and willingness to complete the same questionnaire. Both groups are also likely to be social participators as TI09371695

Faro! Cancer Among Seventh-Day Adventists 1105 opposed 03 loners. A large proportion of SDA tend to be heavily involved in church activities. The ACS stud)' subjects were obtained by local community volunteers for the ACS who solicited frienth or acquaintances to enroll in a stud)" group of persons with similar characteristics to those of the ACS community volun- teers (social parfidpa03rs). In v~ew of the comparability of SDA and AGS non-SDA in regard to characteristics other than diet or life-style and, if one considers the equal risk of fatal colon-rectal cancer in non-SDA and U.S. whites, a reasonable suggestion seems to be that the low risk of fatal colon-rectal cancer among SDA is largely attributable to some unique aspect of the typical SDA life-style or diet. Although dietary habits vary gready among SDA, the typical diet of a sizable proportion of them would be lower in saturated fat and cholesterol and higher in dietary fiber as compared with that of the general population. Sufficient variation in the SDA diet to produce a biologic effect is strongly suggested by a previous report that showed a threefold greater risk of fatal coronary heart disease among young SDA non- vegetarian men when compared with the risk among vegetarian men (8). This difference in risk persisted after adjustment was made for several major risk factors for coronary heart disease. A similar body of evi~nce suggests that diet may be related to female breast cancer (9, I0), particularly in the postmenopause years. Although the overall age- adjusted SDA:non-SDA mortallty ratio for breast cancer was not significantly different from .l.0, the age-specific mortality ratios are consistent with the above possi- bility. The small difference in risk of dying from breast cancer between SDA and non-SDA seems to increase with increasing age. Table 8 indicates that, compared with non-SDA women in the ACS study, SDA women tend to be younger at first pregnancy, have higher parity, later age at first menses, and earlier age at menopause, all of which would decrease their risk of breast cancer. Thus any observed low risk of breast cancer among SDA may be attributable to these differences in risk factor distri- bution. However, SDA women have a slightly higher weight distribution, and fewer SDA women report regular self-examination of their breasts, both of which might increase their risk of breast cancer death. Not shown in table 8 is the fact that the prevalence of breast cancer risk factors tends 03 be higher among SDA females who follow the vegetarian diet compared with SDA women who are nonvegetarians. A subse- quent report will examine the relationship of dietary habits to risk of colon-rectal and breast cancers among subgroups of SDA with different dietary habits. Clearly, this relationship will have to take into account the differing distribution of breast cancer risk fac03rs among these dietary groups. For several cancer sites and all causes of death, the difference in risk of death decreases with increasing age. This phenomenon reflects a delay rather than a complete protection from death due to these causes. It also raises the possibility that some of the decrea~ in cancer mortality among SDA compared with that among non-SDA reflects a more favorable sur¢ival dme after the onset of cancer among SDA rather than an actual decrease in risk of acquiring cancer. This possibility is presently under investigation in an ono going study that compares cancer survival rates in SDA versus non-SDA. TA~Lg 8..--Prment pre~lenee of characteristics related to breoR ect~eer r3k amon~ SDA and no~,SDA white California ~raea during 1980 Characteristic California ACS study participants SDA N0n-SDA n-14,339 Age at tint prego Nullipsrous 10.3' 11.6 <20 21.6" 14.8 20-24 3~.5" 33.9 2~29 20.3" 20.5 :>30 11~* 14~, Par/W 0 10"2" 11.5 I-2 38.1" 43.5 3-4 31.5" 31.7 :>5 22~' 13.2 Age at fh'st mens~ <12 132" 13.8 12-14 67.7* 70.7 :>15 19.1" 15.6 Breast feeding _>1 child for _>2 me~ 49.9* 56.8 Monthly bre~t self-examination 31.9" 38.5 Percent wJnopamal by current age. yr <40 13.9" 9.0 40-44 27.6" 18.9 4~-49 50.3* 37.8 50-54 87.8" 78.2 _>~5 99.5* 98.9 Height, inches <61 11.3* 8.6 61-~2 24.5* 22.2 67-68 8.9 10.3 69-70 1.8" 2.0 ~_71 0.5* 0.3 Mean 64.1" 64.4 Weight index, pounds" <~0 6.5" 5.3 80-89 17.9" • 24.1 90-109 51.8" 54.2 110-119 11.7" 9.0 129-129 o 6.2* 4.1 _>130 5.9* 3.3 " *=P<0.0I by chi-square. b Data collection wa, limited to women who ~ave delivered 1 or more living children. ~ Weight index= 100 (reported wt/mean wt of non-SDA women of the same height). Women who reported "sick at present," hi,cry of em~eer-heart di~ease-~troke, or who lo~t :>10 pounds in the year prior to completing the questionnaire have been excluded. J,~CL VOL ~5. NO. 5. NOVEMBER TI09371696

1106 No. of d~aths Agv-s~x-ad~usted m~rtaliW ratio"'# Primary cause Sex SDA: SDA: of death SDA Non-SDA non-SDA U.S. white~ for 19e0-65 All cancer ~ 134 883 0.67"" 0.59*" @ 228 1,032 0.82*" @, £ 362 1,915 0.76"* 0.67 Stomach cancer ~ 17 48 1.46 0.97 £ 11 48 0.77 . 0.59 ~. ~" 28 96 1.09 0.78 Col0n-reetal cancer ~ 22 118 0.88 0.69 ~ 38 147 0.82 0.67" ~, 9 60 265 0.84 0.68"" Lunar cancer ¢5 5 241 (0.11)" (0.22)'" 9 3 23 (0.45) (0.63) ~. 9 8 264 (0.15)*" (0.28)'" Other smoking-related ~ 13 124 0.44"" cancer¢ 9 21 71 0.96 0.71 ~. 9 34 195 0.66" 0.51"* BreMt cancer 9 52 276 0.78 0.89 Pro~tat~ cancer ~ 22 76 1.06 0.83 Lymphoma.s ~ 9 ,58 (0.92) (0.92) ~ 14 45 1.17 1.06 ~', 9 2~ 103 1.00 1.00 Leukemia c~ 10 45 0.96 1.09 @ 8 30 (0.91) (0,74) ~. ~ 18 75 0.93 0.90 All other cancerd c~ 36 173 0.96 0.54** ~ 81 ~4 0.94 0.69~* ~, ~ 117 507 0.94 0.64** All came~ ~ 827 4,308 0.78® 0.56@ @ L222 3,459 1.04 0.63@ ~', g 2,049 7367 0.92@ 0.60® • This wM ba~d on SDA who completed • brief census quemtionnaire in 1958 but did not complete the questionnaire for the ACS stuffy. ° ", *'ffiP<0.05 and P<0.0L ~ti~ly. ~ on ¢hi-~quare (~iative). Encf~[~ ast~sk indicat~ that the h~th~is of uni- fo~ ~lafi~ risk for ~e ~o~ st~ ~ rej~t~ P<0.05 w~ b~ on chi~qua~ (homogen~). Ratios b~ on <10 deaths are ¢ This ~up includ~ mouth and pha~nx (ICD CMe No. 14~149), ~opha~ (1~), la~nx (161), bl~der ~d other urina~ (1~, 189.~189.9), and ~e~ (15~. " This ~up inelud~ all ~cer si~ not includ~ in the s~ifi¢ ca~ri~ a~ve. The observation ~hat the SDA:non-SDA mormliW ratios for smoking-related cancers and colon-rectal cancer are equally low for both educational groups indicates that the level of education and socioeconomic status do not modify the estimated SDA:non-SDA mortality ratios. However, if the number of deaths allowed a finer breakdown of eductional attainment, the situation might be different. The difference in education or socioeconomic status between SDA and all U.S. whites probably accounts for a substantial proportion of the difference in risk of dying from cancer in sites that are unrelated to smoking other than colon-rectal. The adjustment factor used to inflate the SDA deaths during 1966-76 to account for underascertainment by the computer-assisted, record-linkage procedure may have been too small, which could result in spuriously low SDA:non-SDA mortality ratios. This possibility is not supported by table 9, which is a replica of table 5 except that the data were derived solely from the j.xo. VOL. e. No. 5. NO~.T-MBER 1~ 1960-65 period when the death-ascertainment proce- dures for both SDA and non-SDA were the same. The similarity in the mortality ratios in tables 5 and 9 for all causes of death is particularly important. That the record-linkage procedure would selectively miss deaths from a particular cause is unlikely. Thus the equality of all-cause mortality ratios for 1960-65 and 1960-76 would suggest that differences in the cause-specific mortality ratios for these two periods are most likely due to random variation. However, still bothersome is the fact that the two nonsmoking-related sites {colon- rectal and male leukemia) which show a lower risk among SDA versus non-SDA for 19~0o76 have consid- erably higher ratios for 1960-65. In summary, this comparison of the cancer mortality experience of 22,940 white SDA (1960-76) and 1 white non-SDA (1960-7~) living in California shows that: ~) SDA have a risk of dying from colon-rectal cancer which is 60% that of nonoSDA of similar socioeconomic status. This low risk among SDA is T[09371697

probably accounted for hy s~m~ asF~-t of th~ ~picaI SDA diet or life-style, b) SDA, who virtually all abstain from smok/ng, have a risk of dying from lung cancer and other smoking-related cancers that is 21 and respectively, of the risk among a population of non- SDA that contained a sizable proportion o£ cigarette smokers (51,~ among males and 32% among females) in 19~0. c) The difference in mortality, between SDA and all U.$. whites for other cancer sites is substantially reduced when SDA are compared with a l~O13ulation non-SDA who are socioeconomically similar to SDA. This phenomenon suggests that high socioeconomic status or other factors that characterize persons who choose to join the SDA Church primarily account for the apparent low risk among SDA, REFERENCES I.~ou FR. WAI.~.~ RT, Wooos KW. Cancer of ~he lung and mouth in Se~nth-D-ay Adeentists. A preliminary report on a populado~ study. Cancer 1~4:17:48~.t97. Fatal Cancar Among Savantl~-Day Advant~sts 1107 car.~r am:~g (3) BAIL,,t JC III, Kr'z~ F. S~gaiEc~..~ fa~:~rs fcr ~.- rati) ~f a Poisscn ~ariable to its exFectad~,n. Ei~metrics 1~-~4;,.~:63~- 643. (¢) HA.'.~V.O.~I~ EC. Smoking in relation ~o th~ death rates ef one million men and women. Natl Cancer lnst Monogr 1~55:1~: 127o204. (5) M~..'.-r~L N. data from ~etrospective studies o~ disease. J Natl Cancer Inst I ~59;22:719-7't8. {6) Ft.F.Iss JL. Combining evidence from £ourfoki tames. Chapt 10. In: Statistical methach for tate~ and proportions. New York: Wiley, (7) Gltat-L~M S, (8) PHtt.Ul~ RL, Lf.'aou FR. Br~.~ou WL. Ktr~lA JW. C~ronary. heart disea~ mortality among Seventh.Day Adventi, L~ with differing dfetary habitv---a preliminary report. Am J Clin Nutr 1978"~1 :S191-S198. (9) M.q.L~. AB, KXLLY A. CHO! NW, et aL A study M diet and b~ast cancer. A.m J Epidemlo| 1978;107:499-$09. (I0) DgWAAm) F, B,utunu.s-vAr~ FIALtWUM EA. A prospettlve study in getteral practice on breast cancer r/sk in posnnenopausal women. Int J Cancer ]974;14:15~-160. JNCl. VOL. ~. NO. 5. NOVX.'~mF_R Imo TI09371698

Cancer Mortality in a Human Isolate,.= i Alice O. Martin, 3 Judith K. Dunn, + Joe L+ Simpson, 3 Carolyn L. Olsen, 5 Sam Kernel, e Michael Grace, r Sherman Elias, a Gloria E. Sarto, a Blon Smalley, + and Arthur G. Steinberg ABSTRACT--C~ncer mortalib/(1~5-77) among 12,652 members of In inbred humln religious isolate, the Huttedtss, was com- pared with expectations based on mortality rites for the U.S. white population in 1970. Overall. Hutteritas had slgn;ficsntty fewer deaths from cancer thin expected (P<0.01), due primarily to fewer lung c,+ncars among males. Smoking is prohibited for thl~ religious group. The most frequent b/pec of cancer~ were leukemia and cancers of the digestive system, the prostate gland, and the female brae'eL Preliminary results suggest an as~oclatlon between rece~ive alleles and childhood teukeml& More stomach arid rectal cancers were observed than expected, but d|tflrerl¢ll were generally not significant. Famillal aggregates of cancers of the stomach and breast ere b~ing Investlglted. The tow frequency of cervical cancer Is consistent with currant evidence for an Msoclatlon of cervical cancer with e~rly Iga It first Intercourse lrld promiscuity, neither of which is charectedstl¢ of this pepuletlon.--JNCI 85: 1109-1113, 1~0. Our objectives were !) to ascertain the types and frequencies of neoplasia in the Hutterites, a large religious isolate, and 2) to search for factors predispos- ing to cancer. The role of recessive alleles is being investigated by inbreeding analysis because inbreeding increases the, probability of homozygosity and hence the expression of recessive alleles. Environmental fac- tors are being evaluated by a comparison of the Hutterite life-style and cancer rates with other (non- inbred) human populations. This report describes can- cer mortality among 12,652 Hmterites for 1965-77, which was evaluated by comparison with expectations based on mortality rates for the U.S. white population in 1970. Methodologic approaches and genetic aspects of the study will be reported elsewhere. SUBJECTS AND METHODS Population.--The Hutterites originated in Europe in 1528 and migrated to South Dakota during L874-77. There are now over 25,000 descendants of the 400 indiv/duals who settled in three original Hutterite colonies in South Dakota; these descendants are spread among more than 200 present-day colonies (communal farms) in the United States and Canada. The popula- tion we examined comprises the 12,659 Hutterites in South Dakota and in Alberta and Manitoba, Canada, who are descendants of the Schmiedenleut (S-lem) or the Lehrerlem (L-leut), which are names of the people descended from ancestors in two of the three original colonies. Mortality was not analyzed in the third group (D-leut), but cases of cancer are presented for compari- sons of sites involved in the three subpopulations. Case ascertainment.--We used two procedures to ascertain cases of cancer:. 1) record searches to detect individuals with one of the 15 Hutterite surnames and £) ~ield trips to colonies to obtain family and medical histories. Data were obtained from several sources: the Alberta Cancer Registry, 195~-77; the Manitoba Cancer Registry, 1944-77: the South Dakota Bureau of Vital Statistics (death certificates), 1960-78; and the Sacred Heart Hospital Tumor Registry, Yankton, South Dakota. Additional searches are underway with the assistance of the Saskatchewan Cancer Registry and the South Dakota Bureau of Vital Statistics (death certificates issued before 1960). A visit to all the colonies was impossible, but field trips were made to 27 colonies in Alberta, Manitoba, South Dakota, and Washington for addi- tional information on cancer among the founders of the various kindred subdivisions within each leut (group). During field trips, physical examinations were occasionally performed, but no new cases of cancer were ascertained in this way. Case conflrmation.~Each cancer patient ascertained through record searches required confirmation that the individual was in fact a member of a Hutterite colony. Indeed, the searches uncovered persons with similar surnames but who were not Hutterites in addition to persons whose religious and ethnic origins were simi- lar to those of the communal Hutterites but who did not live in a colony. Many o~ the latter group were "Prairieleut," i.e., Hutterites who did not adopt the Allllltw.~'rlo:~s t'stl~. ICD=,lnternatlonal Classi~i~tlon og Diseases; SMg:astandardized mortality ratiois). t Presented at the Workshop on Populations at Low Risk o~ Cancer condta:ted at Snowbird, Utah, August 23-25, 1978. * Supported in part by Public Health Service research grant R01- C.A19822 from the National Cancer Institute. ~ Depamnent o~ Obstetrics and Gynecology. Northwestern Uni- wrsit.v Medical School. Chicago. Ill. 60611. ' Cancer Center. Northwestern University Medical School. s Bureau o[ Disease Control, New York State Health Department, .~dban.v.N.Y. 122~7. ' Manitoba Ca~er Treatment and Research Foundation, Winni- peg, .Manitoba R.3E 0V9, Canada. ~ Department ol Medicine. Unlwrsity of ,Mberta. Edmonton. Alberta T6G 17.2. Canada. * Department of Biology. Case Western Reserve University. Cleve- land. Ohio 44106. t We acknowledge the imaluable encouragement and adsqce of Dr. Nathan/el Berlin. Di~ctor, Cancer Center. Northwestern L'nlversit¥ Medical School. We thank the personnel of the Tumor Registry. o[ Sacred Heart Hospital. Yankton. S. Dak.. and the Pdberta Cancer HospitaLs Board. Edmonton. Alberta, for assisting in data collection. We also thank the staf~ of the South Dakota Bureau of Vital Statistics. Pier~. S. Dak.. for conducting death certificate searches. 1109 jxcl. VOL. ~. .~O. 5. NOVE311~EP~ TI09371699

1110 Martln, Dunn, Simpson, st aL communal life in South Dakota when they migrated /ram Russia in the 1870's. We confirmed the Hutterite membership of cancer patients ascertained through record searches by matching name, sex, birth date, spouse's name if applicable, and colony or county of residence with similar information in preexisting family records from the colonies (1). Cases ascertained during field trips to colonies were confirmed as cancer and classified by anatomic and histologlc types. The ICD (Eighth Revision) (2) was used by the three cancer registries. South Dakota death certificates from 19~8 to 1978 were also classified by the ICD (Eighth Revision), whereas those before 1968 were classified by the Seventh Revision. Information from hospital records and all anatomic classifications not in the I~D (Eighth Revision) form -.-ere wan.slated hy our group to that [arm. Histologi~ identification was as- certained ~rom record review and was available for all but 7 cases; these 7 cases -.,ere from South Dakota and included the ,t whose ascertainment was solely by death certificate. Alberta histologic classifications were made according to (~)0 but all histologic information was translated according to the nomenclature given in (4). Estimation o! population si~s and preparation of standardized mortalit~ tabl~s.mFamily data collected during recent field trips were used to update the age-sex distributions of the populations (1) for each year with which this report is concerned. Because some recent family records were not available, we estimated the population size during 1971-75 for the L-leut and L--Cancer mormlltll in the Hutterltee ranked b~ site of caner" S-leut Manitoba: 1938-77 South Dakota: L-leut. Alberta: 1955-75 D-leut, Alberta: 1953-75 Primary site No. of Primary site No. of Primary site No. of Primary site No. of Digestive organs 15 Digestive organs 4 Digestive organs 10 Digestive organs 11 Breut (female) 6 Leukemia 3 Leukemia $ Bre--t (female) 6 Respiratory system 4 Breagt (female) 2 Breast (female) 3 Bone 2 Leukemia 2 Pr~tat~ gland 2 Prostate gland 2 Urinary system 2 Prostate gland 2 Female genital system 2 Female genital system 2 Brain 2 Female genital system 1 Urinary system ~ Urinary system I Respiratory system 1 Lymphomg 1 Respiratory system 1 Gum and mouth 1 Lip Connective tissue 1 Other sites 2 Other sit~ 1 Leukemia I Other sites 4 Endocrine gland~ Other sites 2 Total 38 17 25 29 TaSL~ 2.--Observed (Obs) and ezT~et~d (Exp) humbert of cane~n b~ site occurring amon~ males and/emal~ Primary site or cancer S-leut, M~ltoba and South D~kota: L-leut, Alberta: 196rr-75 Males Females Males Female~ Obe Exp Obj. Exp Obs Exp 0bs Exp Total malignant neoplasms 16" 31.7 Lip, oral egviW, and pharynx 0 0.9 Digestive organs, peritoneum 6 R~pim~ sys~m 1* 9.5 Bon~ ~d join~ 0 0.3 Conn~five, m~u~. ~d o~er ~ft t~u~ 1 02 Me~noma. skin 0 0.4 Bre~ (retie) Female ~ni~ M~e ~ni~l ~m 3 2.6 Uri~ s~m 0 1.7 Eye 0 0.0 Ne~ D~m 0 1.4 ~nd~ne glands 0 0.1 L~phom~ 1 1.8 Multiple myelomz 0 0.3 Leuk~ia 3 2.3 Other ~d unknown si~ 1 1.5 18 24.9 13 1L7 5 8.1 0 0.3 0 0.3 1 0.1 7 6.2 6 3.2 2 2.1 1 2.2 0¢ 3.4 0 0.7 0 0,2 0 0.1 0 0.1 0 0~2 0 0.I 0 0.1 0 0.3 0 0.1 0 0.1 3 5.2 0 0.0 2 1.7 3 3.8 0 1.2 2 1~ 1 0.7 I 0.7 0 0.2 0 0.0 0 0.0 0 0.0 0 1.0 0 0.4 0 0.3 0 0.2 0 0.0 0 0.0 0 1.2 0 0.6 0 0.4 0 0.3 0 0.I 0 0.I 2 1.7 4~ 0.8 0 0.5 ' I 1.3 0 0.6 0 0.4 • P<O.O1. ~ P~O.IO. jxo. vOL ¢. NO. 5. NOVEMBER TI09371700

during 1959-76 for the Sqeut in Manitoba by fitting a thircl-d~gree polynomial to valu~ for 1955-70 and I977 and exwapoladng for intermediate years. Because D- leut family data were not as accurate as those of the S- and L-leut families, cases ascertained among the D-leut are presented only to compare the relative distributions of types of cancer among the 3 groups. Only information on cancer mortality during 19~-75 (L-leut) and 1965o77 (S-leut) was included Ix'cause the standard of comparison gas the 1970 U.S. white population. The significance of the SMR was deter- mined with the use of the techniqu~ of Bailar and Ederer (~). RESULTS The ranked distribution of deaths by site of cancer in S-, L-, and Ddeut is shown in ruble 1. Only the 52 S- leut and L-leut deaths occurring in the interval begin- ning with 1965 were the basis for the computation of the SMR calculated from observed and expected values summarized in table 2. The total number of deaths due to cancer was small; however, the size of the popula- tion surveyed was also small. The total population size [or S- and L-leut in Manitoba, South Dakota, and Alberta in the interval included in this report was 12,652. Became families are large, 50% of the popula- tion was le~s than 15 years old, an age group not at high ri~k for cancer. During 1965-77 (S-leut) and 19~-75 (L-leut), 29 cancers occurred in males and 28 in females. Overall, fewer deaths were observed than expected (P~0.01) largely due to lower mortality from lung cancer among males (1 death observed compared with 13 expected) ~P~0.05, $-leut; P~0.10, L-leut). Although not significant in all classifications, excess • numbers of leukemias and stomach, uterine, prostate, and rectal cancer~ were noted. The Hutterites with leukemia appeared to fall into 2 groups based on age at death: parents 18 years old or younger and those older than ~0 years. Table illustrates the pattern of deaths in the first group (6 The most frequently occurring sites for cancer among Hutterites were those of the digestive organs, primarily the stomach. Familial aggregates of stomach cancer were found in our entire sample. The only significant excess in the interval covered by this report was that of rectal cancer among L-lem males. Overall, the number of colon cancers was approximatel~ equal to that rectal cancers. The next most common cancera were those of the breast (female), the prostate gland, and the female genitalia, excluding the cervix. However, the only significant deviation from expectations was in the number of uterine cancers among the $-leut. Most breest cancer occurred in women over the age year;. No deaths from cervical cancer were found among all cases ascertained. Cancar MortaIily in a Human Isotate 1111 ~NCI. VOL. ~. NO. 5. NOVEMBER I~0 TI09371701

1112 Madln, Dunn, Simpson, et aL DISCUSSION There were fewer deaths from cancer among the Hutterites than expected on the basis of comparisons with the 1970 U.S. white population. This difference was significant (P<0.01) and appeared to be due primarily to fewer lung cancers among males (5). This is to be expected in a population who prohibits smoking. Only a 42-year-old male died from lung cancer, whereas 13 deaths were expected. One 78-year- old female with lung cancer was ascertained through a death certificate, but no histologic information was available. Also consistent with what would be expected in a nonsmoking population was the finding of only a single death due m bladder cancer and one due m oral Only 1 patient with cervical cancer was observed in the total Hutterite population; she is still alive and has adenocarcinoma not squamous cell Cardnoma. This low frequency is consistent with current evidence for an association of cervical cancer with early age at first intercourse and with promiscuity, neither of which is characteristic of this population. Compared with the 1970 U.S. white population, the Hutterites generally appear to be at low risk of developing cancer. This finding is consistent with expectations based on Hutterite life-style, but genetic factors may also be involved. Underestimation of cancer mortality due to their less frequent use of medical fadlities than the general population is unlikely, particularly of Canadian Hutterites who have access to a national health system. Hutterites are concerned about their health and do not hesitate to use modern facilities. Because most Hutterites die in the colonies and all are buried in the respective colony cemetery, we believe that death certificate information underesti- mates Hutterite deaths regardless of cause, which could have been a source of bias in our study. However, the only death certificate searches conducted were for S-leut in South Dakota. Only 4 cancer cases were ascertained solely by use of this source, and the South Dakota cancer mortality patterns do not appear to differ from those of S-lent in Manitoba, where ascertainment was through the cancer registry and field trips. The most frequent types of cancer among the de- ceased S- and L-leut were leukemia, cancers of the digestive system (stomach, rectum, and colon), and cancers of the prostate gland and female breast and uterus. Cancers of the digestive system and female breast were also among the most frequent types of Cancer among the deceased D-leut in Alberta. The excess of childhood deaths due to leukemia (P<0.10) in an inbred population may be of special interest. The average coefficient of inbreeding for children dying of leukemia "(i~_0.0't) was higher than that of the total S- and L-leut (F~0.02). Definitive con- dusions regarding the possible role of recessive alleles in leukemogenesis must await the completion of in- breeding analysis in which patients are compared with controls matched for year of birth, completeness of the Ix~ligree, sex, and Ieut. Although generally more cases of stomach and rectal cancers were observed than expected, the differences over the interval were not significant "except for rectal cancer in L-leut males. An elevated rate of stomach cancer %-as expected because high rates have b~en reported in the countries of Hutterite origin (Austria, southern Germany, and the USSR) and among mi- grants from those counta'ies (6). Because Hutterite dietary traditions are likely to have been preserved over the decades more than those of other Canadian and U.S. ethnic groups from East Europe, the rates of stomach cancer may have remained stable, whereas those reported for Canadian and U.S.-born, non-Hut- terites of East European origin have been decreasing. We have ascertained familial aggregates of stomach cancer that exhibited longitudinal transmission and are investigating them further. Because focal is prepared in a common kitchen and individuals rarely eat meals in family units, the possibility that these aggregates of cancer result from familial dietary habits rather than common genes is lessened. However, a geneticoenviron- mental interaction may exist because family members tend to live on the same colonies, and there may be intercolony environmental differences. Hutterite women are obese; therefore, endometrial cancer might be increased. Counteracting this increased risk of developing this type 0[ cancer would be-Lactors believed to lower the risk, i.e., the high parity and infrequent use of exogenous estrogens among Hutterite women. (Contraception has traditionally been rejected.) Although the rate of uterine cancer mortality was significantly elevated in S-leut females, we cannot be certain how many of these cancers were endometrial. The rate of female breast cancer was not detectably different from the standard. In this instance, high parity and relatively late age at marriage may counter- act each other as risk factors. Our data are consistent with those of previous studies of cancer incidence among the L- and D-lent in Alberta (7); these previous investigations also showed that, compared with other whites, Hutterites are at lower risk of cancer. Further investigations are under way with the use of other reference populations matched geographically, temporally, and o~cupatiofially. REFERENCES (1) S'rtlNtt£Ro AG. BL~I~,£V HK. KVKCZVNSKt TW. 3hewn AO. gt'R~s~ EM, ~mdc stud~ on an inb~ huron i~late. In: ~ow JF. N~I JV, ~s. pr~ings of the thi~ inter- national c~gr~s of huron genetks. ~ltimore: Johns Hop kins U~v P~, l~7:~7-~. tional ~ifi~d~ of Di~, Eigh~ R~s/~. ~pt~ for u~ in the Unit~ States. Washington, D.C.: U.S. ~vt Print om I~. (3) THOXI~N E~. K~L~ A~ ~s. Sta~ ~om~clatu~ of di~s a~d o~oti~s. 5~ ~- New York: MeG~-HilI, (4} PER~" CL B~G JW. THO~ ~. 3~n~l of I~or tmmen- jxca. rot_ ~. xo. 5. xovE3mE1~ 1.0s3 T109371702

clinic acd co~r~g. New Ycrk: .-ta~ C~r.cer $oc. 1~9. (5) H~Z~L V;..Mi~ant s~udies. ~n: Fraum~ni JF Jr. Persons a~ Cancer MortaZZly ~n a Human Isolate 1113 J,~CI. VOL ~. NO. 5. NO1.'E~MBE~ T[09371703

American White Protestant Clergy as a Low-Risk Population for Mortality Research, Haitung King 2,9.4 and Frances B. Locke ABSTRACT--An exam|nation of 28,134 c~ergymen in five pre- dominantly white Protestant cle~om[nat~ons. morn Mvo~b~a mo~all~ for the~ cle~ymen for de~ to~l cn~e~, ~d ~Iova~u)ar-~nal d~m com- pared ~th t~ ~1i~ for U.S. white ~)es, wl~ ~ ~pe~ence. and totll U.S. white c~er~en. The mndard~ed mo~ll~ ~tl~ for ~r of ~e lung and non- motor-vehlcIe ~c~den~ ~m pa~lculady depmmd. However, • o~ for dl~etes, leukemia, ~d cmnce~ of the prostate gland and, ~ s l~mr extent, lymphoms B~ ~ncers of ~e Intestine a~ p~cmBs ~re not significantly different from I~. The flndlngs w~ Inm~ret~ In the ~IM cla~ conte~ with emph~ Ms on cm~Uonal Inv~Ugations and crltl~l me~ent of ~e~l ~tlstl~.~NCl ~: 111~1124. 1~0. In 1713, Bemardino ILamazzini, the dean of occupa- tional health, .published a Latin text on "Disease of Workers" (I). A distinctive feature of this classic work is ~he inclusion of a long dissertation dealing with "Diseases of Learned Men." Since then and up to the 20th century, an impressive number of studies on longevity of members of the learned professions (used here to include, as traditionally defined, clergymen, physicians, lawyers, and teachers), almost exclusively of clerics, has appeared in the Western literature (2). In this century, research interest in the health and mortality experiences of persons in the learned profes- sions, both clericaland nonclerical, has been sporadic. However, investigations dealing with clergymen re- main predominant, and a few studies have been de- voted to the medical profession. The official statistics and study findings pertaining to the health of clerics and physicians since earlier years were reviewed in some detail by King and Bailar (2) and King (3). However, many of these works suffered from dam and/or methodologic limitations. The paucity of recent epidemiolog/c research on the health of clerics is certainly unfortunate, particularly because the clergy serves as an ideal profession for the ascertainment of occupation-centerd health risks: The life-styles of minis~-rs in the same religious bodies are likely to be similar, the turnover rate of the clergy is known to be low (as noted later), and good records on clergymen are available in most major denominations. Moreover, the clergy and persons in other learned professions generally share certain common character- istics such as social origin, educational level, and occupational stability. No less importantly, the clergy further is a unique population group for the ~amina- tion of low mortality risks over time. For example, among British clergymen, hhe depressed ratios for overall mortality were nearly constant from 1860 to 1972 (4-I0). A similar trend was shown for cancer and several other causes o~ death (tables 1, 2). In response to the need for a systematic ascertain- mere of mortality risks among the U.S. white Protestant clergy, this paper presents the results of our new analysis of pooled mortality data on 5 clergy groups and speculates on social class effects on clerical health risks. CLERICAL POPULATION-AT-RISK The study was based on a defined population of 28,134 active or retired clergymen who were recorded as being alive on one of three dates in the rosten of the following predominantly white Protestant denomina- tions: 1) the American Baptist Convention, currently known as the American Baptist Church in the United States of America {the clergy of which entered the study on January 1, 1950); 2) the United Lutheran Church in America, known since 1962 as the Lutheran Church in America (for which the date of entry to the study was September 1, 1950); and 3) the Protestant Episcopal Church in the U.S.A., the United Presbyterian Church in the U.S.A., and the Lutheran Church-Missouri Synod (for which the entry date was January 1, 1951). These denominations represent some of the major Protestant religious bodies whose membership mainly originated in above-average social classes. The study group is composed primarily of white married clergymen. Those under 65 years of age at the start of the study constituted 16% of the total U.S. white clergy of corresponding ages in 1950 (11). About 90% of the subjects were born in the United States; of the foreign-born, the distribution ranged from 2~ for Presbyterians and 13% gor United Lutherans (mainly A~S~VI~ONS OSiriS. CMFmcomparative mortality flgmx'(s); IC_.Dm " International Classification of Diseases: PMKatproportional moc~aliw ratio; SMK~standa~lized mortality ratlo~s). i Presented at the Worksho~ on Populations at Low Risk o~ Cancer cond,,eted at Snowbird, Utah, August 2~-25, 1978. z Biometry Brar~h, Division of Cancer Cause and P/evention. Natiot~al Cancer Institute. National Institutes of Health, Public • Health Service. U.S. Department of Health and Human ~wlces, ~ Department o! Community Medicine and Family Health, School of Medicine. Graduate School. and Kenne~ty Center for Population Research, Georgetown University, Washington. D.C. 20~07. • dcldres8 reprint reqtwsU to Dr. I~ng a~ the Natiooal Cancer Instimte. 1115 TI09371704

1118 King and Locke TABU~ 1.--Tre~ in to~l ~.,a~alit~ a~,,c~y clsvg~'=.,t~ a~ acth~ Wales, I3~I972~ Active a~d Clerk- Mortality measure and yr retired n~le~ men CMF (all male~ 1906-02 = 100 for 1860-1902: all ratles 1910-12 - 100 for 1910-12)" 18~0.1861.187o in 7o 18~-32 105~ 60 1890-9~ 11~~ 63 1900-02 100a 54 1910-12 79 44 Ages 1,5-~ yr SMR (all males in given yr'- 1321-2~ I01" 60 1930-32 I0~/ 69 1949-53 81/ 1959-63 99 1970-72 103" 76• • Dtta were compiled from (4-10). ~ Total clergymen are reported for 1959-63 and 1970-72: clergymen of the Church of England only are reported for all other year~. ~ The CMF for • given population is a death rate. standard- lzed to a population among whom exactly 100 (or 1.000} deaths occur. The SMR for • given population is the ratio of the observed death rate to the rate in a standard population. justed to the age distribution of the study group. Both the CMF and the SME are equal to 100 when the study population has the sa,me age-~poeific mortality rate as the standard population. • Value excludes the retired. • Value excludes the ir,-ctive. ! Valu~ are for males 20-64 yr old. • Value ropres~nts males 15-64 yr 01d. German) and Episcopalians (mainly "Canadian). Ap- proximately 90~ of the study group were active clergy- men and 10% were retired. Among active clergymen, the distribution of parish ministers varied from 72% (United Lutheran) to 86% (Lutheran-Missouri Synod and Presbyterian); the rema/nder were such profes- sionals as church adminiswators, college teachers, mill- taw chaplains, and mission workers. TABL~ 2.--SMR by ~e[ected ea,s~s of death, clercrltrnen 15-6~ ~ars old. England and Wales. 19~9--68• Cav,~ of deaths 1949-53~ 1959-63a Tuberculosis (001-019) 43 9 Cancer of stomach (151)48 54 Cancer of lung and bronchus (162-163) 31 17 Oiabetes (260) 33 50 Pneumonia (490-493} 40 37 Suicide (E963: E970-E979) 48 53 Accidents (ES00-E936.0)70 43 • Data were compiled from (8. 9). s .V, mbers i, Ixzrznt~es~s are ICD (Six~ P~sion) c~es for 194~53 and ICD (Seven~ ~s~on) e~es for 195~3. ~ ~e~en ~lcng ~ ~e Chu~h of England. # Values include all cle~en. j.XCl, cOL ~. NO. ~. NO~.q~Mi~ER Ib'39 Pr~#.ant den~innh'~s a~d ~e 5 d~r.o~tinatic~s in th~ stud~° All 1$ Five den~minatior~ Are~ den~minatians in this study All areas I00.0 I00.0 United States 97.0 98.6 Northeast 20.0 31.0 North-central 35.0 3~.2 South 31.0 17.~ West 11.0 13.9 Other areas 3.0 1.4 • Data ~re compiled from an unpublished report of P. tL Scheror. Over 90% of the subjects in the five denominations resided in the United States. Most seemed to have a stable occupational and residential histow. For ex- ample, our findings indicated that approximately 60% of all retired United Lutheran clergymen had either remained in the same position during their entire ministerial career or had changed assignments fewer than three times (12). A survey of a sample of parish ministers of 15 predominantly white Protestant denominations con- ductcd in 1964 (Scherer PR: Unpublished report) pro- vides information on the geographic distribution and selected social characteristic~ of some of the clergy in the 5 denominations whose mortality experience is examined here (tables ~-5). We shall summarize the social origin and educational status of each o~ the 5 groups included in the survey. Forty percent of the ministers came from upper-class families, ranging from a low of 25% for the Bapdsts to a high of 51% for the Episcopalians. The corresponding figure reported for the entire survey group was 3~1%. Nearly 25% of the subjects originated in clerical families, in contrast to 1~% of the total survey population. However, a much lower distribution was shown for United Lutherans, Episcopalians, and Baptists: 13, 9, and 6%, respectively. The proportion of the study population with full college and seminary training compared favorably with that of the entire survey group (47 vs. ,t6%). However, TAn~ 4.--D/art/button b~ population.s~ group of elerl~ntn tt~th charge in 15 l~fotestant denominations and the ~ denominations in this st,dya All 15 Five de- Population-size group denomina- nomina- tions in ~ons this study All areas 100.0 100.0 _>500.000 9.0 II~: 100.000-499.999 16.0 20.9 25.000-99.999 18.0 20.2 2,500-24.999 24.0 23,3 <2,500 29.0 21.5 Unknown 4.0 2.9 Data were compiled from an unpublished report of P. R. Soberer. T109371705

All 15 Five de- Social characteristics den.~rnina- nomina- tions ticns in this study Father'~ o~eul~ttion (clmified ~ 83.1 40.7 profe~ionai, official, and rrmr, ager[al) Full college {4 yr) &nd beyond 75.0 85.4 College and seminary 46.0 College. seminary, and graduate 29.0 work Previous affiliation with other 26.0 80.1 denominations 47.5 37.9 Data were compiled from an unpublished report of P. R. Scherer, compared with other clerical groups in the total survey population, United Lutheran and Presbyterian churches had a higher proportion (,H%) of ministers who did additional graduate work. The distribution of clergy- men with previous affiliations sometime in their lives was 40% each for Episcopalian, Presbyterian, and Baptist clergymen and 5-9% for the 2 Lutheran groups; an intermediate 26% was reported for the 15 denomina- tions combined. Comments on patterns of distribution of clergymen by area and population-size group are in the "Discus- sion." This survey was conducted 10 years after the mid- point of the interval 1951-60 that is examined in our mortality stud),, so it gives only a rough approxima- tion o! the geographic distribution and social charac- teristics of the study cohort. DATA AND METHODS Analysis of mortality was based on 5,207 deaths that occurred during 1951-60. Personal information on clergymen in each denomination for this study period was provided by the respective church Pension Office. Death certificates or certified letters were obtained from the health offices in this counu'y and abroad. The causes of death were classified in accordance with the ICD (Seventh Revision) and coded by a single experi- enced nosologist. The person-years of exposure contributed by the subjects over the 10 years, aher the number of with- drawals by deposition (termination of services) or by death was taken into consideration, were analyzed separately by l-year and 5-year age categories. Inas- much as the results thus obtained were almost identical, only the quinquennial are presented in table 6. The index of SMR was used for the measurement of mcrtality levels. (See table 1, footnote d, for definition.) The expected number of deaths entered into mortality rados of ministers who died at age 20 years and ahov~ was computed on the basis of mortality experience of White Protestant Clerical Mortallt~, 1117 U.S. white males of comparable ages in I955, the midpoint year of the study period [U3); Appendix A in (14)]. Confidence limits were calculated with the use of the 95% tabular confidence factors described by Mantel (I5). The lower- and upper-limit factors for th~ observed numbers on which the SMR were based were multiplied by the SMR to obtain the confidence limits. STUDY FINDINGS Mortality Pattern The number of deaths and SMR for total mortality and selected specific causes of death are presented in table 7. If mortality rates for all deaths were the same as those among the general white male population of the United States, about 7,245 deaths would be ex- pected to occur in the clergy group followed for a 10- year period; however, only 5,207 were observed. Thus the SMR was 72. Total malignant neoplasms, which as a group ac- count for the second largest number of deaths, showed an SMR of 63. Mortality from one of the subcategories of malignant neoplasms, cancers of the lung, was far lower than that in the general population, with an 5MR of ~5. Other cancer sites with low SMR included stomach, rectum, and bladder, for which the SMR were 48, ~1. and 56, respectively. In contrast, the SMR for intestinal, pancreatic, and prostate cancers, as well as for leukemia and lymphoma, were not significantly different from I00. The SMR of 74 for major cardiovascular-renal dis- eases, representing 67% of total deaths, was similar to that for all causes combined. Mortality ratios for tuberculosis and accidents, among others, were also TASLg 6.---A~ distn'bu~ion at entry and wit~idrau, al No. living No. of u of withdrawals Per~on-yean Age. yr. date of exp~ure of entry~ Deposition deaths Total for 28.134 L404 5~07 249,215,5 all ages 20-24 1G0 1 0 248.0 25-9-9 1.542 12 3 3,973.5 30-34 2.830 96 7 14.701.0 3~39 3,309 194 35 26.036.5 40-44 3.470 288 48 31,460.5 45-49 3.190 270 103 32.083.5 50-54 2.,5~3 237 184 29.193.5 55-59 2.827 143 237 24,514.0 60-64 2,346 87 385 21.909.5 65-~9 1,931 44 ~62 20.001.0 70-74 1.639 17 7L2 16.534.5 75-79 1.435 7 828 12.812.5 80-84 913 4 944 9,183.0 -~$5 499 4 1,159 6,564.5 ° danmtry 1. 19S0. for Baptist: September 1. 19~. far United Lutheran: January 1. 1951. for Episcopalian. Presbyterian. and Lutheran-Missouri Synod. JNCIo VOI.. f~5. NO. 5. NO~.T.MB~'R TI09371706

1118 King and Locke C~ of dmth" 0l~rved Exr.e~ted S~.4:R~ deatI~ All caus~z 5~7 7,242.7 M~or eard~ova.~ular-.r~nal di~c~a~ (3~0-334: ~488; 592-594) 3,470 4,~.4 ~s of ¢a~io~u~r s~m (~3~; 40~4~) 3,~7 4,598.4 V~r lesio~ ~f~ng ~nt~l ne~o~ s)~m (33~4) ~5 Di~s cf h~ ~d rheumztio fe~r (4~; 41~43) 2.~ Rheu~fic fear ~d eh~nic rhe~at[c h~rt di~ (40~402; 41~416) ~ 67.3 A~ri~lemffc h~ d~, inci~ing co~ di~ (420~ 1.912 2,493.8 ~H~le~ffc hea~ di~, ~ d~ (420.0) 4~ 955.1 ~r d~s d ~ ~ (420.h 420.2) 1,476 I.~.6 No~e~afic ~nic end~ifis a~ o~er my~a~i~ degene~fion (421-4~) 219 ~her di~ d h~rt (43~4) 68 108.6 H~nsion ~ h~ di~ (~3) 205 ~1.5 H~nsion ~out mention of h~ (~-44~ 57 ~.8 ~e~l ~le~sb (~) 176 ~.0 ~mnic ~ uns~cifi~ neph~tls and o~er ~i ~le~is (59~594) ~ ~.0 Mali~nt ~opla~. including n~pl~ms d l~atic and hema~ietic 681 1,083.8 ~fi~ sys~m (1~1~ 157-159) 2~ 426.6 S~m~h (151) 61 ~6.7 In~sfine a~ r~m (15~[~) ~6 170.4 In~sfine (15~1~) ~ 112.0 ~m (1~) 30 ~.4 ~s (15~ 46 68.3 ~plmW~ ~m (I~1~) 70 19E3 ~ng aM bm~h~ (16~163) 62 176.4 M~# g~ni~ o~ (177-179) ~ 1~,2 ~ gland (17~ 1~ 147.4 ~rina~ o~ (I~181) 38 69.4 Bl~der ~ o~er u~W o~ns (181) ~ 48.5 ~em~ ~d £e~¢mia (2~) 38 41.7 ~phoma (2~2~; ~5) 36 42.9 All o~er ~it~nt ~pl~ms ~ 149.7 ~be~ui~is, all fo~s (~1~19) 6 66.0 8~hills ~d i~ ~quel~ (~9) 9 15.5 Di~ melli~ (260) 83 89.7 lnfluen~ a~ pneumon~ e~ept pne~on~ of newbo~ (4~493) 137 1~.3 ~eumoni~ except pneumonia of newbo~ (4~493) 123 179.9 ~ia of p~ g~nd (610) 44 74A ~c~en~ (~2) 158 ~4.5 MoOr ~hicle ~¢iden~ (ESl~E~5) 82 105.3 ~1 o~er ~ciden~ (~02; E~E~ 77 199.4 Suicide (E~3: E~E979) ~ ~! o~er ~u~ ~91 72 (70-74) 74 (71-77) 7~ (72-78) 77 (71-83) 73 (70-76) 56 (4O-77) 77 (74--31) 46 (42-~1) 96 (91-101) 63 (~9-80) 68 (59--78) 104 (79-135) 62 (55-70) 48 (37-62) 74 (63-88} 86 (70-105) ~I (34-73) 79 (58-105) 35 (27--~) 35 (27-45) 93 (70-110) 94 (79-112) ~ (39-76) 56 {37-82) 91 (64-12J) 84 (59-116) 60 (48-74) 9 (3-20) 58 (27-110) • 93 (74-115) 72 68 (5~-81) 59 (43-79) 52 (44-61) 77 (61-96) 39 (31-49) 3£ (21-46) 89 (83-97) • Number~ in par~t~e~s are ICD (Se~n~ ~ion) ~. ~ U.S. w~ mai~, ~20 ~ old, 1~5, ~us] 1~, ~um~ ~ ~ i~i~ 95~ c~den~ limit. TAB~ 8~A~ SMR ~ ~ ea~e of ~a~ ¢a~ ~g ¢1~ s~du ~. ~0 yea~ ~d and m~ ~ f~m m~or ~a~ f~m all ea~s ~io~a~nal ~aths f~m m~i~ant ~a~! f~m all o~er 4~68; 0~ Ex~ SMR~ O~e~ E~ SMR* 0~ Ex~ct~ 8MR' Ob~ E~ SMRb ~20 5,2~7 7,242.7 72 3,470 4,666A 74 681 1,083.8 63 1,056 1,492.5 71 20-44 93 230.7 40 36 83.2 43 17 29.5 58 40 . 118.0 34 45-54 287 557.0 52 155 286.7 54 49 96.7 51 83 173.6. 48 55...64 622 1.016.1 61 410 5S6.3 70 97 200.5 48 115 239~ 50 65-74 1,273 1,776~ 72 839 1,117.9 75 195 323.5 60 239 334.8 71 ~ 75-34 1,770 2.349.8 75 1,189 1,628.3 73 230 319.2 72 351 402.3 87 >--85 L158 L312.8 88 $38 9~3.9 87 93 114.3 81 227 234.6 97 Unspec- 4 ~ -- 3 ..... I -- -- ined • Nam/~r~ i~ ~varen~eze~ are ICD (S~venth Revision) cc~e~. ~ U.$. white maI~, >-20 yr c!d, 1955, equal 100. ~NCL VOL ~. NO. 5, T109371707

significantly low: However, the mormliW redo for diabetes was not significantly different ~rom I00. Age-Spe¢lll¢ Death Rat. When age-~peclfic death rotes were computed, we found that the depressed SMR of the study group was largely a reflection of the extremely low death rotes of young m/n/seers (table 8). For example, a fairly steady trend in SMR from a low of 43 for the age group 10--4't year~ to a high of 87 for the group over 85 years old was observed for all causes of death combined. Varia- tiom in age-specific death rotes for cardiovascular-renal diseases generally followed the overall mortality pat- tern. No such steady wend was observed for total malignant neoplasms, however. The SMR for all cancers combined dropped from 5S for the group 20-44 years old to 45 for the group 55-54 years old and then rose steadily to 81 for the group 85 years old or more. Examlnadon of age-specific death rates for specific cance~ sites (not shown here) indicated a high consis-. tency between the age profile and the corresponding summarized SMR of a given cause. Consider, for example, cancer of the prostate gland, with a mortality ratio of 94 that approximates that of the standard population. The age-spedfic death rates for clergymen were barely distinguishable from those for white males under age 85 years, after which a slight increase was indicated. In contrast, for intestinal cancer, exhibiting White Prolestant Clerical MortstIlty 1119 a lower SMR of 86, the age-spedfic curve wa~ exceeded by that of whites in practically all ag~ groups. Comparison With Other Profssslonsl Groups The death rates of the study group were compared with tho~e of several professional groups including U.$. white clergymen of all denominations. We com- puted the SMR using as a standard the 1950 age- specific death rates of U.& males of all races who were 20--64 years of age and had work experience (11). Data not shown here indicated that for all causes of death combined and for nearly all Sl:~'cific causes, the SMR for the study group was considerably lower than that based on white males, 20 years old and older. However, the SMR of 48 for overall mortality of our subjects was not much different from that of 61 for teachers and 52 for college faculty members, but substantially below that of 90, 91, and 85 for physicians, lawyers, and white clergymen, respectively. The ~lues for members of the teaching professions should be interpreted with due regard to the greater likelihood of reporting errors as a result of high turnover in this category of employment. For all cancers combined, an SMR of 50 was shown for the subjects compared with ~i for law,/ers and judges, 86 for white clergymen of all denominations, 81 for physicians and surgeons, 65 for teachers, and 48 for college professors and instructors (1]). breed e~ae of 4,eoXk cetegorv a~¢~ cl.rrml e~xi~l mdoje~, ~.0-~ Ue~ old, zg$o-eO C~u~e of desth' Ot~erved Expected S/~* demths de~ths All Major cm-diov~culsa~-mml d/sess~s (3~: ~; ~-6~) V~u~ l~io~ ~ c~l ne~u~ s~m (~) Di~s of h~ ~ rhe~a~c f~er (4~: 41~) ~k~t~ he~ ~m, inel~i~ eo~ d~ (4~) Am~ls~fic h~ d~ ~ d~n%M (420.0) ~her ~ms of ~ ~ (~0.1: ~0.2) H~n ~ h~ d~ (~) Genii ~le~ (460) ~nic and u~ifi~ neph~ ~d o~sr rsnal scle~h (592-594) M~f~nt ~pl~m~ includes n~pl~s of I~phatic ~d hedonistic ti~u~ (14~5) S~h (151) In~sfine ~ r~ Bl~der ~ o~sr u~ o~ (181) ~emia and ~e~emia ~mphoma (2~3: 205) TurqUois, ~1 fo~ (~1~19) S~hil~ ~ i~ ~uel~ (~9) Di~e~ melli~ (260) Inflc~ a~ pneumon~ ~epC pneumon~ of new~m (4~9~) ~ciden~ (E~E~2) Suicide (E~3: E~E~9) 1,002 1,78~.~ ~ (~-60) 601 1,095.0 ~ (51-60) '74 161.3 4-6 (36"-'68) 491 884.9 ~"/(52-62) 421 677.0 62 41 100.1 41 (29...66) ~0 577.0 6~ (~0-73) 29 7L3 41 (27-69) 12 13.1 9'2 (48-161) 5 10.8 46 (1S-107) 7 ~,,0 ~ (9--,~) 16~ ~.6 61 12 '29.3 41 (21-72) 88 67.0 49 22 28.5 7'/(4~116) 10 17,6 57 (27-105) 22 29.1 78 (4S-lI~) 10 12.8 78 (o"7-144) 3 5.9 51 (11-149) 11 10.7 103 (S1-184) 14 13.8 101 2 &~.6 6 (1-2~) 3 7.0 4~ (9-126) 13 33,4 33 (19.-~2) 9 24.3 37 (17-70) 42. 110.8 33 (~-50) 22 22,9 96 (60-145) • ,Vum/~,~ in ~~ are ICD (Seventh Revision) codes. # To~.si U.S. white cler~'men, ages 20-64 yr, 19~0. equal lC0. Numbers i~ /~a~-e~k.eses ind£cat~ 95% JN~. VOL. ~. TI09371708

King end Locke The study subjects, 20-64 years old, were also com- pared directly with the total U.S. white clerical popula- tion (table 9). In computing the SMR0 we used the age- specific death rates for this group of clergymen to generate expected numbers of deaths for the study population (11). Although mortality of the total white clergy in the United States was below average, it was still almo, t twice as high as that in the study group. Thus for all cause, of death combined, the SMR for study subjects compared with that for all U.S. white clergymen was 56; deviations from this figure for most specific cancer site~ could be due entirely to sampling variation in the small number of deaths. DISCUSSION An adequate interpretation of the extremely low mortality of the clergy requires more intensive investi- gations. Some possible qualifying and limiting factors may be noted first. One relates to the extent of inaccuracy in survival data that was likely to be small, as evidenced by the results of survivorship verification. For example, a random sample of 106 persons was selected from the `8,272 United Lutheran study subject¢ who were still on the church roster at the end of the 10-year investigation on August -81, 1960. They were matched agaimt the clergy list in the 1965 church yearbook. Nine had died or transferred their church affiliaiion after August 31. 1960, and 97 were alive as of October I, 1964. The vital status of these 97 subjects was further checked against other church document~ for its accuracy and was confirmed (12). The reporting errors typical of occupational studies would be minimal in an investigation such as this one. No information is available on the magnitude or even the net direction of misclasdfication of occupations, but errors in the basic census and mortality data might explain some of the rado differences in either direction, Another possible limiting factor is that observed death~ among the clergy were distributed over the entire study (1950-60) during which a general decline in mortality occurred, whereas age-specific death rates of the standard population were limited to a single year (1955). The effect of this on the calculated mortality risk for clergymen would be difficult m assess but is probably small. The effect of geographic distribution appeared neg- ligible. As indicated in tables $ and 4, more than two- thirds of the ministers had charges in the highly urbanized Northeastern region, and only one-fifth of them resided in areas with populations of less than 2,500. White male* in the Northeast generally have elevated'mortality rates for all causes of death and for coronary, heart disease and cancer (I4, 16). Therefore, it is unlikely that the geographic distribution of ministers would lower their mortality rates nor would it explain the large mortality differences observed. From dose examination of mortality differences between the clergy and other classes of workers, one gains the impression that the much depressed mortality jNc~ vow. ~. No. 5. NOV~-MBER may be a reflection of the low death rate generally observed in higher socioeconomic clasms, though most studies of such groups have not shown such differences as large as those noted here. For example, for all causes combined, total malignant neoplasms, and cardiovas- cular-renal diseases, the mortality level of study sub- ject~ 20-64 years old was from 46 to 52% below that for U.S. male~ in corresponding age groups with work experience. An extremely large deficit of 60% was shown for cancer of the lung. All such experiences were in keeping with the favorable mortality, observed in varying degrees among other learned professionals, such as physicians, lawyers, and teachers. One possible explanation would be that persons not in robust health may be less likely to enter or finish college and professional training. Other contributing [actors would include the subsequent elimination of unfit individuals and the high quality of medical care constantly re- ceived by those who remain. The noticeable differences in health ri~k exhibited between the study group and U.S. white clergy may be interpreted as mainly a reflection of.suix:laa deviations in mortality. Unlike the study subjects, the U.S. clergy is composed o[ persons of heterogeneous sodoeco* nomic standing and includes many part-time ministers who have other occupations such as retailing and woodcutting. One may also infer from the British experience, to be noted later, that similar inconsis- tencies might exist between census and church data on cleric populations. The generally less depressed, mortality from cancers of intestines, prostate gland, and pancreas, leukemia, and lymphoma observed in the study group calls [or further comments. To be sure the finding~ on these cancers are consistent with those reported elsewhere for the upper social classes, notably the British mortality experience (table 10)..Of particular interest is that a positive social gradient (e.g., a higher SMR for class ! than for class II) was indicated for British males when the age-adjusted PMK was used as a measuring index. The PMR has the advantage over SMR of circumvent- ing such intervening factors as occupation upgrading in reporting of deaths and social class differences in survivorship, inasmuch as it depends only on internal variation in the distribution of ¢xcupation by spedfic causes. • More recently, using the regression method, S. Devesa (unpublished data) demonstrated an association of cancer incidence with income and education; particu- larly for intestinal and prostate cancer*, she observed a positive association with education among white males even after adjustment xvas made for income. This finding is in keeping with the prostate cancer-educa- tion hypothesis reported earlier by Kitagawa and Hauser (I7), but contradictory observations are reported elsewhere (18, 19). The intricate relationship between income-education and mortality risk may be amplified to illustrate another dimension of the dynamic confounding effects of subclass characteristics noted elsewhere. For example, T[09371709

Wh|te Protestant Clerical Mortality 1121 Can=er si~ ~ehl cla~: 195~s $~ial cI~ 197~~ ~1 ~nce~ ~3 ~O 1~ 1~ 1~9 7~ ~ 91 113 116 131 ~ in.stOne 120 99 105 ~ 1~ 105 1~ 105 1~ I01 109 except ~m P~c~ I12 93 I01 I01 ~I I~ 97 I~ II0 I01 Pr~ gland 1~ 1~ ~ ~ 116 91 89 99 115 108 115 ~emia 1~ 1~ 1~ 97 108 113 1~ 1~ 101 104 95 ~mph~oma 110 96 101 98 114 123 101 105 1~ 108 A~es 65-74 rr PMR~ All eancen 95 99 101 100 I00 96 98 99 102 100 101 Ltrgt intestine 148 116 100 95 84 139 112 115 102 84 81 except rectum Pancreas 105 112 103 95 ~8 121 109 111 ~4 95 99 Pro~ttte gland 116 118 98 96 93 125 118 113 99 93 80 Leukemia 156 122 99 94 82 138 128 108 98 90 72 l.~vmphomreoma 156 127 98 91 86 134 115 124 103 85 79 • Dam were compiled from (9. 10), + Clm I i~ the highest elm. CLt~ IIIN=nonmanual worker~. Class lllMffimanual workers. c Simil~rly to the SblR, the proportion~I mortaliW me~ure is derived by ind|re~ standardization. only 25,% of the Baptist ministers in our study came from upper-class families, compared with 51% of the Episcopalian clergy (20, 21). This finding is not in keeping with the SMR indicated for the 2 groups of clergymen (72 vs, 69). However, the proportion of Baptist ministers with a full college education, semi- nary training, and graduate work was on par with their Episcopalian counterparts. Presumably, some of the relatively unfavorable health effects stemming from the ascribed status of Baptist ministers, based on "father's occupation." would have been attenuated by other positive influences associated with their achieved educational status. Again this observation is consistent with the findings of Kitagawa and Hauser (17) on mortality with respect to socioeconomic disparity, which suggests an interdependent relationship of multisocio- economic characteristics. One other possible compli- cating factor is that both the Baptist and the Episco- palian groups had a high proportion of ministers who had previous affiliations with other denominations. Further studies of the intricate mechanism of these interlocking factors are needed. From a different vantage point, one is tempted to suspect that occupation itself may have positive and negadve effects on an incumbent's health. However in most instances, the health effect of occupation can hardly be entirely separated from that of social class. As reported in a recent analysis of mortality variations between occupations iri England and Wales. on the basis of social class standardization, only 18,% of the overall difference between occupation orders was asso- dated with work or a majority of over 80% was related to social class (i.e.. life-style). The corresponding fig- ures were 1£ and 78%, respectively, for cancer and 30 and 70%, respectively, for circulatory and respiratory diseases (22). Admittedly, the interaction between direct occupa- tional effect and indirect influence of life-style goes beyond the oversimplified calculations cited above. The lack of sufficient information makes it particularly difficuh to speculate in specific terms on the confound- ing effects of both types of influences on the health of clergymen under study. Also, at the current state of knowledge, a number of interdenominational differ- ences in mortality cannot be resolved in the study population (table 11), such as the Presbyterian-United Lutheran disparity in mortality rates from cancer of the pancreas (SMR: 73 vs. 175) and diabetes (SMR: 102 vs. 48) and the Presbyterian-Lutheran (Missouri Synod) contrasts in lung cancer mortality rates (SMR: 20 vs. 51), all of which necessitate field im, estigadons. In conclusion, Protestant clergymen constitute a population at low risk of developing most diseases and cancers at various sites that may be partly a reflecdo'n of their being predominantly from the upper classes. In any case, further studies on levels of health and mortality among various clerical groups, with a focus on differences in life-style and cross-national compari- son, could provide new insights into the proverbial -longevity of clergymen. One would therefore need to 1) conduct macroscopic mortality and/or morbidity sur- veys of various "religious bodies and ministerial spe- cialties, 2) initiate comprehensive investigations of a retrospective and a prospective nature, and 3) assess the adequacy of clerical population and mortality statistics. A detailed proposal for such studies, and a suggested conceptual research scheme were pr~-,ented elsewhere [King H: Unpublished manuscript; (.o4)]. Brief com- JNCI. YOL ~. NO. 3, NOVEMBER T109371710

1122 King and Locke Cause of deaths United Lutheran- United Protestant American Mis~uri Presby- Episcopal" Baptist# Lutheran• Synod' terian" i All muses 73 Vascular lesions affecting central nervous 84 system (330--Eq4) Arter[oschrotic hear~ disease, including 76 coronary di~Me (420) Hyl~rmnsion with heart disease (440-443) 99 Maligramt neoplasms, including neoplesm of 79 lymphatic and hematopoietic tissues (140-205) Stomach (151) 41 Int~sfin~ and rectum (152-I~) 97 Pancreas (157) 175 Bronchus. trachea, and lung (16~-163) 27 Prostate gland {177} 102 Leukemia and aleukemia (204) 100 Lymphvma {20G-203; 205) 74 Tubereuloeis. all forms (001-019} 12 Diabete~ mellitus (260) 48 Influenza and pneumonia, except pneumonia 53 ct' newborn (480-493) Accident* (ES00-E962) 58 Suicide (E963; E970-E979) 45 71 72 72 69 73 82 70 76 79 76 76 78 59 71 59 68 62 59 61 62 63 46 39 66 60 73 73 69 NC 73 39 130 51 20 51 29 97 96 86 96 NC 103 83 NC NC 83 90 141 NC NC NC NC 88 102 126 47 49 76 84 65 48 58 44 51 NC 21 59 NC • Data were from King and Bailar (I~), King ('2I), King et al. (~$), and Locke and King" (20). b ~Vumbft~ in parentages are ICD (Seventh R~vision) codes. ¢ U.S. white males. >20 yr old. 1955. equal 100. NCffiSMR were not computed: deaths were fewer than 5. • Value~ are for 19~0-60. • Values are for 1951-60. ments are in order on cross-national studies and assessment of clerical statistics. The need for studies of clerical health and mortality at the cross-national level is well demonstrated by the noticeable mortality differential reported between such groups as the British and U.S. Catholic priests (table .12), the U.S. and British clergy of the Anglican Communion, and the total Anglican clerics in England and Wales as documented in government (census) versus.church statistics (table 13). As another example, the health of Catholic priests as well as Jewish rabbis has been studied least in any country. A systematic examination of both groups at the cross-national level would add some significant new. dimensions to our knowledge of clerical health risks. An adequate explanation of the wide mortality disparity shown for most diseases between U.S. and British clergy of the Anglican Communion would require, aside from the consideration of any possible statistical artifacts (see table 14), rigorous testing of a host of propositions, including geographic distribution (most British vicars sen, e in less urbanized com- munities), degree of role s~ress (British parish clergy- men were reported to be "doing far too much"), marital status (the proportion of married clergy is lower for the British compared with the U.S. clergy), socioeconomic origin (the characteristics of recently ordained British clergy were said to be much more typical of the general population), personal characteris- tics (consumption of tobacco and alcohol), perpetual tenure held by the British clergy, the much cherished eccentricities in teaching by vicars, the availability and quality of medical care in the community, and general environmental effects (21). The need for a c~itical assessment of the reliability of clerical statistics may be seen from the two sets of • markedly dissimilar SMR computed for Anglican cler- gymen; these sets were based on government and church statistics (table 13). Such a disparity seems to suggest possible errors in reporting on age and occupa- tion in census and mortality statistics and/or data gaps in the church documents (table 14). One would also suspect that the census definition of clergymen ex- cluded those who were not engaged in primary church TAI~LE 12.--SMR by broad age ~r~oup among Roman Catho~i~ prie~t~. Engh~ a~ Wa~s. 19~, an~ ~e~ ~ a Roman Caddie ~up. ~nited State~. 1958~7 SMR at ages:# Country _>20 yr 20-64 yr >65 yr England and Wales 109 103 112 United States 70 59 89 Data were from (8) a~d special tabulations furnished by Mndigun {.°5). SMR for b~th countries are computed on the bmis of age- sl:ecifio death rates of U.S. white males ages >_20 yr. 1955. ,.~NCI. VOL. ~5. NO. 5. NOYEMBER Tl09371711

Csun~y. profe.isn. ~d year SMR at a~:~ P.20 ~" 20-~ ~" ->65 ~ United Stat~: EpL~c~al cleft, y- 76 (71-79) 67 (~0--76) 78 (73-83) men. 1951-60 Engknd and Wales: Ckr~Tmen, Churoh of ]~ngland 1959, 1963-65,1967¢'" 70 (67-72) 80 (27-33) 93 (89-97) 1949-~3" I~6 (102-110) 78 (73-83) 114 (110-119) 1959--61~4 81 (79-83) 71 (68-75) 87 (84-90) 1949-68"`/ 107 (103-111) 87 (82-92) 119 (114-124) • Data were adapted from King (~I). b SMR ~re eomlm~d on the buts of age-~peeifi¢ death rat~ at U.S. whi~e mal~. _>20 yr old. 1955. ~Vumba~ in ;~ar~e=~ indicate 95% confidence limit& ~ Data are from the Church or bfedical Register. ~ Value includes thorn ~20 yr old. • Data are from sovernraent records. ~" Value excludes tho~ 20-24 yr old. TABLE 14.---Awrug~ annual dcatl~ rates per 100,0o0 ~ Of~ce of ~s~r ~ne~: Cen~ ~ of Fiance. ~e Chur~ of En;l~d: 195~ 1959, 1~, 1~7 Age. CIe~en. ~ Chu~h of ~a~s Cler~en ~ Dea~~ Engla~ No. P~" No. Per- No. Per- Ha. Per- ~nt ~nt ~nt cent Total 97,890 I00.0 3,064 100.0 104,064 100.0 2,346 100.0 <45 32,905 33.6 60 2.0 32.8~1 31.6 20 0.9 45-54 16,825 17.2 112 3.7 23,@8 22.6 88 3.8 58-64 18.880 19.0 382 11.5 22,000 21,1 252 10.7 65-69 10,130 10.3 386 12.6 6,712 6.4 191 8.1 70-74 8,045 8.2 452 14.8 6,710 6.4 337 14.4 >75 11,405 11.7 1,702 58.5 12,324 11.8 1,4~;8 62.1 • Data were adapted from King" (~I). b Age unknown was proportionately distributed over the known age ~-oups, activities. Interestingly, the inconsistency in SMR com- puted from government and nongovemment data is also shown in table 14 for British physicians: the rados based on the Medical Register data were, as with the .clergy, much lower than those based on government statistics. Presumably, these apparent inconsistencies between government and nongovernment data may not be confined to clerical and medical populations as presented here. One should keep in mind such a ~possible hidden bias in data in planning mortality :'~tudies in general and occupational groups in par- t'icular. RtqFERENCES II) R.~.%t~im.~l B. Disease of workers. The ~fin icum of 1713 ~'i~. with t~nslation and notes, by Wilmer ~ Wright (off,hal biling~i ~idon in ~tin and English published in 1940 as No. 7 o~ the History of Medicine Series). New York: Ha£ner. 1964, (2) Kt~G H. I~L~n JC IlL The health of the clergy: A review of demo~mph|c literature. Demography (3) KI.~G H. Health in the medical and other learned pttdesslons. J Chronic Dis 1970:.'~:~7-2~!. (4) Reglstrm- General [or England and Wales. Supplement to the 65th annual report o~ the Registrar General of births, deaths, and marriages in England and Wales, 1891-1900. Part I. Ion- dan: H M Sat Off, 19~7. (~)--. Morlallty o[ men in certain occupations in the three years 1910, 1911 and 1912. Supplement to the 75th annual ~pon at the Regisuar Gene~i for England and Wales. Part IV. London: H M Slat Off, no date. (6) ~. Occupational mortality, fertility, and infant monallty, England and Wales, 1921. The Registrar General's decennial supplement. Part II. London: H M Slat Off, 1927. (7) --. Occupational mortality, England and Wales, 1931. The Registrar General's decennial supplement. Part lla. London: H M Slat Off, 19a8. (8) --. Occupational mortality, England and Wales. 1951. The Registrar Genesal's decennial supplement. Par~ 11, Volt 1 and 2. London: H M Slat Off, 1957. (9) --. Occupational mortality, England and Wales, 1961. The Registrar Gene~l's decennial supplement. Tables. London: H M Slat off. 1971. (10) ~. Occupational mortality, England and Wales. 1970-1972. The Regim~r Generai°s decennial supplement, Part lla. l~n- don: H M Slat Off, 1977. (11) G~gAL';ICI~ L Mortality by occupation and industry among men 20 to 64 years at age, United States, 1950. Vital statistics special report (National O/rice of Vital Statistics). Vol 53, No. 2. 3. and 5. Washington. D.C.: U.S. Govt Print Off. (12) I~I~G H, B.',lt.~g JC lIl. Mortality among Lutheran clergTmen. Miibank Mere Fund Q 1~8:46:527-548. (13) National Office of Vital Statistics. Mortality (tom selected cauls by age. race, and sex. United State~. Vital statistics special report. Vol 46, ,'go. 5. Washington, D.C.: U.S. Govt Print Off. 1955. (14) G~tno.,;T, C~rrr~nr.,,~ M, I-L~SS~L W. Cancer mortality ~rends in the United State~. 1930-1955. Nad Cancer Inst Monogr 196h&133-350. (1~) MA.'rr~L N. Tabular values of 95 percent ~onfidence limit tars for estimates of a Poisson-distributed variab|e (Appendix C). In: Haens~l W. Loveland DB, Sirken MG. Lung<ancer mortality ~ related to residence and smaking habits. L White ma~es. ~ :gad C, aneer In~t i~2:28:947-1~31. (16) E.'~'-~RL~Xr-PE. STeWArt WH. Geographic patterns in j~cl. VOL ,~. NO. 5. NOVEMBER 1980 T[09371712

1124 £rc~ c~r~nary hart ~iL~-a~-_ (17) ~A~A~VA ~.i, ~t~ PM. D~f~e~nt~l m~lhy in th~ Uni~ 5~. ~ph6dg~ ~ Univ ~s. 1973. (18) ~ss~z VL, SZLW~ S. SAc~ ~, Avs~ DE. B~o~'~ SM. mt~ by ~ a~ ~al c~, ~ J Epldemiol 1978;107:311- 319. (19) ~K GW. TOy~A JA. ~tion and mortality in Wash- ington ~anty, ~land. J H~hh (~0) ~ ~. ~G H. Mo~lily among ~pt~t cler~. J Chmn~ Di~ ~) Fox AJ, ~ AM. ~cu~or~l mo~lh~ Work or ~y cf life. J Ep~iol ~mmunhy Heal~ 1978~7~-78. (23) ~yo H. Z~os G. H~ ~ Further inqui~ in~ Pro~nt cle~l ~o~]ity ~t~em~ J Bi~ ~ 1975;7:243-~4. (24) ~z~ H. W~t ~ know a~ d~n'[ kn~w a~ut the heal~ of cler~. In: Nix ~. F~h~ CJ. ~, S~na for A~to~te. Washington, D.~: ~n~r for Appli~ K~arch ~e A~[olate. 1970. (25) ~vI~A~ FC. Role ~tisfaaion~ and length of life ~n a clo~ j~ci. ~.'OL. 63. .~0. 5. NOVEMBER If~.3 T109371713

Cancer Patterns in Ethnic Groups Moderator: Kenneth Weiss, Ph.D. TI09371714

-- Cancer Patterns of Four Ethnic Groups in Hawaii Laurence N. Kolonel a.4 ABSTRACT~Cancer r~ek$ wsr~ tempered in the Jepane=e, Chlne=~, H=waHan. and Fll~ptno population= of Hawaii w~th t~se In the U.S. white ~pufafion. The a~ls u~ s~ndardlzed Inctdln~ ~tlo(=) (SIR), ~s~ on ~e U.S. whl~ i~cidence rates for 1070. In general, the rliks for m=ny miJor ~ncers In the Jitney, Ch~ne~, and Filipinos were low In re;affon to U.S. white, whereas tho~ for Hawaii=n= were h~gh. However, the SIR ~ue= for the vsflou= ~ncer sites and ~p~ varied ¢onslder~ly wi~ln e=~ e~ni¢ group. Ethnic~peciffc data on ~oklng, drinking, and die~W habits and on s~lologl¢ and demographic factor= wine e~mln~ in ~latlon to t~e SIR findings. Some of the ~=ul~ ~re consistent with cu~nt etiologic hypotheses, where~ othem =uggosted ames for fu~erepldemiologic =tudy.~ JNCI 65: 1127-1139, 1980. Among the 5 main ethnic groups comprising 95% of the population of Hawaii, i.e., Caucasians, Japanese, Chinese, Filipinos, and Hawaiians, the patterns of cancer incidence and mortality eary considerably. Many of the site-specific incidence races for the nonwhite groups are significantly lower than those of the U.S. mainland white population. An examination of these rates in relation to corresponding dam on smoking habits, alcohol consumption, and diet, as well as socioeconomic and anthropometric measures, makes possible a number of inferences about potential causes of cancer at various sites. METHODS Incidence cases of cancer in the entire State of Hawaii have been recorded by the Hawaii Tumor Registry since 1960. By 1967, the first year of the 10- year period examined in this study, ascertainment of cases throughous the State was almost entirely com- plete. Few cases (<1%) are picked up only by death certificate, and more than 94% of cases overail are histologically confirmed. Diagnoses were classified by the ICD-O (Geneva: World Health Orgahlzation, 1976, first edition), and all cooling was done by Registry staff. Inasmuch as the population of Hawaii is less than I million and is divided into several ethnic subgroups, it does not yield sufficient numbers of cancer cases for analysis of short intervals. Hence this report examines all cases diagnosed during the 10-year period from the beginning of 1967 to the end of 1976. Expected numbers of cases were computed with the • use of incidence rates from the Third National Cancer Survey (I) for whites on the U.$. mainland. These rates were applied to estimates of the size of each ethnic popula6on in Hawaii. The estimates were ob~ined from the HSP of dee Hawaii DOH rather than from the U.S. Bureau of the Census, because census forms in 1970 did not permit an appropriate classification of the popula6on of Hawaii by edmidty. The HSP sample comprises 2-3% of the households in Hawaii and is drawn randomly from continually updated Iistings of housing units representative of the entire population. The HSP population estimates were derived with an expansion formula that made use of the census totals in each category of age and sex. Inasmuch as the Filipino and Caucasian populations have rapidly in- creased in recent years, two midinterval estimates were used in the computations: the 1969 and 1974 estimates for the first and second 5-year intervals, respectively. (Population estimates for 1974 are given in Appendix table 9.) Expected numbers of cases in each category of age, sex, and ezhnicity in each of the 5-year in- tervals were summed to ~ve the overall expected numbers; observed and expected nuJnbers were then used to compute the SIR. Statistical significance of the ratios was assessed under the asumption of a Poisson dis~xibution (2). The patients classified themselves on ethnicity at the time of hospital admission, whereas the population-at- risk (surveyed by the HSP) was classified by ethnicity based on parentage. The effect of this difference in method of classification is probably minimal. Only persons of mixed ancesm/ could be allocated differ- ently; with the exception of individuals of part-Hawai- ian ancestry (who are grouped with the pure Hawai- ians by both the Registry and the Hawaii DOH), more than 90% of the population is ethnically pure. The percentage is even higher in those age groups among whom most cancers are diagnosed. The variation in the SIR was evaluated in the light of exposure data obtained from spec/al studies based on the HSP population sample. Exposure rates of each ethnic group were age adjusted by the direct method with the 1970 U.S. white population as the standard, because this population provided the cancer incidence rates for computation of the SIR. A||I~VMt'I"ION$ USED: DOH,=Dcparunent of H~I~; HSP=Hmhh Su~eil~n~ ~o~m; ICD-O=~t~o~l Classifi~ of Dis- ea~ for Oncolo~; NOS=not o~e~ise s~[i~: SIR=s~n~i~ incidence ~s). z Presemed =t the Workshop on Populations at L~w Rhk of ~er co~uc~ at Snowbi~. U~h, Aunt ~-~, 1978. ~ Sup~ in ~ by ~bl~ H~hh ~ {PHS) ~n~ I-N01- ~15~5 and 1-R01-~0897 [r~ the Natio~l ~n~ Institute (~) and by PHS ¢on~ct N01-O5~511 h~ t~ Di~i~ ~ ~¢er ~¢ a~ Pr~tion. NCI. = Epi~io[o~ ~o~m. ~n~r ~nt~ ~ ~ii. Unit~nity of Hawaii. I~ ~ala S~.. Honolulu. ~ii ~813. ' I a~nowl~ ~e ~l~bIe assis~ of Ms. S~n 51u and Ms. Virginia Siebu~ in ammblin$ ~e ~u [nr this ~n. 1127 .]NCI. VOL ~. NO. 5. .~OX%MBEK I~) T109371715

1125 RESULTS The observed and expected numbers of cases and the SIR in each category of disease are shown in Appendix rabies 1-4. The SIR of the more important categories ar~ summarized in ruble l, the main features o£ which are: 1) The SIR for all cancers combined among Japanese men and women and among Chinese and Filipino men only were significantly lower than those among whites, whereas those among Hawaiian men and women were significantly higher than those among whites. 2) Among males, the rates of four cancers (lip, prostate gland, and bladder cancers and melanoma) were significantly lower in all 4 ethnic groups. (In other words, U.S. white males are at significandy higher risk for developing these four cancers than are all 4 ethnic groups in Hawaii.) Among females in these ethnic groups, only the rates of melanoma were significantly lower than the rates among white women. 3) Besides the four cancer sites just mentioned, the onIy site for which Hawaiians were at significantly lower risk was the colon (both sexes). 4) Filipino men had lower risks of developing cancers of the colon, pan- cr~s, lung, larynx, kidney, testis, and brain and of developing lymphatic leukemia, in addition to the four cancer types noted above. Filipino women had signifi- cantly lowe~ risks of developing melanoma and cancers of the colon, I~r~ast, cervix (invasive), and bra~n. 5) Besides the four cancers mentioned above, the tyg~-s for which Chinese men were at lower risk were: lympho- sarcoma, Hodgkin's disease, leukemia, and cancers of the lung, larynx, mouth, and tongue. Besides mela- noma, the ovary is the only site for which Chinese women were at a significantly lower risk than white women. 6) Japanese men and women had more sites of low risk than any other ethnic group. They had significantly lower rates of cancers of the mouth, lung, larynx, bladder, kidney, and brain, as well as lympho- c/tic leukemia, melanoma, and Hodgkin's disease. In addldon to these nine and the other cancers for which all ethnic males w~re at lower risk than whites, Japanese men had significandy less pancreatic and testi~ular cancers and lymphosarcoma. Besides the above nine cancers which were low for both sex~s, Japanese women had siguificantly less granuloc/tic leukemia and cancers of the colon, breast, cervix (invasive), and ovary. 7) The SIR were significantly higher for a number of sites, but I will not comment on this finding because Dr. Fraumeni discusses these high-rate sites elsewhere in this monograph (~). In addition to table 1 that summarizes the detailed data on observed and expected numbers (Appendix tables 1o4), Appendix tables 5-8 show age.specific incidence rates of cancers at sites for which at least 100 cases were L--SIR for #eleaed c~ncers in H~zg~aii b~ s~ and ahnidt~ of patient Female" Sit~ or WI~ of ~neer Japanese Chinese Filipino Hawaiian Japanese Chinese Filipino Hawaiian All malignant neoplasms 84# 75# 65# 127 90# 110 93 147 Lip 9# 11# 0# 0• 82 0 0 0 Tongue 71 14 t ~5 I~ 108 160 43 171 Mouth and gum 43• 11# 118 163 43# 24 237 169 Nuopharynx 188 1,500 270 792 94 1,833 667 40~ Esophagu~ 103 150 144 373 51 33 120 263 Stomach 341 87 80 373 405 196 169 527 Colon 103 103 69• 64• 85• 96 56* 70# t R~tum 131 132 108 84 109 101 86 74 Pancreas 74# 90 ~6~ 168 80 92 46 131 56# 42• 25~) 0 Larynx 58# 63• 31• 101 48 1~2 Lung 38# lf~4 79# 210 143 218 Skin, melanoma 9• 20~ 31• 51" 7 I0• 14# 25• Brmt 51 0 25 ~3 75 • 91 47# 114 Cervix. inv~iw NA NA NA NA 62# 86 65~ 120 Corpus uteri NA NA NA NA 108 133 99 196 Ovary NA NA NA NA 55# 62~ 70 100 Prorate gland 69~ -~0# 49# 71# NA NA NA NA Testis 34# 29 14# I01 NA NA NA NA • Bladder 66• 43 # 27# 48# 66# 80 78 124 Kidn¢y 56 81 56 77 Brain 41# 65 62" 76 45# 71 17• 85 Th~'oid gland 187 250• 365 202 153 246 399 233 " . Lymphoaareoma 73" 33# 104 101 83 112 121 145 Hodgkin's di.s~a~ 39* 34¢ ~ 66 65 37b 33 53 75 ~aaphocyti¢ leukemia 3~• 33~ 51b 65 51~ 4~ 99 58 Gcanuloc~ic leukemia I00 ~0" 99 95 70' 133 158 I07 NAb=not applicable. SIR was sig'aificantly low (P<0.01). SIR was significantly l~)w (P<0.05L j.xo. rOL. ~. .xo. ~. NOVE.MBER Tl09371716

Cancsr Patterns In HawaH 1129 7 Fema!e ,)'almnese Ch/n~e Filipino Hawaiiz. Cauc~ian Japan~ Chinese Filipino Hawaii~.n Caue".~sian Median family in- 16.242 16.412 11,127 11.711 12.~7 14.~S 16,424 11.112 9A30 Cclle~ ~u~ % ~.9 46.8 19.7 12.7 41.9 ~.5 ~.4 16.9 9.2 Het~ in~ ~.1 67.1 ~.0 68.4 69.8 61.0 62A ~.9 63.8 64.2 Wei~h~ ~unds 148 144 142 181 1~ 115 115 119 1~ O~iW index~ 3.4 3~ 3.4 3.9 3.4 3.1 3.0 32 3.8 Ci~ smoking C~nt + ~- 52.7 ~.6 47~ ~.3 61.0 24.6 17.3 27.8 ~.1 48.9 smcke~. ~ U~ ~ong ~ 16.2 162 20.5 24.9 13~ 11.2 13.7 16.2 18.4 smoke~, ~- • Data have been age adj~tod to the 1970 U.$. white population. ~ Formula for this index is [weight in pounds/(height in ineh~/] × 100. Ale~ol and dietary intake Male F,male Jaim- Chi- Fill- Hawai- Cauca. Japa- Chi- Fili- Hawai- Cauea- nes~ ne~e pino Jan sian nese n~e pino ian sian Beverage consumption >6 ear~ ~r/w~ % 19.6 7.9 17.7 26.5 2L5" 0.8 0.3 1.3 7.8 4.2 ~6 g~ ~ li~or/wk, % 4.8 4.1 3.2 5.4 1L6 1.3 1.3 1.1 2.5 8.8 ~5 ~e~ e~ol/wk, % 14.8 5.6 12.6 24.7 23.6 1.6 0.8 0.7 5.5 8.7 >2 eu~ ~ffe~y, % 26.8 13.0 15.9 ~.1 37.0 19.1 8.5 12.7 17.7 31.5 F~ ~n~pfion B~ ~wk ~ ~2 ~5 ~ 371 189 198 152 ~ ~9 Por~ ~wk I~ ~1 151 1~ ~9 ~ 149 117 139 97 Dri~ ~ ~h, ~wk 3.9 3.6 5.$ 9.3 1.0 2.4 1.7 3.2 3~ 0.3 M~ ~ ~d~ nixie, ~wk 189 152 137 201 172 118 102 91 116 112 Ri~ ffwk 2~ I,~ 2,~1 1,~7 5~ 1,425 1,146 1,982 ~7 377 0~en~ pickl~ ve~mbl~, tim,/wk 3.7 1.1 0.9 1.0 0.6 3.9 0.9 0.5 0.6 0.4 ~w ~ge~les, fim~/wk 6.3 4.6 4.3 ~.7 5.7 6.7 5.3 4.4 6.0 6.1 Fmh f~iL Ume~wk 5.0 5.6 4.6 4.7 5.2 5.8 6,9 5.1 5.3 ~.8 Nu~ent in~ke To~l ~, ~y ~ ~ ~ 70 71 ~ 51 47 52 53 ~im~ p~, g/~y 40 ~ ~ 49 52 31 ~ ~ 37 39 To~ ~ ;/~y ~ ~ ~ 76 ~ 51 ~ 46 57 64 ~L g/~y 37 41 ~ 49 ~ ~ ~ 28 35 36 U~n~. ;/d~y ~ ~ ~ 40 45 29 30 24 31 35 C~i~I, m~y 317 309 319 ~0 368 247 253 2~1 ~ ~4 ~rb~ ~id (t~), ~/day 103 115 ~ 1~ 119 113 113 ~ 1~ 110 • Data have been age adju~tod to the 1970 U.S. white population. recorded during the period of study, 1967 through 1976, and Appendix table 9 shows the age distributions of the ethnic populations of Hawaii. Data on smoking habits, alcohol cor~umption, diet, sodoeconomic status, and anthropometric measure- ments of the,e 4 ethnic groups and of Caucasians in Hawaii are given in tables 2 and ~. The information in these table~ can be used in a qualitative manner in the evaluation and interpretation of the SIR in table I. DISCUSSION Low-risk sites for all 4 gro-pz.~Cancers that pc- cured at lower rotes than expected among all 4 ethnic groups may reflect increased genetic suscepti- bility of the U.S. white population from whom the expectations were derived. For example, lip cancer and melanoma risks are suspected to be increased by exposure to intense sunlight (4, ~). Thus the low rates of these cancers in all 4 groups may be explained by their generically determined, heavier skin pigmenta- tion, which affords them more protection from solar radiation than is enjoyed by Caucasians. Support for this notion may be found in the significantly elevated SIR of Hawaii Caucasians, i.e., SIR o[ 151 and 227 among men and women, respectively, for lip cancer JNCI. VOL e~. NO. 5 NOVEMBER I.~.'30 T109371717

1130 Kolonel and corresponding SIR of 185 and 150 among men and women for melanoma. A second explanation for uniformly lower ris-ks at some sites might be an infectious agent, exposure or susceptibility to which is lower in Hawaii than on the U.S. mainland. For such an explanation, the risks for Caucasians in Hawaii would also be expected to be lower. Of the cancers with lower risks in all 4 comparison groups, only Hodgkin's disease and chronic lymphocytic leukemia showed noticeably lower (though not statistically significant) SIR among Caucasians in Hawaii as well. An infectious agent has been suggested [or both of these cancers, particularly among children and young adults (6-9). The younger age groups do account for much of the overall difference between observed and expected numbers of cases of these cancels. In one report (10) a lower risk of developing Hodg- kin's disease and lymphatic leukemia was associated with a history of common contagious diseases of childhood as well as with coffee drinking and heavy cigarette smoking; both cancers had parallel predictive factors. Data from a serologic survey among children in Hawaii indicated low levels of rubella immunity. However, the prevalence of Epstein-Barr virus antibody in children in Hawaii (particularly among Asians and Hawaiians) was higher than in children in Connecticut (11). This virus has been implicated in the etiology of Burkitt's lymphoma and nasopharyngeal carcinoma 112, 13). A third explanation for uniformly lower risks among the 4 groups would be differences in exposure to general environmental factors in Hawaii in compari- son with the U.S. mainland. Hawaii does not have significant amounts of air or water pollution, for example. Differences in occupational exposure or in personal habits (such as diet and smoking) must also be considered. Hawaii is not heavily industriali2ed, so that uniformly lower risks for cancers with known or suspected occupational etiologies (e.g., bladder cancer and leukemia) would not be surprising. Although bladder cancer risks among males of these 4 ethnic groups were lower, Caucasian males in Hawaii had a significantly higher SIR (119) for bladder cancer. Inas- much as bladder cancer is also associated with smoking (14), this difference may be related to smoking habits (table 2); not only were more Caucasian males smokers, but also their average cigarette use was higher. Further- more, in comparison with mainland whites, those in Hawaii ~ppear to smoke more [45% of male Caucasians in Hawaii are current smokers compared with 39% on the mainland; corresponding rates for females are 37 and 29% (13)]. Also consistent with the bladder cancer pattern is the observation that Hawaii Caucasians drink more coffee than do the other ethnic groups (table 3), an association that has been reported (16, 17). The prostate cancer pattern parallels that for male bladder cancer, including an elevated risk for Hawaii Caucasians (SIK--19_9). In addition to coffee. Hawaii Caucasians consume more beef than do the other Jxct. VOL. 65. NO. 5. NOVEMBER ethnic groups and have higher intakes of fat (total and unsaturated) and perhaps of animal protein (table Thus the pattern in these data is consistent with other investigators' observations that patients with prostate cancer appear to have a higher intake of foods rich in animal fats than do matched controls (18). Low-risk sites for the 3 Asian groups (Japanese, Chinese, and Filipino).~One striking anthropometric observation in Hawaii is the generally slight body build and absence of obesity among the Asian groups and the greater stature and high prevalence of obesity among the Hawaiian group. A comparison of body measurements among these groups is shown in table 2. The obesity measure (Quetelet's index) clearly distin- guishes the Hawaiians from the other groups. Al- though heredity may contribute to obesity in Hawai- ians, diet no doubt plays a significant role (table 3). Hawaiians consume greater amounts of protein, fat (total, animal, cholesterol), and ethanol (particularly from beer) than the Asians. Furthermore, a greater proportion of them have ever smoked (table 2); these smoking differences may explain three of the cancer sites for which the $ Asian groups have low risks: larynx, the lung in males, and the bladder in females. Testicular and ovarian cancers, which also fall into this grouping, ought to be studied for possible dietary etiologies. Ovarian cancer mortality has been correlated with national consumption of fats and oils (19). Heath and Daniel (20) reported that testicular tumors occurred in animals given injections of cad- mium, and Schroeder et al. (2I) determined that diet is the main source of exposure to this element for most people. One food with a high cadmium content is oysters, which are consumed less in Hawaii than on the U.S. mainland where they are harvested. Oysters are also eaten more commonly in Japan, and the rote of testicular cancer among Japanese in Japan is higher than that among Japanese in Hawaii (22). The low risk of developing invasive cervical cancer among Asian women and the correspondingly high risk noted for Hawaiian women may be related to ethnic and cultural differences not only in sexual behavior but also in attitudes toward medical care (23). Low-risk sites [or either the 2 Oriental (Japanese and Chinese) or the 2 non-Oriental (Filipino and Hawaiian) groups.~The Oriental population and the Filipino-Hawaiian population in Hawaii differ most notably in socioeconomic status. Of the 4 groups, the Hawaiians have long been economically most de- pressed. The Filipinos are the must recent migrants to Hawaii, which characteristically places them low on the socioeconomic ladder as well. Table 2 includes data on education and income that support these observa- tions. Socioeconomic status alone, of course, is not an etiologic factor, but it may be indirectly related to various important exposures. One such relationship is occupation. For example, many Filipinos and Hawai- ians are employed in agriculture; the proportion of the Filipino work force engaged in agriculture is four dmes greater than the overall average for the State (24). T109371718

Whether lower than average ~¢p~sures to such agents as p~sticides can explain the lower risks of devdoping lymphosarcoma and granulocydc leukemia among the Oriental maies has not been determined. Com-incing evidence of human cancers' being attributable to pesti- cide exposure is lacking, hut cardnogenidty of these substances has been demonstrated in animal experio ments In addition, some dietary patterns common to the Filipinos and Hawaiians are not shared by the Oriental groups. These include a higher intake of dried salted fish and a lower consumption of fresh fruit by the Filipinos and Hawaiians. Hawaiian and Filipino men have lower risks of developing colon cancer, whereas the women have lower risks of developing rectal cancer. Cancers of both of these sites prohably have dietary etiologies (26, 27). Other low-risl~ combinations.mA few of the ~main- ing low-Hsk patterns are worth noting in relation to the available exposure data. For most groups, includ- ing Hawaii Caucasians, stomach cancer SIK are ele- vated significandy. A dietary edology is considered most likely for cancer of the stomach (gS, 29) and is consistent with "the higher rates among whites in Hawaii, who consume more Oriental food than do whites on the mainland. Ingestion of nitrite-containing tood~ in particular has been implicated in the edology of this cancer on the basis of presumed nitrosamine formation in the stomach {30, 31). Some comparative data on meats with nitrite additives are included in table 3. The intake by Chinese and Filipino men, whose SIR for stomach cancer are low, appears to be notably less than that by men in the other groups. Because ascorbic add inhibits the reaction of nitrites and secondary amines to form nitrosamines (32), intake of this vitamin is worthy of careful observation. How- ever, the data in table $ do not indicate that higher intake of this vitamin by Chinese and Filipino men is an explanatory, factor for their lower risks. Chinese and Filipino women have low SIK for gallbladder cancer in contrast to the elevated values for Japanese and Hawaiian women. Although cholesterol intake is lower for Chinese and Filipino women than for Hawaiian women, Japanese women also have a low cholesterol intake. However, coffee consumption by Chinese and Filipino women is less than that o[ both Japanese and Hawaiian women. These observations attract our attention because cholesterol gallstones have been associated with an increased risk for 'gallbladder cancer (33), and a higher coffee consumption is associ- ated with elevated serum cholesterol levels (34). Whereas Japanese and Filipin~ women have reduced risks for cancers of the breast and pancreas, those [or Hawaii Caucasian and Hawaiian women are elevated. In contrast, Chinese women have risks of developing these cancers similar to those of the rderence group, U.S. mainland white women. These observations imply the possibility of a common etiologic factor for these two cancers, both of which have shown a gradual increase in incidence among women in the United Cancer Patterns In Hawaii 1131 States over ~he past 40 years (33). An association wi~h dietary fat is suggested by the intake patterns of total fat, unsaturated fatty adds, animal fat, and beef among these groups (table 3), The interesting patterns of low SIR among these -~ ethnic groups discussed here and their correlations with various exposure data support the belief that environmental agents cause most cancers. The frequent and plausible assodations with dietary factors in par- ticular offer further evidence that diet and nutrition are important elements in the causal web of cancer. Several of the observations in this report can perhaps serve as the basis for further epidemiologic research in this REFERENCES (I) CL'Tt.~t SJ, YOt*.~G JL Jit. ed~. Third National ~er Su~.ey: Incide~ ~m. Nati ~nc~ Inst Mono~ 1975:41:1~M. {2) ~ JC III, ~R~t F. Si~ifi~nce factor$ for t~ ratio of a Poison ~ble to i~ excretion. Bi~etfi~ 1~;20:~ {3) F~v~lg~l JF Jt. High-rite ~nc~ among Iow.fi~k ~pulations. JNCI l~0;~:118~-1189. (4) ~Xt~s K. lncid~ce of ~ii~n* ~lan~a of ~e ~kin in [i~e No~ic countff~: $i~ifi~nce of ~r ~afion. lm J ~ncer (~) ~ .~ ~llu~r ryes, s~i~l, ~ce. ~d~ty. ~u~fions, habit, a~ as~iated di~ in the ~tho~is of lip can- cer~ .~ J Epid~iol (6) ~C5~o~ B. Epld~iolo~ of H~g~n's ~. ~nc~ Res 19~:26:1189-12~. (7) ~om~ ~. H~v~ B. H~[~n's di~se cl~tering in famiHo ~d communi~. ~ncer K~ 197~:~:1161-1168. (8) G~N RW, B~o~ ID. GR~HAS~ $. et ai. ~emla in chii- ~n ex~ to mul~ple ~sk [acm~. N Engl J M~ l~;~ (9} At,~ OF. ~ S. DwO~Ky R, H£~D~ BE. E~s leuke- mia in coho~ o[ child~n ~m [ollowing influenm epidemics. ~ J Epidemiol 1975;101:77-88. UO) P~a~¢R ~ JR. W~ ~ HYw RT. ~a~c~tics in you~ pr~cti~ of aduli~n~ ~li~ant l~ph~as, mela- n~as, and leukemias. J Had ~er ~st 19~8:~:8~. (11) J~n~In6s ~ ~nce of EB ~rm anti~y in Hawaii. Ph.D. ~is. New ~ven: Yale Uni~iff ~1 of M~i~ne. (12) Jtmm~ SC. H~s~ W, ~n~ ~. A clm~r of Eps~in.~ff ~rus-as~ted A~n Bu~kiit's I)~ph~. ~ Engl J i 977:~7:~468. (1~) Ho JC. N~ MH, K~t~ HC, ~at" J~ ~s~in-~ff-~-irus-s~- cif~ I~A ~ IgG ~ anti~i~ ~n n~phaDm8~l n~a. Br J ~n~ 1~6;~4:~. . {14} w~m JM. Dvs~ JK JR. Smoking and m~lity: A prm~tive seu@. ~r 197~:105-112. (I5) Morbi~ty ~d mo~lity w~y ~rt. 1977;26:1~ [DHEW publi~tion ~o. (CDC)77~017]. (16) BRo~ ~, TIDINGS J..~othe~ l~k at coff~nking and ~n- cer o[ ~e ~naff bla~er. Prey M~ 197~;~5-451. (17) Sl~o~ D, Y~ S. ~ P. ~f[~ drinking and ~ncer lower u~ tract. J Nail ~er lnst 1975;~:587-591. {I8) Ro~ls ID. Smd~ in the epidemiol~y of ptosmle ca~er: ~ sampling. ~ncer Trot Rep 1917:61:17~i80. {19) ~A AJ. Dietary factors ~ated with d~lh rotes from c~min n~plasms in man. ~ncet 1~:2:~2-3~. (20) ~ JG. DA~L MR. The pr~uction of ~li~ant t~ou~ by ~mium in the rot. Br J ~ncer !~:!8:124-1~. trace elem~ts in man: Zinc. Re~tion to environmental mi~. J ~ronic Dis 1~7:~17~210. ~) ~K JA. H~osI'GI 5I. ~s~x G~ ~ in m~lity lrom jxo. yOU ~. NO. 5. NO't'E~,IRER I~0 TI09371719

1132 Kolonel v..zm~rs c[ &e r.~.s in ~a~n. 1~7-70. J ~ad ~=c~ ~ ~hu~s in Hawaii--an in~ucdon for mmml h~l~ wo~ ~nolulu: Univ Hawaii, 197~. (2#) ~ Matke~ ar~ KmpIo~ent ~e~ce ~h ~ction. [o~ infa~fion [~ ~fi~a~ acdon pr~. S~te o[ ~ii, Honolulu. ~ii: ~t~em of ~r and ~- dm¢~l ~ladons, 5~y i~. (2~) ~v ~ ~u~donal ~et tis~ for ~stici~ ~tk~s. ~- (26) ~m~ 5~. Diet as an etiologi~i [actor in ~e developm~t of ~ of ~e col~ and ~tum. J ~nic Dis 1975:~:67-80. (27) Moo~s 8, ~ V, ~m~ F. MoD~ M, G~E~a~ZG ~. G~H~ S. ~w-fi~r in~k~ as an etiol~ic factor in ~n~r of the ~lon. J Haft ~ncer Inst 1975~5:i5-18. (2~) ~'~ T. A stay o[ epidemiol~ of stomach can~r, wi~ s~hl RfeRnce Io ~h~ e[[~c¢ of ~e ~et hcto¢. Bull Inst PukHc Heahh fJpn) 1~63;12:~-~. (29) ~L~L W, Ku~RA~ M, S£cl ~L ~ ~ S~ach ~ncer (39) B~E~ P. Nhd~, ni~ro~n~ and ~n~r. F~ ~ 1976;35: 1~-1326. epidcmiol~y. ~ncct 1975;2:5~. (32) WEISaU~G£~ ~. Vi~min C and prevention of nitro~mine for- ~ti~n. ~n~t 1977;~. (33) ~tT J, M~ B. SUASl M. ~olelith~i~ in the a~ol~ of ~llb~dd~r ~op~sm~. ~ncet (34) ~OADS GG, ~&~ A, Y~O ~ ~tion~ ~tw~n ~cton a~ p~ li~p~in~. ~rcutation 197~:~ ~uppl ($5) D~ SS, S~L~A~ DT. ~ncer indd~ce and ~en~ in the Unit~ Smte~ 1955-74. J Nail ~ncer ln~t 1978; APPENDIX AI, PR,m~IX T~Lg 1.---Observed (Ob~) and expected (Exp) numbers and SIR for cam bu site occurring among males and females: Japanese in Hawaii. 1967-76 Primary site* Male Female Obs Exp SIR Obs Exp SIR Total malignant neopl~ms 2,997 3.581.0 85' 3,010 3,358.4 90~' Lip, oral cavity, and pharynx 121 Lip 4 Tongue 2.5 Salivary glands 31 Gum and mouth 19 Nasopharynx 15 Other and unspecified pharynx 27 Digt~tive organs and peritoneum 1,350 Eiophagus 51 Stomach 473 Small inteltine 11 Colo~ 848 l~etum, rec~z~igmoid junction, and anal 246 ~ntl Liver 69 Gallbladder ll O~her billary l~ges 41 Psncreu 93 Retrop~rltoneum. peritoneum, and un- 7 specified digestive organs Respiratory system 514 Lxr~x 52 Lung 444 Other r~iratory organs including pleura, , 18 mediaatinum, tnd intrathora~ie site Bones ~d joints 6 Connective. subcutaneous, and other ~oft 18 tiMues Skin. melanoma ' 5 Brea~ 4 Female genital system Cervix, in situ Cervix Corpus u~eri Uterus NO$ Ovary Vulva and cliteria Other fema|e genital 190.1 64' 60 6'7.8 88 45.8 9' 1 3.1 32 35.2 71 15 13.9 108 18.0 28 1. 44.1 10 23.1 8.0 188~' 3 3.2 94 44.1 61' 3 13.2 23~ 926.3 146' 929 732.0 127' 49.4 103 8 15.8 51 138.7 841~ 305 75.3 405~' 10.7 103 5 7.7 65 338.9 103 292 345.5 85 187.5 131~ 145 131.3 109 30.3 228s 31 15.8 196" 11.1 99 39 24.7 158.~ 16.8 244~ 33 12.3 268' 126.1 74~' 68 84.6 80 16.6 42< 5 18.7 27 877.3 59~' 155 200.0 78* 92.4 56' 3 11.9 25' 769.7 58' 142 180.3 79' 15.3 118 10 7.6 132 9.4 64 2 7.5 27~ 24.2 74 14 19.1 73 53.7 9~ 4 56.7 7.8 51 732 976.5 75~ 685 707.4 97 152 121 195.2 62~ 286 £64.3 108z, 1 29.8 100 182.0 553, 7 6.9 101 17 21.3 80 1 7.5 13' • Diag~ were clarified by the ICD-O ' P<0.01. ¢ P<0.05, JNCL VOL. ~. .~O~ 5. .XOVEMBER lf~3 T109371720

Cancer Patterns In Hawa~! 1133 Primary site" Ot~ Exp SIR Oh~ Exp SIR ~4ale genital s~atem 393 591.2 Fr~tate gland 372 541.1 69~ Testis 13 33.0 34~ Peni~ 6 I0.2 59 S~ro~um and other male genital system 2 1.7 118 Urinary system 230 344.4 67~ 71 125.9 Bladder 158 240.9 66~ 48 72.8 66 Kidney and renaI pelvis ~0 93.9 20 48.0 42# Other urin-ry organs 12 9.4 128 3 5.1 59 Eye 3 9.3 ~3~: 2 9.0 22' Nervous sys~rn 31 67.3 s~ 25 50.5 50 Brain 26 63.3 41b 21 46.4 45 Other nervous system 5 4.3 116 4 4.2 95 Endocrine system 54 30.4 178s 104 6$.1 153~ Thyrvid gland 47 25.1 187* 99 64.8 153' Other endocrine glands 7 5.3 132 5 3.6 139 Lymphoma~ 90 160.5 ~ so 1193. 67# Lympho~rcoma and reticulum cell sar- 64 87.3 73• ~ 67.6 83 Hedgldn's disease 11] 46.6 12 32.2 37 Other lymphomas ~ 26.7 30~ 12 19.5 62 Multiple myeloma 13 41.0 32, 21 31.7 66 Leukemia 95 131.2 ~ 54 86.1 Lymphatic (lymphocytic) 20 $6.3 * 17 33.6 Acute 16 14.6 1180 I0 8.7 115 Chronic 3 36,7 ~ 4 21.0 19# Other 1 5.0 20 3 3.9 77 Other leukemia (e,g.. granulocyti¢ or 75 74.8 100 37 52.6 70~ monocytic) Acuta 29 39.2 74 15 28.2 Chronic 32 20.8 154• 13 13.5 96 Other 14 14.8 95 9 10.9 Other sad unknown primary 70 117.3 60) 72 101.4 ; Diigno~,~ were classified by the ICD-O. P<0.01. • P<0.0~. APPENDIX TABL~ 2.---Ob,qerred (Obs) and e~pe~d (E~p) n~nb~r~ and SIR fo~ concern by site centring am~# mates and females: C~in~e in Hgu~zii, 1967-76 ~ims~ site° Fem~e Obs Exp SIR 0bs Exp SIR v" Total malignant neoplasms 586 778.1 75~ 666 606.8 110" LiPLip~ oral cavity, and pharynx 381 39.19.4 .~ b 210 12.10.6 174"0 Tongue 1 7.2 14" 4 2.5 160 Salivary glands 9 2.7 5 1.9 263 Gum and mouth 1 9.0 11' I 4.1 24 N~ophary~x 24 1.6 1,SO0~ 11 0.6 1.833' Other and unspecified pharynx 2 9.2 22: 0 2.3 0 Digestive cry-an, and peritoneum 231 204.3 113 149 138.7 . 107 E~phagus 16 10.7 150 1 3.0 33 S~'nanh 27 30.9 87 28 14.3 196~ Small intestine 3 23. 136 1 1.5 67 Colo~ 77 75.1 103 63 65.4 96" Rectum, recto~ig-moid junction, and anal ~4 41.0 132 25 24.7 101 Liwr 21 6.7 313~ 6 2.9 207 Gallbladder 2 2.5 80 4 4.8 63 ~ Diagnmes were classified by the ICD-O. * P<0.01. • P<0.0,5. jxct. VOL f~. NO. 5. NOVEMBER I~C~'0 Tl09371721

1134 Primary site" Male Female 01~ Exp SIR Ohs Exp SIR Other billary l~L~ages 5 3.7 135 5 2.3 217 PanereM 25 27.8 90 15 16.3 92 Retroperitoneum, peritoneum, and un- I 3.6 28 I 3.5 29 specified digestive or~ns R~piratory system 124 188.1 66b 71 35,9 198b , 1 .I 421 o 2.1 oo ~ng 105 165.8 63 68 3P,4 21 * Other relpiratory organs includinl~ pleura, 11 32. 244* 3 1.4 214 med~Mtlnum, and intrathoracic site ]~one~ and joints 1 1.3 56 2 1.4 143 Connective, subcutaneous, and other soft 8 4.9 163 4 3.4 118 t~ue$ Skin. melanoma 2 10.2 £0" 1 9.8 l0b BreMt 0 1.7 0 164 170.1 91 Female Eenitai system 154 125.3 129" Cer~x, in situ 85 Cer~x 29 33.6 86 Corpu~ uteri 63 47.5 133" Uter~ N0S 1 5.3 19 0~ary 20 32.2 62" Vagina 2 1.3 164 Vulva and clltofia 4 3.9 i03 Other female ~enital system 0 1.4 O Male genital system 68 136.7 50* Proetate gland 64 127.3 50* Te~tie 2 6.8 29 Peni~ 1 22 45 Scrotum and other male genital system 1 0.4 250 Urinary system 35 75.4 46* 16 23.7 68 Bladder 23 53.6 43* II 13.8 80 Kidney and renal pelvis II 19.7 50 5 8.9 56 Other urinary organs 1 2.1 48 0 1.0 0 Eye 2 1.9 I05 0 1.6 0 Nervou~ system I0 13.2 76 12 92. 130 Brain 8 12.4 85 6 8.5 71 Other nervous system 2 0.8 250 6 0.8 750* Endocrine system 14 5.9 237* 28 12.0 233* Thyroid gland 12 4~ 250* 28 11.4 246* Other endocrine glands 2 I.I 182 0 0.7 0 L.vmphom~ • I0 32.3 31' 18 222 81 L.vmphoearcoma and reticulum cell sat-- 6 18.1 63~ 14 12.5 112 ¢omM Hndgkia's disease 3 8.9 34' 2 6.0 ~3 Other lymphomM 1 5.4 19 2 3.6 56 Multiple myeloma 4 9.0 44 3 6.1 49 Leukemia I£ 282 43* 16 16.2 , 99 Lymphatic (lymphocytic) 4 12.£ 33 ~ 3 6.5 46 Acute 3 3.0 100 2 1.6 125 Chronic I 8.1 L2b 1 4.1 24 Other 0 1.1 0 0 0.8 ff Other leukemia (e.g., ~ranuiocytie or 8 15,9 50~ 13 9.3 133 monocle) A¢ute 2 82 24" 7 5.3 132 Chrenic 4 4.4 91 5 2.5 200 Other 2 3.3 61 l 2.0 50 Other and unknown primary 27 25.5 106 17 19.2 89 i : Diagnoses were clarified by the ICD-O. P<0.01. " P<0.05. JNcI. VOL. &5. NO. 5. NOVEMBER !_~ T109371722

I

1138 female." Filipino~ in tta~.~ai~ l~eT-Tt~ P~m ~'7 site" Male Female 01~ Exp SIR O~ Exp SIR Leukemiffi 52 67.0 78 26 19.4 134 Lymphatic (lymphocytic) 1S 29.7 51b 8 8.1 99 Acute 14 8.0 17~ 6 4.1 146 Chroni= 1 19.1 5# 1 3.2 31 Othor 0 2.6 0 1 0.8 125 Other leukemia (e.g., granulocytlc or 37 37.2 99 18 11.4 158 monoey~ie) A~te 19 19.6 97 11 6.9 159 Chronic 14 10.0 140 7 2.5 280' Other 4 7.6 53 0 2.0 0 Other and unknown primary 47 59.6 79 14 16.6 84 • Diagnoses were classified by the ICD-O, # P<0.01. ¢ P<0.0& ApP~.N~IX T,~aLF, 4.--Obsert~d (Obs) and ez~ctcd (Ez'p) numbcn and 8IR fo~ ¢a~c~ by ~it¢ occurring arumg mal¢¢ and females: Har~aiia~ in Ha~:aii. 1967-76 Primary site" Mate Female Obs Exp SIR Ob~ Exp SIR Total malignant neoplasms 1,122 881.8 I~p, oral cavity, and pharynx 69 46~ Lip 0 11.1 Tongue 13 8,4 Salivary ghn& 7 3.7 Gum and mouth 17 IOA Nasophawnx 19 2.4 Other and unspecified pharynx 13 10~ Digestive organl and peritoneum 346 208.1 Esophagu~ 41 11.0 Stomach 113 30.3 Small intestine 4 2.6 Colon 48 P,~tum. recto~igmoid junction, and anal 35 41.9 canal Liver 40 7.5 G~libladder 9 2.4 Other biliary pM~ages 3 3.8 Pancreas 47 28.0 R~troperitoneum, peritoneum, ~nd uno 6 4.8 ~peeifi~d di~ztive orgen~ R~pimtory ~y~l~m 304 204.0 Larynx 22 21.8 Lun~ . 274 177.7 Other respiratory organs including pleura, 8 4.6 mediMtinum, and in~rathora¢ic site Bone~ and/ointe 10 5.1 Conne~tive~ subcutaneous, and other soft 22 8,8 ti~ue~ Skin. melanoma 9 17.6 Breaat 1 1.9 Female genital system Cervix. in situ Cervix Ut~ra~ NOS Vulva and clitoris Other femak genit~l 127# 1.379 938.4 147' 149.~ 44 18.5 238' o* 0 0.7 0 155 6 3.5 171 189 20 3.5 ~71b 163 10 &9 169 792# 4 1.0 40~~ 127 4 3.5 114 166# 244 170,0 144# 373# 10 3.8 373~ 38 16.7 527~ 154 7 2.0 350# 64# ~ 80.1 70# 84 23 31.0 74 533~ 16 4.4 38@ 375# 7 5.2 79 10 2.7 37O 168~' 25 19.1 131 125 2 4.8 42 149~ 115 83.1 217b IOl 5 3.3 152 I~4# 103 47.3 218~ 174 7 2.4 292~ 196 6 4.2 143 2,50) 12 7.8 154 51~ 5 19.9 25~ 53 307 288.9 I14~ 400 206,4 194~ 125 79 66.1 120 138 70.3 196# 0 8,1 0 52 ,52.0 10o 1 1.8 56 3 5.3 57 2 2.4 83 • Diagnoses were classified by the ICD-O. s P<0.01. " JXCl. vO- ~. NO. ~. NOVF..M~ER 1~3 TI09371724

Pffm~ry sita~ Male Fent~le Ob~ Exp SIR Ohs Exp SIR 4 Male genital system ~ I~0.7 75b Prostate gland 77 10S.8 71~ T~tis 19 18.8 101 Pen~ I 2.4 ~ S~m ~nd o~er ~ ~ni~ s~m I 0.6 167 Urinz~ ~s~m 47 81.S ~ $6 32.6 110 BI~dde~ ~ ~.3 ~ 21 17.0 124 ~dney ~ renal ~l~ ~ 24.? 81 11 14.3 ~ O~er urinsW ~ 1 22 45 4 1.2 ~ Eye 2 SA 59 S 3.8 79 Ne~us ~s~m 24 ~ S8 19 22.5 ~ B~n 19 ~.0 76 17 20.1 8S Other ne~o~ sya~m 5 2.4 ~S 2 ~5 ~ End~ne sysmm ~ 11.7 214~ ~2 ~.9 249# ~id g~nd 18 8.9 ~2~ 71 ~.0 263~ ~er end~ne glands 7 2.8 ~0" I ~.0 ~ ~mpho~ 47 ~A $8 43 38.9 111 ~mph~reoma and re~@ulum cell ~r- 25 ~.7 101 27 18.6 145 com~ H~gkln's di~ 13 19.9 65 I1 14.6 7~ ~her l~phom~ 9 8.8 102 5 5.9 ~ Mul~ple ~eloma 19 9.1 2~ ll 7,8 . 151 ~ukemia ~7 ~.I 80 ~ 31.6 ~ ~mpha~c (l~ph~e) 14 21.7 65 8 l&8 ~ Ch~nic 1 7.9 13t 3 4.6 ~ O~her kukem~ (e.g., ~nul~ic or ~ 24~ 95 19 17.8 107 monroe) A~ 14 14.6 96 8 11~ 71 ~nic 9 5.8 155 10 3.6 ~8t Other 0 3.8 0~ 1 3.0 ~ Other ~d un~o~ primz~ 62 ~.4 ~6s ~ 24.2 145~ ~ Dlagncees were classified by the ICD-O. P<0.01. APPENDIX TABLE 5.---Age-cpe~'f/~ inffdence rates for cancer amonq Ja~ane~e in Ha~zii. 196~-7~" Ske of cancer Incidence rates/100,000 population for age groups:. 0-24 25-34 35-44 45-54 55-64 6,5-74 • >75 yr yr yr yr yr yr yr Males Total ratli~nant neopl~ms 14.1 35.0 89.3 251.1 590.3 1,306.3 2,877.5 Lip. oral cavity, and pharynx 1.3 6.5 5,3 16.2 19.4 47.9 75.1 Digestive orgar~ and peritoneum 0.0 8.1 41.7 132.4 298.1 629.0 1.123.5 Stomach 0.0 2.9 1S.0 33.9 88.6 237.4 503.7 Colon 0.0 3.6 11.3 45,5 80.1 141.9 234.1 Rectum. reeto~igraoid junction, 0.0 0.9 10.6 28.7 65.3 100.8 149.3 and anal canal Respiratory system 0.0 2.5 9.1 46.3 99.7 226.8 550.3 Lung 0.0 1.9 8.3 3&6 84.4 193.1 493.4 Male genital system 0.6 2.7 1,5 5.7 51.8 195.7 626.9 Prostate gland 0.0 0.0 0.0 5.0 47A 192.1 615.2 Urinar~ s~mm 0.0 4.1 10.6 17.5 53.1 82.3 228.1 Bladder 0.0 4.1 5.8 8.5 36.3 59.0 173.9 Only sims for which at les.st 100 ca~es were recorded during this l:eriod are included. jNa. rot_ ~s. No. 5. NOVEMBER T109371725

1138 ~nc~dence ra~es/lO0,(~O ~cp~]aticn/'~r a~e 0-24 25-34 35-44 45-~4 5~..~4 F~-74 >--75 yr yr' yr yr yr yr yr Females Total rrmlignant neoplums 13.5 99.9 254.3 379.2 596.5 D|ge~ti~e organs and peritoneurn 0.7 11.0 38.1 73.8 190.5 Stomach 0.2 4.7 14,9 18.8 59.3 Co,on 0.2 3.2 12.0 30.8 61.8 Rectum. rectosi~,moid junction. 0.0 0,7 6.0 15.2 33.0 and a,~i canal Respiratory system 0.0 0.0 4.8 12.5 26.4 Lung 0.0 0.0 42 11.6 23.3 Breast 0.5 17.8 93.8 144.1 162,7 Female genital system 3.2 41.7 85.9 103.1 147.3 Cervix, in situ 1.6 24.6 45.3 11.7 5.5 Cervix. invasive 0.0 5.4 16.1 15.9 22.8 Corpua uteri 0.2 7.0 14.4 55.6 90.6 Ovary 1.I 3.2 9.6 17.0 33.2 Endocrine system ?-5 12.6 8.9 9.0 15.8 848.0 1.152.8 415.1 596.0 142.1 215.0 115.9 159.2 55.7 77.5 60.6 127.5 59.0 118.0 112.1 46.3 108.1 106.9 13.4 1.9 29.8 34.5 44.2 27.5 15.5 21,8 25.9 25,6 • Only sites for which at least 100 cases were recorded during" this period are included. lnddlnce rates/100.000 population for age SToups: Site of cancer 0-24 25-34 3,5-44 45-54 " 5,5-64 65-74 _>75 yr yr yr yr yr yr yr Males : Total malignant neoplasms 16.0 ,53.7 135A 335.3 661.5 1.165.2 1.586.8 Digestive organs and peritoneum 2.3 7.8 42.3 148.2 233.5 510.8 602.2 Respiratory system 1.2 13.7 12.6 42,3 183.5 230,4 442.8 Lung 1.2 13.7 12.6 35.3 1$8,0 200.1 343.9 Females Total malignant neoplasms 17.5 100.4 315.5 517.6 830.7 1.049.1 1.071.4 Digestible organs and peritoneum 1.2 0.0 30.8 77,8 155.1 386.0 461.2 Breast 0.0 19.4 142.7 137.2 229.6 149.6 74.5 Female genital system 1.2 50,6 78.9 185.1 177.4 136.5 183.1 See footnote. Appendix table 6. APPENDIX TABLE 7.--Age-epedfic incidence rates for ca,cer amo, g Filipinos in Hawaii, 1967-76" Incidence rate~liX}.000 population for a~e groups: Site of cancer 0-24 25-34 3~44 45-54 55-64 65-74 _>75 Males Total malignant neoplasms 14.2 30.8 75.6 205.9 544.0 1,079.2 1,795.9 Digestive organs and peritoneum 0.7 5.0 29.5 57.2 204.9 440.2 649.5 Colon 0.0 0.0 6.3 9,0 47.6 130.4 219.9 Rectum. recmsigmoid junction. 0.0 1.4 5.1 14.5 74.2 98.5 75.4 and ar~l canal • Respiratory system 0.0 3.6 23.1 49.7 102.5 I32.9 234.9 Lung 0.0 3.6 18.5 46.6 84.8 125.5 195.7 Male genital system 0.0 0.0 0.0 5.4 37.2 167.1 394.6 Prmtato gland 0.0 0.0 0.0 5.4 35.5 161.9 394.6 Females Total malignant neoplasms15.8 119.3 221.2 303,0 573.9 982.8 915.3 Female genital sys~m 3.7 56.3 80.2 95.1 114.2 114.0 81.2 • See fo~tnct~e, Appendix table 5. TI09371726

Cancer Palterns rates f~r e~r~er ~'~.~,g H~z~i~.~.~ in How:,~il, 1139 Ir.ddence rat#..~10~),~O ~puladan f~r age groa~ Site d car, oct 0-24 25-34 35-44 45-54 55-~4 65-74 >--75 ~ yr yr yr yr ~al~ Te~I ~li~nt n~pl~ms I4~ ~.0 150.7 518.4 Di~sti~ o~ans ~d ~ri~ne~ 1.3 11.7 45.5 134.5 4~.1 ~7.6 ~.1 S~ma~ O~ 4.7 13.5 48.5 11~5 316.2 1~.1 ~pi~ ~m 0.0 2.5 ~.7 ~3.8 379.0 ~1.0 ~8.5 Lung 0.0 2.5 ~.8 197.0 ~7.7 ~.3 ~.0 remal, Total ra~lignant neoplasms 15.8 141,2 297.0 629.4 1,042.4 1.639.6 1.501.4 Digestiv~ org-an~ a~d peritoneum 0.5 9.9 23.7 100.0 212.9 497.7 468.4 Respiratory system 0.0 2.0 22,6 62.8 88.2 171.9 14L3 Lung 0.0 0.0 22.6 57.7 84.6 142.8 113.2 Breast 0.0 17.1 80.2 170.9 236.4 346.7 353.1 Female genital system 4.1 78.7 127.3 183.8 257.9 282.4 169.7 Cervix. in aitu 2.5 45.5 58.9 45.3 19.3 26.2 14.1 Corpus uteri 0.9 13.9 23.8 76.9 136.0 145.6 56.3 See footnote, Appendix table 5. APPENDIX TABLE 9.---A~e, sex. a~d ethnic digributfon of ~e ~u~ti~ of Hawaii. I97~ Male F~ale ~ ~up. ~ C~uca- Fill- Hawai- O~er Cauea- j~m~ Chin~ Fill- Ha~i- Other s~n Japan~ Chin~ pino Jan s~n pino ian 0-4 10.081 6~21 1.348 6198 10~94 2.419 9.148 6~o12 1,280 5.801 9.550 2.151 5-9 9.721 7.861 1.686 6.629 11.936 2.184 10.124 7.553 1.735 5.796 11.397 2.233 10-14 12.608 10.927 1.502 6.128 11.658 2.391 10.459 10,220 1.715 6.476 12`049 2.140 15-19 12,318 12,143 2.002 5.647 9,145 1,973 10,378 11.295 2.148 5,011 8,968 1.808 20-24 28.839 10.(~ 2.062 4,246 5.993 3.167 14,237 9.962 1.774 4.304 5.774 1.827 25-29 15.201 8.807 1.652 3.917 4.794 2.035 13.345 7.186 1.744 4.533 5.632 2.135 30-34 12,415 62.56 918 3355 4.393 1.412 11,086 6,218 1.252 3,725 4.588 1,419 35-89 10.028 6.192 1.119 3.108 3.386 1.064 8.069 6.693 1.168 3.533 3.703 1.009 40-44 7.544 7.339 1.310 2.723 2.977 853 7.053 9.758 1.618 2`963 3.358 968 45-49 6.712 9,318 1.303 2`889 2,382 816 5,687 11.761 1,419 2.460 3,422 760 50-54 6.470 9.911 1.525 2.685 2.371 882 5.593 10.251 1.420 2`096 ' 2.431 691 55-59 5.382 7.687 I`276 1.488 1.335 430 4.458 7.817 1.265 1.021 1.346 447 60-64 3.340 5.337 1.163 4.103 1.338 345 3.693 5.267 995 958 1.331 238 85--69 2.569 3~97 822 3.532 727 165 2.882 3.078 731 725 1,183 298 70-74 1.684 2.947 866 1.903 537 192 1.952 3.068 862 543 530 283 ~75 1.885 3.751 742 1.654 616 308 3.530 5.263 939 730 710 350 Total 148,047 118.158 21-o96 60,305 73.882 20,636 122.174 121.602 22`066 50.675 76.472 18.747 JNCI. VOL fiS. NO. 5. NOVEMBER 19dO T[09371727

- Cancer Mortality Among Chinese in the United States, Ha|tung King 2, 3, ~ and Frances B, Locke F ° ABSTRACTmA total of 1,824 cancer deaths among ~o Chinese In C=Iifcrn|l. Hawaii, ~'~¢! New York C[~. 1~-72, was e~mlned against ~.6~ T~wane~ dying from ~ncer for the co~espond- log y~ Emphasis wM pl~ on pa~ernl of displacement by ~t;~ ~#een 1~0 sod 1970. Mu~ of the tr=ns~t~on~! ~denc~ were similar to those m~d~ for Japane~ and European migrants to ~e Unlt~ Sta~s, such =s the ds~ of ~nce~ of ~e lung and colon in melee However. the upward dl~lacement of ~ncem of ~e female breast and corpus uteri flll~ to occu~ among the Chinese. In general, the ps~em of ~ltion for Idai and ENIi was less apparent, compar~ with • e mo~ll~ experiences of Issei ~d Nisei. Pe~sps longer ~iod= Im ne~ed to achieve full displacement of cancer risks In The significance of the examination of mortality experiences among a migrant population with refer- once to nativity cannot be overemphasized. As one of nature's unplanned experiments, the migration of popu- la6ons provides an unusual opportunity for researchers to link differing mortality levels among residents of the same areas with d/ssimilar histories of exposures and to speculate on endogenous environmental confounding effects in the development of specific diseases. Previously, we systematically reviewed for the first rime nadvity differences in cancer mortality among the Chinese in the United States during 1959-62 (I). At that time, 16,% of the Chinese males and 11,% of the Chinese females of native (U.S.) birth were over 45 )'ears old, an age after which cancer becomes a numeri- cally important cause of death. By 1970, 18 and I'L% of the Chinese males and females, respectively, born in the United States had attained this age. An updated stud), of mortality would perhaps shed some new light on nativity differences in cancer among the Chinese. This paw is a summary of some of our findings, but a more complete analysis of mortality and incidence dam, supplemented by detailed information on Taiwan Prov- ince, Hong Kong, and Singapore, will appear later. DATA AND METHODS The study group included Chinese in California, Hawaii, and New York City, who constituted about 70,% of the entire Chinese population in the United States in 1970. Their age distributions by sex and nativity in the 3 areas (table l) and those of all U.S. Chinese are similar. Cancer mortality dam for 1968-72 were provided by the respective health departments. The total number of 1,824 deaths (1,26£ males; 562 females) included in the study represents 75.9 and 79.0~ of all deaths among U.S. Chinese males and females, respectively, For comparison with cancer mortality experiences among As/an Chinese, comparable statistics on Chinese in Taiwan Province during the same years were exam- ined. Mortality data were obtained from (2), and population figures were obtained from (3). Of the Taiwan population, 85% originated in Fulden, whereas most U.S. Chinese were migrants from Canton and its environs. Presumably, some of the present differences in mortality may reflect dissimilar genetic and/or cul- tural traits that characterize the 2 Chinese populations. Classification of diseases was in accord with the ICD (Eighth Revision). For general comparability, adjustments were made to certain disease categories in the 1968 dam for which the ICD (Seventh Revision) codes were used. The SMR was used as a summary measure of mortality, adjusted to the 1970 age-specific death rates of U.S. white males and females. For the comparison of transitional trends, a new set of SMR was computed for the 1959-62 data, adjusted to the same standard. Reference to these ratios, not shown in tabular form, will be made in the text. A Poisson-distributed variable was calculated on the basis of tabular values of 99 and 95,% confidence limits proposed by Bailar and Ederer USED:. ICD=Internadonal Classification of Dik.a~es; SMR=standardired mortality rations). t Presented at the Workshop on Popuiation~ at Low Risk of Cancer conducted at Snowbird. Utah. Auffus~ 23-25. 1978. ~ Biometry Branch. Div/slon of Cancer Cause and Pr~ention, Na- tional Cancer Institute, Natlorml Institutes of Health. Public Health Sew/co. U.S. Department of Health and Human Services. Bethel:in. Md. 20205. ~ Departtuent of Community Medicine and Family Health. School of Medicine. Graduate School. and Kennedy Center for Population • Research, Georgetown University, Washingtun, D.C. 20007. . * ,4ddress reprint requests to Dr. King at the National Cancer Institute. ~We extend our appreciation to Mr. John E. Patterson (Director) and Mr. Arne B. Nelson. Division of Vital Statistics. National Center for Hmhh Statistics, Rockville. Md., for the initial assistance in seek- ing cooperation for this study frotu the Io<al health authorities and to the following person~ for providing mortarity data for our analDis: Mr. Merle Shield~ {Chief) and Mr. Roger Smith. Vital Statistics Section. California Depantuent of Health, Sacramento, Calif.; Dr. Thomas A. Butch (Chie0"and Dr. George H. Tokuyama (.Mdstant Chief). Research and Statistics Office, Hawaii Department. of Health, Honolulu. Hawaii: Dr. Melvin 5. Schwartz (Assistant ComtuL~sioner for Biostatistic~l and Ms. l~rieda Nelson (A~sistant Direftorh Bureau of Health Statistics and Analysis, the Chv of New York Department of Health: Dr. C. M. Wont {Direaor General}. National Health Administration of Taiwan: a~d the late Dr. Kung- Pei Chen. Natiorml Taiwan University College of Medicine. We are also g~ateful to 3h. Frieda Nel.~n for her assistance in providing estimates of the age distribution of Chinese in New York City. by. ~ex and nati~'ity. 1970. 1141 JNCJ. VOL 65. .NO. 5. .~OI.'.'.'.'~MBER 1£80 T109371728

1142 <5 11,~8 10,946 1,012 10,971 9,991 980 5-14 2~,~% 21,~39 6,99~ 26.983 20,572 6,411 ~ 2L~9 9,~8 ~,~1 ~,~5 8,4~ ~,~6 ~ 17.~0 8,~ 9,~2 14,675 6,~2 7,~3 ~ 1~o ~ 7,~7 10.140 3,491 6.~9 ~74 8,~ 2,~7 ~,186 6~2 1,~1 4.301 ~75 3,172 1,~2 2,110 ~6 772 1.7~ To~! 152.765 ~,~4 ~,~1 141,~1 W.~I ~70 By convention, the first generation of Japanese migrating m the United Smms is denoted as "lssei," and their descendants born in the United States are designated ~"Nisei." To facilimm discussion, we sug- gesmd in our earlier report (1) the use of the terms "Idai" for foreign-born Chinese and "Erdai" for Chinese born in the United Smms: These two terms are used for nativity comparisons in the present study. FINDINGS Comparisons of Mortality by Nativity and ~x The SMR for major cancer sites and the expected values derived from standard ram tables are presented in tables 2-5 for the study populations. The 1968-72 dam showed that the overall cancer mortality ratio for Chinese males wa.s similar to that for U.S. white men, whereas Chinese females had a lower ratio than did U,S. whim women. Compared with 1959-62 dam, those for 1968-72 revealed no substantial reduction in mor- talit7 except for mormliw among Erda/ males, whose ratio was about one-half that for the U.S. whim male population and two-thirds that for male Taiwanese. As was true I0 years earlier, among the sites associated with high risks of cancer for males were the nasopharynx, liver, and, to a lesser extent, the esophagus (the esophagus ratio had shown a noticeable decline). For females, an elevated ratio was shown for nasopharynx, stomach, and liver cancers. The risk of lung cancer remained high dcspim a substantial ratio reduction since 1960. The sites with low risks of developing cancers were urinary organs and skin for both sexes, prostate gland for males, and breast and genital organs for females. Differences in mortality by nativity generally followed the 1960 pattern. For Chinese males, a higher mortality among ldai than among Erdai was observed for all cancer~ combined. The elevated rado of 132 for Idai was also noted for its being the only sex:nativity mortality ratio that exceeded the level for U.S. whites. In contrast, the overall cancer risk among Chinese females was similar for the 2 nadviry groups. JNCI. VOL ~. NO. $. NOVEMBER I.C$0 As with the overall cancer mortality, Idai males also exhibited an excess risk for cancers of practically all specific sites, including lung and colon. However among females, the ratio disparity was not as promi- nent except for cancers of the nasopharynx, liver, and pancreas, which was apparently due to the small number of observations. Inspection of age-specific curves. further confirmed this finding, because higher rates prevailed among ldai at older ages for all the cancer sites with elevated ratios. Compared with the mortality experience of Issei and Nisei, the pattern o[ transition for Chinese was less apparent. Perhaps the time needed to achieve a full displacement of cancer risks is longer in the elderly. More specific comments follow on the transitional experience by selected cancer "site. Specific Types of Cancer end/or Site Nasopharynx.--The lower risk of nasopharyngeal cancer reported for Erdal than Idai during 1959-62 appeared to have been greatly amplified for 1~68-7~, esperially among males. This impression was further reinforced by age-spedfic rates between the two inter- vals, which showed a much steeper leveling off in those approaching 60 years of age in 1968-72. The findings were consistent with the results of the case- control study of Zippin et al. (5) of California Chinese, which suggests that the lower risk among Erdai seemed to account for the apparent decline in this type of cancer. Among the Chinese in Taiwan, the nasopharyngeal cancer mortality was much lower than that of U.S. Chinese, though it still far exceeded that of U.S. whites. In the absence of further investigations, interpre- tation is difficult because Chinese in the United State~ and Taiwan do not share proportionately the same origin as noted earlier. Esophagus.--Mortality from cancer of the esophagus showed a decrease among Chinese males between 1960 and 1970. Although the ratio remained excessive for Idai, which was in keeping with the experience of Japanese migrants to the United States, both Idai and Erdai males experienced a leveling oil in age-specific Tl09371729

Cancer MortaII~ Among U.S. Chinese TA~L~ 2.---O~erre,d (0~) ar.d ex~ecfecl tF_,xp) n~:m~ers =r,~ S.~I2~ fir car,~.ers ~j sf'.e fe77",.~leac ~-~awa~, C~lif~rni=, and .Vew ~rk City, PMma~si~* Female Obs Exp SMR 0bs Exp S,~IR Total malignant n.~oplasms 1.262 1,2~.4 104 Lip, oral ca~ity, and pharynx 82 37.0 222 Lip 0 1~- 0 0 0.0 0 Tongue 4 8.6 47 3 ~2 136 Salivary giands 1 2,5 40 1 12. 8,3 Gum and m~uth 2 7.8 26~ 0 2.4 0 Nazopharynx ~I 2.8 22~6s 20 0.8 2,~0~ Other and unspecified pharynx 11 14.1 78 5 2.8 179 Digestive organs and peritoneum 527 ~41.0 155# 169 185.4 91 Ezophagus 50 25.6 195~ 2 5.6 36 Stomach 89 59.8 149# 39 23.3 167~ Small intestine 3 2.4 125 2 1.4 143 Colon 118 109.2 108 48 78.2 61# Rectum, rectosigmoid junction, and anal 47 38.8 121 15 18.2 82 canal Liver 97 11.9 815~ 15 3.6 417# Gallbladder 16 8.4 190¢ 10 11.6 86 Pancreas 66 67.6 83 24 81.5 76 Retroperitoneum, peritoneum, and un- 61 17.4 293~ 14 11.8 119 specified digestive organs Respiratory system ~63 391.7 93 91 64.1 142~ Larynx 5 18.0 28# 0 1.6 0 Lunl~ 339 367.5 92 8S 60.4 141# Other respiratory organs including pleura, 19 6.2 306# 6 2.1 286= m~llastinum, and intrathor~cie site ]~nes and joints 0 7.~ 0 1 4.3 23 Connective, subcutaneous, and other soft 0 6.0 0 0 3.7 0 tissues Skin, melitnoma 2 14.0 14.3# 3 8.4 36 Breast Female genital system 68 106.6 64s Cervix 19 30.4 63~ Corpus uteri and uterus unspecified 17 22.4 76 Ovary 30 49.6 60~ Vagina 0 1.2 0 Vulvg and clitoris 1 2.0 50 Other female genital system 1 1.1 91 Male g~nital system 41 106.8 38~ Prostate gland 37 98.6 Testis 3 6.6 45 Penis ~ 1.4 71 Scrotum and other male genital system 0 0.2 0 Urinary system 43 70.0 61n 14 20.9 67 Bladder 25 41.Z 61 .Kidney, renal pelvis, 18 28.9 62" 8 11,1 72 and other urinary organs Eye 0 12. 0 1 0.8 125 Nervous system 29 35.4 82 15 20.4 74 Brain 23 27.0 85 6 15.6 38~ Other nervous system 6 8.4 7~ 9 4.8 188 Endocrine system 4 4.4 91 2 4.1 49 Thyreid gland 3 2.6 115 2 2.9 69 Other endocrine glands 1 1.8 56 0 1.2 0 Lymphom~ 38 54.4 66~ 18 29.8 60¢ Lymphosareoma and reticulum cell sat- 22 28,4 77 12 15.3 78 eom~ Hodgkin's disease 5 15.8 32~ 4 8.6 47 Other lymphomas 9 10.2 88 2 5.8 34 Multiple myeloma 5 14.7 34~ 7 8.5 82 Leukemias 39 57.6 68" 32 32.0 100 ,Other and unknown primary 91 66.6 137~ 34 39.0 87 = The ICD (Eighth Revision) codes were used except for California ~nd New York City in 1968 were used. * Risks were significantly different [rom 100 at the 1% level. " Risks were significantly different frem 100 at the 5% level. when the Seventh Revision codes j.x¢~ VOL ~. .XO. 5. NO~.'E.MBER IS~0 TI09371730

1144 m=[~ a~ f~" Hu~ii. Cal~o~. .z~ New ~r~ Cit~. Female Primary site" O~s Exp S~L~ Obs Exp eMR Total n~lignant neoplasms 282 4~.0 Lip. oral cavity, and pharynx 15 14.2 Lip 0 0.4 Tongue 0 3.3 Salivery glands 0 1,0 Gum and mouth 0 3.0 Nasophar~x 12 1.2 Other and unspecified pharynx 3 5.4 Digestive organs and peritoneum 108 128.2 Esophagus I1 9.7 Stomach 17 22.4 Small intestine 3 1.0 Colon 32 40.8 • Rectum, rectosigmoid junction, and anal 10 14.5 Liver 11 4.0 Gallbladder 0 3.5 Pancreas 15 25.5 Retroperitoneum, peritoneum, and un- 9 6.8 specified digestive organs l~e~p|ratory system 62 149.8 L~rynx 0 6,8 Lung 60 140.6 Other re~picatury organ~ including pleura. 2 2.4 mediastinum, and intrathoracic site ~ones and joints 0 3.6 Connective, subcutaneous, and other soft 0 2.6 Skin, melanoma 1 5.8 Breast Fomale g~nital system Cervix Corpm, uteri and uterus unspecified Vagina Vulva and clitoris Other femal~ ~enital system Male genital system 18 38.3 Prestats gland 15 34.7 Testis 2 3~ Penis 1 0.5 Scrotum and other male genital system 0 0.1 Urinary system 14 26.3 Bladder 6 14.9 Kidney, renal pelvis, and other 8 11.4 urinary organs Eye 0 0.5 Nervous system 12 16.2 Brain 9 12.0 Other nervotm system 3 4.2 Endocrine system 0 1.7 Thyroid gland 0 1.0 Other endocrine glands 0 0.7 Lymphomas 14 22.7 L.vmpho~areoma and reticulum cell mr- 7 11.8 Hedgkin's dist~se 3 6.8 Other lymphomas 4 4.1 Multip]~ my~Ivma 1 5.5 L~ukemia~ 16 25.7 C~her and unknown primary 21 24.5 49# 192 252.9 76) 106 8 3.4 235: 0 0 0 0 0 1 0.8 125 0 0 0.4 0 0 0 0.8 O~ 1,000~ 7 0.4 1,7~0 ~6 0 L0 0 84 54 63.2 ~5 113 0 2.0 0 76 12 7.9 152 300 0 0.5 0 78 23 26.6 86 69 5 6.2 81 275~ 3 1.4 214 0 5 3.8 132 59~ 3 10.8 28~ 132 3 4.1 73 41~ 34 23.6 144 0 0 0.6 0 43~ 31 22.3 139 83 3 0.8 375 0 I 2.0 50 0 0 1.6 0 17: 1 3.2 31 29 56.2 52b 22 39.1 56# 4 11.6 34~ 6 7.6 79. 12 18.4 65 0 0.4 0 0 0.6 0 0 O.4 0 47~ 63 200 0 63c 3 7.3 41 40~ 2 3.2 63 70 I 4.1 24 0 0 0.4 0 74 7 9.1 77 75 4 6.8 59 71 3 2.4 125 0 1 1.6 63 0 1 1.0 100 0 o 0.5 o 62 7 11.3 62 59 3 5.7 53 44 3 3.5 86 98 1 2.1 48 18 3 2.8 107 62 14 14.4 97 86 8 13.6 59 " See footno~ a. table 2. v Risks ~re signili~antly different from I00 at the c Risks were significantly different from I~0 s~ the 1% level. 5% level. JNCI. VOL. ~. .NO. 5. NOVEMBER T109371731

Cancm" Mortality Among U.S. Chinese 1145 F Male Female Ol~ P.xp SMR Obs Exp SMR Total malignan¢ n~plasrns 977 742.4 ~p, oral ~vhy, and p,harynx 67 22.8 Lip 0 0.8 Ton~ ~ 5.4 Sgli~ glands 1 1.6 Gum ~ N~p~x . ~2 1.7 ~r ~ u~e~ pha~x 8 8.6 Di~s~ o~ and ~H~nv~ 419 212.8 ~ ~ 15.9 S~m~h 72 37.4 S~ll in~sfine 0 1.6 Colou 86 ~m, ~i~oid jun~on, ~d a~l 37 24~ ~r ~ 7.9 Gailb~der ~6 4.9 P~¢~ 41 41,8 Re~neum, ~riWneum, ~d ~- 42 10.6 ~imm~ s~m ~8 241.8 Lung ~6 ~,9 ~er m~pim~w o~ incl~ing pleura, 17 3,8 m~num, and intm~omcic ~nes ~d join~ 0 ~nn~fi~, s~cumn~, ~d o~er ~us~ B~t Female ~niml ~x Co~ u~ tnd u~ms unsp~ V~ Vul~ ~ clim~s O~er fem~ ~ni~l ~m Male ~ni~ s~m ~ ~.2 ~mm~ gland ~ 63.8 T~is 1 3.4 ~n~ 0 0.8 $e~m a~ o~er ~k ~ni~ s~m 0 0.1 U~naw ~m ~ ~.8 Bl~d~r K~dn~. 132# 370 4~4.8 294# 21 6.0 0 0 0.0 0 74 2 1.4 143 62 1 0.8 125 42 0 1.6 0 3.059~ 13 0.5 2,600' 93 5 1.8 278 197~ 115 122.2 94 245~ 2 3.6 193~ 27 15.5 174~ 0 2 0.9 222 126' 25 51.6 48# 153c I0 12.0 83 1,089~' 12 2.3 327# 5 7.8 64 98 21 20.8 101 396~ II 7.8 141 123# 57 40.4 141c 45# 0 1.0 0 122 54 38.1 142' 447# 3 1.4 214# o 0 2.3 0 0 0 2.2 0 12~ 2 5.1 39 49 94.0 52# 46 67.4 68~ 15 19.0 79 11 14.7 75 18 31.2 58" 0 0.8 0 I 1.3 77 I 0.5 200 0 0 66~" 11 13.6 81 73 4 6.6 61 57 7 7.0 100 0.5 2O0 11.3 71 8.9 2..4 2.6 38 2.0 50 0.6 0 18.4 60 9.6 94 5.I 20 3.7 27 5.6 71 17.6 102 25.5 102 ttHnary organs Eye 0 0.7 0 I Nervous system 17 19.2 89 8 Brain 14 15.0 93 2 Other nervotm sy~_m 3 4.2 71 6 Endocrine system 4 2.6 154 1 Thyroid gland 3 1.6 188 1 Other endocrine glands 1 1.0 100 0 Lymphoma~ 22 31.7 69 11 Lymphomtrcoma and redculum cell sar. 15 16.7 90 9 com~J Hodgkin's dise~e 2 8.8 23 1 Other lymphomM Multip]e~ myeloma 4 9~] 43 4 Leukemlas 23 31.9 72 18 Other and unknown primary 70 41.4 169~ 26 • Se~ footnote a. ruble 2. ~ Risks were significantly different from 104} at the ~ Risks were significantly different: from I{X) at the level. level. .~NCI. VOL ~. NO. 5. NOI.~MBER 19~.9 T109371732

1148 Tai~ t~roz'irxe, .1971-75 Primary Male Female Ol~ Exp SMR 0bs Exp SMR Total malignant neoplasms 27,843 ~¢1,378.6 Lip, oral cavity, and pharynx 1,663 1.126.2 Lip 33 31.3 Tongue 148 260.5 S ~livm~" glands 50 53.9 Gum and rr~uth 272 232.0 N~opharynx 861 96.9 Other and unspecified pharynx 299 436.6 Digestive orpns and peritoneum 16.717 8,960.1 Eaophagus 1,643 751.9 Stomach 5,388 1,602.7 Small intestine ~ 73.9 Colon 1,152 Rectum, rectosigmoid junction, and anal 643 1,046.7 Liver 6,882 2/6.8 Gallbladder 111 2/0~ i~ancreas 364 1,911.7 Retroperitoneum. peritoneum, and uno 278 141.8 ;l~ified digestive organs Respiratory system 4,389 II,819.~ Larynx 393 533.9 Lung 3,445 11,088.7 Other respiratory organs including pleura, 751 196.6 medlastinum, and in~athoraeie site Bones and joints 425 299.8 Connective, subcutaneous, and other soft 239 227.1 tissues Skin, melanoma 66 514.0 Breast Fenmle genital system Cervix Corpus uteri and uterus unspecified Va~na Vulva and clitoris Other female genital system Male genital system 349 2,320.3 Prostate gland 212 1.957.7 Tends 53 316.9 Penis, ~crotum, and other male genital system 84 46.7 Urinary system 638 1,864.3 Bladder 459 9~6.2 Kidney, renal pelvis, and other urinary 179 898.6 orgzns Eye 51 41.3 Nervous system 354 1,467.0 Bnin 305 1,092.9 Other nervouz system 49 374.1 Endocrine system 62 148.6 Th~'oid gland 44 80.3 Other endoer/ne glands 13 68.3 Lymphomas 667 1.933.5 Lymphosarcoma and reticuhm cell ~ar- 3~5 2/2.3 coms~ • Hodgkin~a disease 86 634.3 Other lymphcmzs 216 326.8 Multiple my~loma 36 403.3 Laukemias 1~12 2,065.9 Other and unknown primary 775 2.187.5 79b 18,792 23,814.7 79b 148~' 617 321.9 192~' 105. 20 2.0 Io000' 57' 53 73~9 72* 73* 34 39.1 87 I17~ 110 78.2 141" 889~ 296 30.7 964# 68# 104 98.0 106 187` 7,307 5,797.0 126' 219b 361 189.3 191~ 336~ 2,499 758.5 329' 40' 1,085 2,557,4 42~84} 61# ~04 597.4 2,486~ 2.123 130.3 1,629} 41b 66 373.4 18# 19} 197 1,035.0 19~ 196# 267 ].].0.2 2 39' 2,045 2.].99.1 93 74' 115 52.9 217# 31' 1,588 2.371.3 77" 382~' 342 74.9 457~' 142' 332 138.2 176~' 105 146 147.7 99 13~ 51 303.3 17' 1,049 5,184.8 20' 3.~94 3.671.3 98 1,766 738.3 239# 15' 11# 4~ 422 696.8 61# 237 ~09.5 77# 29# 185 387.3 46# 1~3 42 31.8 132~ 24~ 301 850.9 35 28* 268 628.6 43# 13' 33 222.3 15~ 42~ 128 143.6 89 55: 113 96.8 26 15 46.8 34~ 337 1,089.4 31' 38# 186 544.6 34 14# 34 343.9 10# 66' 117 291.0 13 377.9 59~ 868 1,358.3 $5~ 1,540 L552.4 99 • The ICD (Eighth Re;4sion) codes were used. b Risks were significantly different from 100 at the 1% level. r Risks were siEnifi~mntly different frem 1~0 at the 5% level. JNCL I.'OL ~5. NO. 5. NOVEMBER l__r~3 Tl09371733

curves at about age 70 years. The h/gher male than female risk noted for 1959-52 reappeared during 1959-72. For Taiwanese males and females, the mortality re- mained stable at a high level. Stom~h.--Unlike the ris'ks of stomach cancer of other migrant groups to the United States, those of the Idai were more closely aligned with those of the host country population than with the risks of the home- land population, which represented a continuation of the 1959-6~ pattern. Erdai males showed a lower ratio than the U.S. whites, whereas the SM.R for Erdai females greatly exceeded that for U.S. whites but was considerably lower than the ratio for the Taiwanese, Colon and recturr~Consistent with the pauerns of tramition reported for migrant populations to the United States from Europe and Japan, the Chinese have continued to display since 1960 an elevated colon cancer Yisk with one exception: The risk for Idai females remained at a Iow level, similar to that for the Taiwanese. For rectal cancer, the magnitude of upward displacement observed among Idai males was similar to that of 1960. An apparent increase in SMK was shown for Taiwanese women in 1970. riv~ and gallbladder.mAs in 1960, mortality from liver cancer continued to be extremely elevated, par- ticularly among Idai males. The much higher rado shown for Taiwan was in keeping with the conspicuous elevation of this cancer observed among Chinese in other areas. For cancer of the gallbladder, however, the high incidence reported earlier (6) and in this mono- graph [Appendix table 9 in (7)] for Chinese females in Hawaii was not supported by mortality data on Taiwan. The ratios computed for U.S. Chinese, based on small numbers, are subject to sampling variations. tung,~A decline in lung cancer mortality since 1960 was noted in all sex-nativity groups except Idai males, whose stable SMR surpassed the mortality rado for U.S. white males by 22%. However, Chinese females cominued to display a higher ratio than Chinese males due to the lower mortality among females than among males in the standard population. Lung cancer mortality was increased in Taiwanese males. Although the reverse was true for females, the female rado exceeded that for males by 1.5, a reflection of the standardized rates mentioned above. Bre~t (female).~The much depressed mortality from cancer of the female breast observed among both nativity groups demonstrated once again the apparent persistence of a low risk of developing this cancer among Orientals. As further reinforcement of this observation, age-specific rates appeared to level off among the 60-year-old Erdai. Genital organ~ (female).~As in 1960, mortality rados for cancers of each genital site (cervix, oth'er uterus, and ovary) were lower among Chinese than among U.S. whites, particularly in Taiwan. One noticeable observation was that since 1970 mortality has decreased for each nadvity group, except that the risk of develop- ing cervical cancer became slightly elevated among the Idai. Ctne.~r MortalIly Among U.S. Chinese 1147 Data limitations for Taiwan precluded computation of separate mdos for cancers of the corpus uteri and other subsites. However, the cervical cancer rado for Taiwan has sharply increased since 1959-62, perhaps due to a decrease in the number of physidans trained in the Japanese tradition and hence a change in diagnostic practices. The ratios for ovarian cancer suggested that mortality was on the rise among Chinese migrants and their descendants, a transitional experience shared by ml- grants from Japan and other countries. The small number of deaths did not permit dear-cut nativity contrasts for this cancer site or for the cervix or uterus. However, the collective experience of Idai and Erdai females seemed to indicate that risks of developing cancer at these three sites failed to rise to the levels of U.S. whites. Prostate gland.~As with the experience of Japanese migrants, Erdai mortality from cancer of the prostate gland remained far below the SMR of U.S. whites. Moreover, the age-specific curve leveled off before this nativity group was 70 years old. The extremely low mortality from this type of cancer in Taiwan is shared by the Chinese at home and abroad (King H: Unpub- lished observations). Leukemia.~Mortality from leukemia among Chinese males remained lower than that in the standard popu- lation, but the ratios for females declined slightly to be the same as those for U.S. whites. Among the Chinese in Taiwan, a much lower risk still prevailed. Lymphoraa,~Since 1960 some reduction seems to have occurred in the mortality rado for lymphoma among U.S. Chinese. The substantially depressed mor- tality noted for Hodgkin's disease was consistent with the low risk usually reported for the Chinese in other areas, including Taiwan. DISCUSSION The distinctive profile of cancer risks among Chinese generally resembled that seen in 1960. The rise of colon cancer among Chinese to the level of U.S. whites was a likely environmental effect and was consistent with the experience observed among mi- grants from Japan and from those European countries where a low risk of cancer at this site has prevailed. With regard to the high risk of cancer of the naso- pharynx, the usual hypothesis, i.e., endogenous prediso position, seemed to have been confounded by our obo servations of a continuing decline since 1960 in this cancer among Erdai males and by earlier reports based on limited data (8). More investigations of U.S. Chinese are needed to 1) focus on place of origin in the search for genetic markers that might discriminate between regional dialect groups and 2) systematically study their living habits, with emphasis on patterns of accultura- tion. The increase of lung cancer among Idai males is presumably due to environmental effects and par- ticulariy to smoking. The findings of Lin et ai. (9). JNCI, VOL. 6~. NO. 5. NOVEMBER Tl09371734

1148 King and Locke indicating a steady rise of this c~ncer among male Taiwanese during 1954-74, further support such a hypothesis. For the perplexing high risk of lung cancer among Chinese females at home and abroad, however, no immediate interpretation seems to be available. Earlier, incense smoke was regarded as a possible etiologic agent (10). More recently, speculation has broadened to include, among other exposures, smoking, cooking ~apors, and kerosene stoves; defidency in vim- rain A and ascorbic add; low consumption of dark-green leafy vegetables; and such confounding factors as soc/al class and dialect groups (11-14). One recent hypothesis linked smoking to ~he development of adenocardnoma (including scar cancer), but the findings remained inconsistent U1, 15-18). In Singapore, MacLennan et al. (11) reported that the high incidence of lung c~ncer observed in Cantonese males who exhibited elevated rotes of adenocardnoma seemed unrelated to smoking. Nevertheless, the fact remains Chat Chinese females with lung cancer represented a high proportion having this histologic type, i.e., 33-~4% for Singapore and Hong Kong (1#, 16), 59% for Taiwan (19), and ~0% for the U.S. Chinese female population (National Cancer Institute, Third National Cancer Survey: Unpublished dam). Inddentally, a higher frequency of adenocarci- noma was also observed among Jewish females with lung cancer: their eleva~'d risk also has been known to exceed that in males (20). More extensive studies are needed. Studies ~f migrant populations generally indicate a dose correlation in the risks of developing cancers of the breast, corpus, and ovary. The absence in the Chinese of a rise in risk of developing breast cancer to the level of the host population, though representing a deviation from the experience of several European migrant groups to the United States, was shared by another Oriental population, the Japanese. However, as did their European counterparts, both Orlenml migrant groups exhibited an elevation in ovarian cancer. The upward displacement of cancer of the corpus uteri shown for .Japanese and European migrant groups failed to occur among the Chinese; only com- prehensive acculturation studies and/or prolonged ob- servations would provide an adequate explanation. In view of the presence of. major Chinese settlements in various parts of the world, the Chinese population serves as a natural experimental group for epidemiologic observations of the interplay of host characteristics and environmental factors in incurring health risks. One major, largely untapped, source of epidemiolog/c infor- madon is the vast population of the People's Republic of China with many unique opl:~rtunides for in- vestigation (21). REFERENCES (l) K~,~G H. [-L.,~'~.S~L ~,V. C~nce/" mon=lity among fo~i~- 25:62~5. (2) ~ o[ ~1~, T~i~n ~o~n&l ~en~ T~n. (~) Minb~ of ~e line,or, T~w~n. T~wan. Table ~. In: T~n (4) 8m~ JC Ill, ~z~ ~. S~iE~ce ~cm~ for t~e ~do ies ~ he~i~y ~d en~ronmen~ in ~neer o[ ~e ~ (~) W~ous~ ], Mum ~ &~ P. POWZLL ~. ~neer in~- ~e in li~ ~n~n~m ~RC S~ Publ 1976;3:5~9. (7) Ko~ ~. ~ ~t~s of four ethnic ~oups in ~ii. (8) BV~ P. N~pha~x ~ncer in Chine~ of ~lifom~. Br ~li~ from ~li~nt n~s in T~n. J Fo~o~n (1o) ~O~L ~. G~s~ S. ~r~n~ens [n Chine~ incen~ smoke. Natu~ (11) MAc~s~s R, ~A JD, DA~ NE, ~w CH, ~ ~, ~x~ ~ ~ ~cmn fo~ lung ~er in Sin~ ~ine~, a ~pula~on wi~ high f~le inc~e~e ~s. Int (12) ~us~ Js. ~reue smoking, ~he kem~ne stove and ~r in H~g Kong. Br ~ Dis ~ (~) BJ~ ~ Die~ vi~min A ~d h~n lung ~cer. lnt J s~ic hi~l~i ~y~ of lung ~ncer in Sin~ ~in~ d~l~ ~ups, ~d ~r ae~olo~! si~i~i~e. Int J ~n- (1~) W~ E~ ~c~ S. ~. Lung ~n~r. ~n~: UICC, 1976 ~hni~i Ke~n ~ No. ~). {16) ~ ~ ~c~s~ R. Lung ~nc~ in Hong Kong: Momlit~ (17) ~s~o~ A. Rela~onship ~tw~ ~ue smo~ng a~ his- mlo8ic t~ of lung caner in womm. Tho~x D~ a~ ~me ~io~hips am~ ~uiar ~i~ smoken and lifel~s n~moken. J Epid~iol ~mm~i~y H~lth 1978; ~:~0]-31S. (1~) L~H ~, K~o SH, ~ CC, Y.~s~ SP, ~t~ KP. Prima~ ~n- ~ ~ ~e I~ in Talon. I. Epid~iol~c inv~ti~tions: (in ~ine~). (20) Hozw~ H. L~g ~n~r ~ong ~e ]~s. Am ] Public H~lth ~ubl~ of ~. J ~ad ~nc~ lnst j.~cl. VOL. ~. NO. ~. ."~OVE31BER TI09371735

Cancer Mortality Risk Among Japanese Frances B. Locke 2 and Haltung King in the United States, i ASSTRACT~Earller fIndlngs on cancer morta~ity experiences of Nisei and Issal were upda~sd to ~ound 1970. ~mp~ w~ U.S. w~Re~ J~lne~ in Jlpln hid I high standardized mo~lt~ ~tlo (SMR) for canclm of the ~oph~gus, ~oml~. p~ge¢ ~d coax, where~ ~W had 1~ r~t]os for ~cem of molt othlr ~lt~ A~ng U.S. J~p~, ~e dlm~lon magnitude of me SMR ~snsl~on vsd~ by site. but generally the SMR m~erst~ tows~ tb~t of U,S. whR~. Sp~iflcslly. mop ~11~ tram =;omach ~n~r ~s elevat~ though the rl, o has ~n r~u~. In aoaiUon, mon~l~ from ~ncem of ~e liver ~d blllW p~.g~ mmaln~ high. In cont~t to I sh~ d~lne in the SMR for Isophlge~ ~r, the SMR for ~ncer of ~e colon ~¢ ~mphomil m~ ~oser to the ~vel for whi~. T~ 5MR for ~n~e~ of the o~w a~ prelate gl~d rep~ent~ a rise that of Japan. A hlgh~ II~i ~ NilIi SMR wM ob~ for ~at cancer slte¢ ~ardlm of ~e risk Isvel In J~ T~ revere was tree for ~em of the liver a~ nmphawnx ~ong ~mpholarcoma and leukemll among ~th sex~NCI 1149-1156. 19~. Cross-national comparisons of the cancer risks among migrants and their offspring with those reported for the homeland and host populations have contributed greatly to the elucidation of changes in magnitudes and directions of site-specific cancer rates among rati- ons ethnic groups (1-5). Japanese migrants to the United States provide a rich resource for such analyses, particularly because of the presence of a sufficiently large number of foreign-born (Issei) and native-born (Nisei) persons needed for observation, in addition to the availability of comparable data in the homeland. Following up on Smith's (1) initial analysis of cancer risks among the Japanese in the United States for 1949-52, Haenszel and gurihara (4) took advantage of the noticeable increase in the number of Nisei at older ages reported in the 1960 census to examine nativity differences in mortality of this minority during 1959-62. By 1970, the proportion of Nisei more than 55 years old had risen to 10~, which suggested the advisa- biliw of the reassessment of cancer risks for Japanese for 1968--72 (table 1). As a supplement to the detailed SMR computations (tables 2-5), this presentation is mainly descriptive and profile oriented. An in-depth analysis in which all associated causes of death and incidence trends in both homeland and host countries are considered will ap- pear in another paper. DATA AND METHODS Information on Japanese in the United States by nativity is not available in published national mor- tality statistics. Mortalit.v data for 1968-72 used in this study were obtained from local health authorities in Hawaii, California, and New York City. The number of Japanese in the three areas combined, as reported in the 1970 U.S. census, comprised 75,% of the total 1970 U.S. Japanese population. 80% of Nisei, and 60% of Issei. The age distributions of the study group and total U.S. Japanese population were similar, except that the study group had a smaller proportion of Nisei 75 years old and over and a larger proportion of lssei in the older age groups. The total number of deaths available for analysis for the 5-year study was I,,H6 among males and 995 among females, which represented 86~ of all cancer deaths among U,S. Japanese. For comparison with the homeland, dam on cancer mortality in Japan were obtained for 1969-7~ [(6); Segi M: Personal communi- cation, 1978]. The ICD (Eighth Revision) was used for dassiflcation of cancer deaths, ~cept for 1968, when deaths in California and New York City were initially coded by the ICD (Seventh Revision). Tf e SMR. were based on the 1970 age-specific death rates of U.S, whites for each sex. The 1 and 5~, confidence levels noted in the tables were determined on the basis of the significance ~actors prepared by Bailar and Ederer (7). Mortality ratios for several popu- lation groups, although direcd.v comparable to those for the standard population, are less precisely com- parable to one another when the study groups have Aa~,agWAClONS ust~. ICDa=Internationai Classification SMg'astandardized mortaliw ratio(s). ~ Presented at the Workshop on Populations at Low Risk o| Can. car conducted at Snowbird, Utah, August ?J-25, 197B. z alometry Branch, Division o~ Cancer Cause and Present/on. National Can~er Institute, Nadonai Institutes of Health. Public Heahh S~rvice. O.$. ]Department of Health and Human l~thesda. Md, ~0205. ~ l~parmsem o[ Community Medicine and Family Health, School of Mdidne, C, mdua(~ School. anti i~nndy Canter fo¢ Population Rematch, C~o~,etown University. Washington, D.C. 20007, • We exp~ss our ~rat~tude to the iollowing individuala |or their ~ssistance in either contacdn~ the loGal health authorities or in pro- ,dding momllty and population data for this analysis: Mr. John I:. Patterson (Director) and Mr. Ame B. Nelson, DDision oI Vital Sta. tildes. National Czntcr Io¢ Health Statistics, Rock*~lle. Md.: Mr. Merle Shldds (C~ie0 and Mr. Roger Smith. Vital Statistics S~:don, Calit'omia l~partment oi Health, Sacramento. C~li[.; Dr. "L"homas A. Tlurch (Chi~) and Dr. C,¢org¢ H. "Foku!rama (A~sistan¢ Chief). search and Statistics Oflice. I-lav.'aii Department o~ Health. Hono- lulu, Hawaii; Dr..Mels~n S. Schwartz..Msistam Ccnnmissioner for Biostatistics and Ms. Frit'da Nelson..Msis:ant Ditx-ctor, Bureau Health Statistics and Analysis. the C/W of New York Deparmaent ef Health: and Dr. Mitsuo Segi..Mchi Cancer Center. Nago~.a. Japan. 1149 JNCI, VOL E~, .NO. S. NOVEMBER 1,~0 T109371736

1180 Locke and King ~e F~e ~ ~rn To~I ~ ~ <5 15,140 14.368 772 14.643 13,940 708 5--14 39.710 38,454 1,256 38.764 97,591 1,173 15-24 35.907 33.796 2,111 37,1~6 34,968 2,198 25-34 27,51£ 22,898 4,614 31,511 38,037 8,474 3~-44 30.824 27,124 3,700 43.908 29,067 14,841 45-54 32,176 30.356 1.820 33,198 29.756 3,442 ~ 14.885 12.915 1.970 13,985 11.637 2.348 85-74 9.108 3,r~oO 5,518 11,337 3,596 7,741 ~5 6,002 924 5,078 7,358 1,317 6,041 Total 211.264 164,425 26.839 231.875 164.909 46.966 disparate age distributions, as for the age distributions among lssei and Nisei. Age-rpecific rates for each' site were examined for confirmation or amplification of the sununarized rados. FINDINGS Mortality From Cancera of All Sites As of 1970, U.S. Japanese males and females have continued to have a lower overall cancer mortality than do U.S. whites with SMR of 84 and 73, respectively. Among lssei and Nisei, the mortality rados for males since 1960 have sharply declined, whereas the SMR have remained stable for feraales. In Japan, males showed an intermediate redo of 93 between the Issei and Nisei levels of 105 and 66, respectively. However, the ratio of 87 for females, the same as that of the Issei, was above the Nisei figure of Generally, Japanese males and females in both the United States and Japan exhibited a higher mortality than did the whites from cancer of the digestive system, notably cancer of the stomach and biliary passages. In contrast, a lower mortality than was expected was shown for cancers of the respiratory and nervous systems and genitourinary organs and for lymphoma and leukemia. lssei, when compared with Nisei, continued to show, as in 1960, a higher mortality ratio for most cancer sites; the reverse was true for cancers of the naso- pharynx and liver among males, cancers of the breast and several genital organs among females, and leu- kemia among both sexes. Mortality From Cancem of Spaciflc Sites Canc~ o/the buccal cavity.mMortality from cancer of the buccal cavity has remained low since 1960 for U.S. Japanese, with higher ratios for Issei than for Nisei. In Japan, the rado for males was lower than that for Issei and Nisei, in contrast to the intermediate rado shown for females, As with the homeland population, U.S. Japanese showed a low mortality ratio [or cancer of all specific sites within the buccal cavity except for the SMR for jxca: VOL SS. NO. S. NOVEMBER cancer of the nasopharynx, which exceeded that for U.S. whites, particularly for raales (males, 256; females, 153). Although the observed numbers were small, the male ratio was statistically significant. Cancer o/ the digestive ~aem.--Ovemll mortality among U.$. Japanese, especially Nisei, was much below that among Japanese in Japan, but it greatly surpassed the U.S. white level. However, significant variations were found among the subcate~ories within the digestive system, as indicated below: a) Cancer of the esophagus: The risk among U.$. Japanese males was much lower than that for Japanese in Japan, particularly Nisei males, whose ratio was at par with whites (Nisei, 113; Issei, 197; Japan, 245). The ratios for U.$. Japanese females were below those for U.S. whites and for females in Japan. b) Cancer of the stomach: In contrast to the risk 'of developing stomach cancer in the United States, which is among the lowest in the world, Japan has the highest risk, with an elevated ratio of over 800 among both sexes (8, 9). The ratios for U.5. Japanese, intermediate between those for Japanese in Japan and U.S. whites, ranged from 350 to 400 and have remained fairly stable since 1960. The ratios for U.& Japanese sex-nativity groups were similar, except for a lower ratio of 287 for Nisei males. c) Cancer of the colon: In contrast to the high vulnerability to developing cancer of the stomach, the risk of developing colon cancer among Japanese in Japan ranks among the lowest in the world. However, a moderate rise in mortality has been observed since 19~0, which is in keeping with the trends in most countries that have low rates for this type of cancer (10). The mortality ratios for the U.$. Japanese, al- though remaining below 100 (male, 75; female, 79), represented an upward displacement from the home- land level of about 30. As for stomach cancer, the ratios for colon cancer for U~. Japanese sex-nativity groups were similar. The apparent convergence of male and female mortality risks, even though a slight elevation in ratio was observed among males, was a change from the 19~0 profile. d) Cancer of the rectum: Mortality risks among the U.S. Japanese -,'ere about average (male, 113; female, T[09371737

1151 O~ Exp SME 0ha Exp S~R Total n',aligr, ant Lip, ~1 ~W. ~d ~p Ton~e Gum ~d ~p~ ~her ~d uns~ifi~ p~x Diz~fi~ o~ns ~d S~ma~ S~II in~s~ne ~ion ~mm. ~t~o~d ~unc~on. ~d ~r Gallbkdder Re~neum, ~neum. a~ un- ~x ~num. ~d in~omcic si~ ~n~ s~ joln~ ~n~, ~bcumn~, ~d o~er ~ft S~. ~t Fe~le ~iml ~x V~ V~ s~ eliWris O~er ~msle Male ~niml ~mm g~nd P~ni~ S~mm Bl~d~r Eye Ne~ ~m Endmdne LymphomM ~ph~o~ ~gkin's di~ ~er l~phomM Mul~ple m~lo~ ~emiM 1,445 1,727.6 84~' 3'7 53.0 70< 0 2,0 0 6 12.1 50 2 4.0 50 8 11.2 71 10 3.9 256< II 19.8 56 737 498.8 143~' 54 36.1 150~ 314 89.4 351s 7 3.6 194 121 162.1 75b 65 57.4 113 ~8 12.8 2Sl~ 3~ 14.,5 241# 84 ~.0 88 21 26.8 78 319 5£9.2 14 24.8 297 495.6 8 8.9 1 10.8 0 8.6 5 20.6 ~6 1.357.3 73b 8 18.4 43s 0 0.2 0 3 4.3 70 1 2.4 42 2 4.6 43 2 1.5 133 0 5.3 0 613 388.0 132~' 6 11.4 205 60.2 40~8" 2 2.8 71 129 164.1 79b 33 37.8 87 14 7.2 194~ ~0 24.2 207# 58 65.2 89 16 25.1 64 60~ 80 123.8 65~' 0 3.0 ~00~ 7, 116.6 90 I 4.2 24 9 0 7.8 0 0 0 70 0 24# 3 16.5 18' 91 £96.4 31~ I05 209.1 50~ 33 60.4 &5~' 13 44.4 29~ 61 95.6 53' 4 2.4 167 2 4.2 48 2 2.1 , 9~ 83 174.4 48~ 77 162.8 47' 4 9.0 44 2 2.2 91 0 0.4 4~~ 25 43.6 57 57 102.9 36 62.2 ~ 14 21.8 64 21 40.7 62~ 11 21.8 0 1.8 0 0 1.5 0 23 48.8 47~' 21 86.0 17 3?,4 45~ 13 27.8 47~ 6 11.4 53 8 8.2 98 3 6.2 48 5 8.1 62 3 3.7 81 4 6.0 67 0 2.5 0 1 2.1 48 57 76.6 74~ 43 57.1 75 37 39.9 93 £8 29.8 94 14 22.1 63 8 15.8 51 6 14.6 41~ 7 I1.5 61 5 20.4 25' 9 16.6 54 65 83.4 78" 39 60.5 64' 54 92.1 59~ 54 76.9 70~ • The ICD (Eichth Revision) code~ were u~ed except for California and were u.~ed. See Apl~ndix, p. 1191. ~ Risks were significantly different from I00 at the 1% level. ~ Risks ~r~ significantly different fr~m 1~) at the 5% level. New York City in 19~8 when the Seventh Revision codes JNCL VOL. ~. NO. 5. NO%T--~BER T109371738

1152 Locke and King Female Obs Exp S~4R Obs Exp S~4R Total malignant neoplasms 618 937.0 ~6b 453 744.0 61~ Lip, eraI ~ty, ~d pha~x ~ 31.8 ~: 3 10.4 Up 0 0.8 0 0 0.1 0 Ton~ 4 7.3 ~5 2 ~4 83 S~l~ ~lands 1 1.9 ~ 0 il 0 Gum ~ mou~ 3 6.S 46 0 ~5 0 Hmp~x 8 2.8 ~ 1 L0 100 ~er ~d uns~ifi~ pha~x - 4 12.5 ~' 0 3.3 0 D~d~ o~ns and ~Wneum 3~ 246.2 ~* 197 1691 I16~ ~pha~ ~ ~.4 113 1 5.7 18 S~mach 121 42.2 ~* ~ ~.8 404~ S~ll in~sfine 3 2.0 160 1 1.4 ~Ion 63 75.0 ~ 51 70.9 72~ ~c~m, ~i~o[d junc~on, and anal ~ ~.6 I05 15 16.8 89 ~r ~ 7.3 3~s 5 3.6 139 Gallbladder 8 6.8 I18 13 9.8 Pa~ ~ Sl.S ~ ~ ~.I 76 Re~mneum, ~n~m, ~d un* 8 13.6 59 6 11.1 45 ~pi~ ~m 1~ ~.5 ~ ~ 76.2 29* ~x 3 14.8 ~* 0 L8 0 ~n~ 1~ 3~.3 ~* ~ 72.0 31~ ~er ~pim~ o~ i~ludin~ pleura. 4 5.4 74 0 ~4 0 m~i~num, and infrasonic siM ~n~ ~d join~ 0 7.2 0 0 4.9 0 ~nn~ ~bcumneo~, ~d o~er ~ft 0 ~.8 0 0 4.4 0 ~ue, S~n, mela~ma 3 14.8 ~ 0 10.4 0 B~t 74 183.9 40* Female ~niml s~m ~ I~.6 ~x 18 ~.9 46~ Co~ u~rl and u~s u~ 7 21.5 33 ~a~ 37 58~ V~n~ 2 1.2 167 V~ a~ cli~s 0 1.5 0 ~er ferule ~ni~] s~m 0 1.4 0 M~le ~ni~l s~m 18 ~.6 ~ g~nd 14 ~.8 31~ ~ T~tis 4 7.6 fi3 Penk 0 0.9 0 Scream ~d o~er male ~nimi s~m 0 0.2 O Urina~ ~m 17 ~.2 ~ 8 18.9 4~ B~der 10 ~.8 ~ 3 7.7 ~her ~ ~ 7 24.4 ~ fl 11.2 Eye 0 1.0 0 0 0.8 0 Ne~ ~m 19 ~.2 48~ ~ ~.5 47~ Bnin 15 ~.0 ~ 7 19.5 36~ ~er ~us ~m 4 9~ 43 5 6.0 End~ne s~m 0 4.1 0 1 4.0 ~id ~and 0 2.2 0 0 2.5 0 ~h~r e~ne ghnds 0 1.8 0 1 1.5 67 L~phomas ~ 51.3 74 21 32.3 ~ph~oma and reficulum ~11 ~- 26 26.0 1~ 16 16.2 com~ H~gkin~ di~ 8 16.8 ~ 1 10.3 10b ~er ~phomas 4 8.5 47 4 5.8 69 Multiple ~loma 2 10.6 19~ 2 7.8 26~ ~emias 46 49.2 93 ~ 34.6 72 ~her ~d unknown p~ma~ 36 49.4 73 24 ~ Se~ footnot~ a. t~ble ~- Risks were significantly different from 100 at the Risks were significantly different frvm I00 at the 1% level. 5% level, J.NCL VOL 6~. NO. ~,. NO%'E.%IBEK I_~'.~3 T109371739

Cancer MortaI|ty Among U,S, Japanese 1153 Oi:s Exp SMR Obs Exp SMR ! I Total rr.~,lil'nant r~eoplasms 823 7~.6 ~p. emi a~, =d ph=~x 17 21~ ~p 0 Ton~e 2 4.8 S~i~ glands 1 2.1 Gum ~ mou~ 6 4.6 ~h~ 2 1.1 ~her ~d un~ifi~ p~x ~ ?.4 Di~esd~ o~ ~pha~ 31 15.7 S~ch 193 47.2 Small in,st{he 4 1.6 Colon ~ 87.0 ~, ~c~i~oid j~cfion, and ~1 ~ ~.9 ~r ~ 5.6 ~l~ladder ~ 7.6 ~m~Wneum. ~neum. ~d us- 13 13.2 s~ di~sti~ o~ns ~pi~ ~m 212 ~.7 ~x 11 9.9 ~n; 1~ 191~ ~er ~spim~ o~ns including pleu~ 4 3.6 m~finum. ~d in~thomeic si~ ~n, ~d join~ 1 3.6 Conn~fi~. subeumn~us. ~d o~er ~ft 0 S~n, m,la~ 2 5.7 Femal, ~niml s~m ~x Co~ u~ and u~s u~ffi~ Vul~ a~ eli~ds ~her femak ~ni~l ~e ~niml s~m 65 I~.9 ~m~ gknd ~ I18.0 T~ 0 1.5 Penis 2 S~mm ~d o~er ~e ~ni~ s~Wm 0 0.2 Uri~ ~m ~ ~.7 Bla~er 26 ~.4 ~her urina~ ~ns 14 16.3 Eye 0 "0.8 Ne~ s~m 4 9.6 Brain 2 7.4 Other ~ s~m 2 2.2 End~n, ~id gland 3 1.6 ~her end~ne g~nds 0 0.6 ~phom~ 19 2~.3 ~ph~oms and reticulum cell ~r* II 13.9 comM H~kin*s dim , 6 5.3 ~her l~phom~ 2 6.1 Multiple m~lom, 3 9.7 ~emi~ 19 O~er ~d un~o~ p~ma~ 18 105 543 623.3 87~' 80 5 8.0 63 0 0 0.I 0 42 1 1.8 56 48 1 1.3 77 109 2 2.2 91 182 1 0.5 200 95 0 2.0 0 168~ 316 218.8 144~' 197~' 5 5.6 89 409~' 121 £9.5 410b 250 I 1.4 71 b 78 93.2 84 120 18 21.0 86 214~ 9 3.5 25?= 355b 37 14.5 255* 114 36 36.2 99 98 11 13.9 79 104 58 4/.6 122 111 O 1.1 0 103 57 44.8 128 111 1 1.9 ~3 28 0 2.9 0 0 0 2.6 0 35 3 6.0 50* 17 112.5 15~ 41 86.5 47* 15 21.4 70 6 23.0 26~ 14 37.4 37* 2 1.3 154 2 2.7 74 2 0.7 286 0 167 0 73~ 17 24.7 69 68~ 11 14.1 78 86 6 10.6 57 0 0 0.6 0 42 9 10.5 86 27c 6 8.3 72 91 3 2.2 136 136 4 4.1 98 188 4 3.5 114 0 0 0.6 0 75 £2 24.9 88 79 12 13.6 88 113 7 ,5.5 127 33 3 5.8 52 31< 7 8.8 80 56b 13 25.9 50b 42~ ~0 38.8 77 • ~ee footnote a. table 2. ~ Risks ~r~ significantly different frvm 100 at the c Risks were significantly differvnt from 100 at the 1% level. 5% level. jNCl. VOL. ~. NO. 5. NOVEMBER I.~ TI09371740

1154 Locke lnd King Prima~ site" Female Obs Ezp S~.P/ Ohs Exp SbfRb Total maligr--nt necplMms 346,442 371,070.7 Lip, oral cavity, and pharynx 3,816 11,331.8 Lip 68 379.6 Ton~oe 1,370 2,624.7 Salivary glands 354 778.1 Gum and mouth 669 2,398.2 Na~pharynx 377 877.2 Other and unspecified pharynx 9~8 4~74.0 D~ge~five or~n~ and peritoneum 217.896 100,401.6 F~ophagus 15,573 7,596.6 Stomach 1~0,840 17.823.8 Small intestine and colon 10,423 33.213.2 Reatum. r~t~ig~no|d junction, and anal 13~239 11,503.3 Liver 2,612 5.708.2 Gallbladder 7,312 1~19.1 Pancreu 13,409 19,945.1 Retroperitoneum. peritoneum, and un- 1,493 3,387.3 R}eeiRed digeative organs "Respiratory eystem 49.680 119,565.3 Larynx 3,673 5,492.8 Lung 40,846 112,107.4 Other respiratory or~n| including pleura, 5,161 1,955.1 mediastinum, and intrathoraeie site Bones and joint~ 2,459 2,390.5 Connective, ~ubcutaneous, and other soft 572 1.916.1 tissues Skin, melanoma 417 4,537.9 Brealt Female genital system Cer~x Ccrp,,- uteri and uteru~ unspecified V~rina Vulva and clitoris Other female ~enital system Male genital system 6,141 32"319.4 Prostate gland 4388 29,408.8 Testis 884 2,412.5 Penis 489 404.0 Scrotum and other male genital system 30 94.1 Urinary system 8,127 21,230.7 Bladder 5,424 12A19.4 Kidney and renal peivi~ 2,390 8,395.1 Other urinary organs 313 416.2 Eye~ Nervous system 1,404 11,349.5 Brain 721 3,625.6 Other ner~ct~ system 683 2.723.9 Endocrine system 983 1,381.7 Thyroid gland 714 809.5 Other endocrine glands 269 572.2 Lymphemas 7,396 17,434.0 Lympho~arcoma and reticulum cell ear- 3,949 8,933A Hodgkin'e disease 1,307 5.305.8 Other lymphomas 2"140 3,194.8 Multiple m~elcma 1,523 4.581.4 Laukemlas 10,085 18,988.7 Other and unknown primary 35,943 24,552.1 93 273.207 314,548.9 87 34 1,932 4,372,2 44 18 40 34.5 116 52 742 1,028.1 72 47 237 546.5 43 28 316 1,111.5 28 43 173 358.9 48 23 424 1,292.7 33 217 149,317 85,373.6 175 245 6,027 2,614.8 230 846 97,587 11.087~ 882 31 11.867 37,305.8 32 115 11,598 8.597.3 136 46 1.289 1,658.3 78 600 9,573 5,616.6 170 67 9,631 14,668 ~- 66 44 1,745 3,935.5 44 42 20,414 28,747.0 71 67 737 702.9 105 36 16,0~1 27,0~9.7 59 264 3,628 984.4 368 103 1,715 1,778.0 96 30 425 1,589,8 27 9 315 3,629.0 9 13,140 6~.904.4 20 38,598 47,904.1 81 7,598 13~97.8 55 23,418 10.447.7 224 6,154 22,142.2 28 309 562.7 55 454 957.5 47 916 496.2 185 19 16 37 109 38 4.538 10.068.5 45 44 2.954 4,966.3 69 28 1,323 4,743.9 28 75 258 356.3 72 12 1,019 8,467.3 12 8 533 6,520.4 8 25 486 1,946.9 25 71 1,800 1,903.0 95 88 1.668 1,415.4 118 47 132 487.6 27 42 4,373 13,281.1 33 44 2,365 6,954.2 34 25 660 3,646.7 18 67 1,348 2.680.2 50 33 1,268 3,992.1 32 55 8.136 13.386.8 59 146 26~220 22,6~3.7 116 • The ICD (Eighth Revision) codes were used. See Appendix. p. 1191. s All SMR a~e sicnifl~ant at 1% level, except for bones and ~oints of both (significant at 5%), and penis. ¢ Values for eye were not available at the time of the study. sexes, female lip and larynx. female endocrine system .~NCL VOL ~5. NO. 5. NO~.TJ, IB£K TI09371741

87) but represented an increzs~ over dmse seen in 1~0. Nadvity ~£f~ren~s w~e sm~l, ~ouzh previo~ly a hi~her Issd ¢~n Nisei ~k amon~ mal~ was in~- ~t~. In Japan, a ~se in th~ ~n~r since 19~0 has ~n ob~; the ~le m~o was up to ~ l~el o[ U.S. Japan~ ~les and the female ratio to a l~el higher ~an ~[ of ~e U.S. Japan~ women. ¢) ~ncers of [he liver and ~llbladden Mortify was high among U.S. Ja~ne~, parfi~lady males. With • e ~cepfion of liv~ ~nc~ among ~l~, the mdos for I~ei gr~dy ~ed~ ~ose for Nisei, which was consistent with the a~-s~cific profile for ~th ~ncers of the liver and ~llbladder. In Japan, mortality from liver ~n~r was below the U.S. level, bm the reverse ~s ~e for ~llbladder ~n~r. ]) ~n~r of the pan~as: Mortality ratios have ~ea~d since i~ for U.S. Jap~e~, wher~s ~e Nhei SM~ has remain~ ~low the level for whi{~ (males, ~; females, 76). ~ndnuing m ~ depressed, ~e mdos for Ja~n~ in Japan were similar ~o ~ose for Ni~i. • Canc~s of the lung a~ bronch~.--Mor~lity ~dos among Ni~ (~le, ~S; ferule, SI) were as low as tho~ among Japan~ in Ja~n, but ~e I~i SM~ (male, 10S; female, 12~) cl~rly in~ted an upward ~splace- mere. Funhe~ore, a ~r~sing risk has ~n shown sin~ I~0 for U.S. Japane~, more so ~or Ni~i than for Issei, ahhoug~ mortality for the~ ~ncers has ri~n in ~h homeland and hos~ Canc~ o] the breast (]emale).--Mormli~ was tremely low among Ja~n~ [em~es in the Unit~ Stmes and Ja~n, compar~ whh that among U.S. whi~e women; ~th coun~es showed a stable ~end over the y~ (~, 11). However, among the U.S. Japane~, a higher Nhei ~han I~i mortality ~do was obse~ed (Ni~i, 40; Issei, 17). ~is oblation was in accord with ~e findin~ of a higher age-s~fic mortality for almost eve~ a~ group and the ~sing incidence in U.S. Japane~ re~r~ by others (12). Mor~ver, in con~ {o a s~dy ~ in mortality rotes wi~ age among whi~e females, a l~eling oH in age- sp~ific cu~ ~s no~ed among Japane~ women over ~ yea~ of age in Ja~n and in the Uni~ Sm~. Similar obviations were re~rt~ by other inves~- ~mrs (I~, 14). Canc~ o] the g~ital orga~ (]emale).--Mor~ity from can~ of the ce~ix and uter~ remaln~ low ~or U.S. Japanese (ce~ix, 55; uteri, 29). In ~o~idon of the ~/ferem diagnos~c pracdc~ in Japan and the Uni{ed S~es, a combin~ ~do compm~ ~or ~nce~ o~ the ce~'ix and uterus in~ted tha~ the ~cess ratio for Japanese in Japan over the redo for tho~ in the U.S. h~ cominu~ since 19~. The redo for o~rian ~n~r was Iow~ for Japanese in J~pan than [or U.S. Japan~ and for I~i ~an ~or Nisei. In [act, Issei m~ have &op~ close ~o ~e homeland level since 19~, whereas a ~ has ~n ob~'~ in Nisei ra~os. Canc~ o~ the p~o~tate g/and.~The low ~sk among U.S. Japnn~e ~s~s only slighdy higher than that Cancer MorlaIIt'I Among U.S. Jap~mese 1155 ~erved in Japan~e in Japan, notably among Issei (Japan, 16; Issei, 5~; Nisei, M). Insl:ecdon cf the age- s#tic rotes confirmed this observation, inasmuch as the Nisei curve remained constantly above the low rotes for Japanese in Japan and was exceeded only by the elevated rates for Issei elderly, i.e., the groups with the highest ri~ -ks. Cancer of the urinary orgam.~All Japanese groups exhibited a low mortality risk, with a slight elevation over time, but the ratios for U.S. Japanese, particularly the Issei, exceeded those for the Japanese in Japan. Although environmental factors are of significance in bladder cancer, the important role of host factors was emphasized in a recent Japanese study (15). Lymphomas.~Mortality among U.S. Japanese was between the low level among Japanese in Japan and the more elevated level among U.S. whites. Except for Hodgkin's disease, for which Nisei ratios were ex- tremely low (males, 48; females, 10) and Issei ratios exceeded 100, the nativity differential was reward the slightly higher Nisei leveh. Leukemia.--Mortality among Japanese in Japan and U.S. whites was similar for young people, but for those over 45 years of age, U.S. white mortality ratios. increased more sharply than did ratios in Japan. The SMR for Issei and Nisei were intermediate between the homeland and the host country. However, the rates in Japan for the elderly have risen sharply in the last decade, which is a likely indication of environmental influences. DISCUSSION A detailed explanation of the transitional trends noted above will be attempted in a later paper. We will only comment on the confounding effects of host and environmental factors on selected cancer sites. Of particular interest is the substantive reduction in the SMR of cancer of the stomach, accompanied by a rise in the ratio of colon cancer. The absence to date of a convergence in the mortality rates of Nisei and those of U.S. whites for stomach cancer might indicate a lingering genedc influence, possibly complicated "by partial adherence of Nisei to the traditional Japanese diet. The blurring of nativity differences noted in our findings apparently represents another step toward more complete transition. The rise of colon cancer mortality close to that of U.S. whites is in keeping with the transitional experi- ence of other migrant groups to the United States. The apparent environmental effect is particularly reflected in our data, which indicate that Issei and Nisei ratios are fairly dose to one another as is the female to male mortality. Breast cancer among Japanese females has always been of special interest because of the dissimilar transitional experience exhibited among European and Japanese migrants to the United States. The failure of a rise in female breast cancer risk to the U.S. level may be partially attributable to initial settlement of early JNCI. VOL f.,5. NO. 5. NOVEMBER 1953 TI09371742

11E8 Locke tnd Klng Japan~ migrants in rural areas and to the incomplete accuhumdon o[ the Japanese migrams in dietary habits. Although a slighdy higher Nisei than Issei mortality was observed in our dam, a full ~ransition in the near future seems unlikely, particularly in view of the noted leveiing off of the age-specific curves among U.S. Japanese women over 60 years of age. The ele}'ated mor~alhy level among U.S. Japanese for cancers of the lung and bladder and leukemia seems to reflect differentia] effects of the urban environment, industrial employment, and modern living, although host ~actors cannot be ruled out. A fuller interpretation of the findings noted above would require systematic investigations of the differing roles that host and environment play in the ~ransio tional experience of migrants [e.g., the role of dietary and other cultural practices in relation to gas~roimes- dnal and breast cancers (I6-2I) and cancer-related diseases like ischemic heart d/sease and s~roke]; these will be examined in an extended analysis that is in preparation. REFERENCES (1) SMn'H RL. l~e~orded and expected mortality among Japanese of the United States and Hawaii, with special re(erence to cancer. J Natl Cancer Inst 1956;17:459-47~. (2) BV£Lt P, Dt.','4.~ JE. Cancer mortality among Japanese hsel and Nisei of Califomla. Cancer 1965;8:656-664. (J) STAST~WSK! J. HAENST-EL ~,V. Cancer mortality among the Polish- born in the United Stat~. J Natl Cancer Inst 19~5;~5: 291 °297. (4) HAENratL W. KttRIHA~A M, Studle~ of Japanese migrants. I. Mortality from cancer and other di~ases among Japanese in the United States. J Nad Carmer (~ W~rld H~hh O~don. W~IA H~lth Smthfi~ A~I. 1~-1~7~. ~nm: WHO, (7) BmL~X JC HI. ~ F. Si~ifi~e ~ao~ Poh~n ~hl~ to iu ex~don. Bi~t~ !~;~:639-~3. (S) SEGI M, Graphic pr~nmdon of ~ncer indden~ by lite and by ar~ an~ ~pulation. Nago~, Ja~m ~ lmdtum of ~n~r Epid~iol~, 1977, (9) S~ct M. KL'mH~ M. ~nc~ m0~li~ f~ coun~, No. 6 (1~1~7). Tokyo: Jpn ~c~ ~, 1~2. (10) ~ta~ P, ~NS~L W. ~e epidemiolo~ of la~e.~wel ~n- (11) D~v~ SS, StL~X&~tAN DT. ~ncer ind~ and m~lity tren~ in ~e Unit~ S~tes: 1955-74. J Nad ~ In~t 1978; ~:54~571. (12) ~'t~ P. ~anging in~de~e of br~t ~ncer in Ja~n~- Ame6~n women. J Natl ~ncer Imt 1973;51:1479-1483. (13) D~W~ F. ~ent time t~nd* in bt~st ~ncer ~ev M~ 1978;7:1~-167. Ja~n~ women li~ng in Hawaii. Gan (15) OHso Y, A~t ~ Epidemiol~ of bladder ~ncer ~th~ in Ja~. Gan 1977:~:715-729. ~ong Ja~n~ in Hajji. J Nad ~nm Inst 1972;49: ~9-988. (17) ~L W, B~to ~V. S~GI M, Kcttu~ M, ~wei ~er in Ha~ihn Jaunt. J Natl ~ Inst 1973: 51:!7~-1779. (18) ~N~L W, Kctma~ M.~ ~,SutMu~ K,S~Cl M.Stom- ach ~ncer in Ja~n. J Nad ~n~ Imt 1976~:26~278. (19) S~MMt~MANN G, ~S~k W, ~ F. The epidem~ol~ic ~thol~ M ~ ulcer and ~c ~rcinoma ~ong Ja~n~ in Hawaii. J Natl ~ncer ]mr 1977;58:13-20. (20) ~N~L W, L~ ~, Szc~ M. A m~conuol ,tudy of bowel ~n~r in Ja~n. JNCI 19~:~:17-~. (22) ~YAMA T. Epi~miolo~ of br~st ~n~r with s#al refer- ~ to the role of diet. Prey M~ 1978;7:173-195. j.~cL VOL. ~, xO. 5. NOVE31BER, TI09371743

- Cancer in Alaskan indians, Eskimos, and Aleuts' V Anne P. Lanler, M.D., M.P.H., 2 Wiiitam J. Blot, Ph.D., ~ Thomas R. Bende¢, M.D., M.P.H., ~ and Joseph F. Fraumanl, Jr., M.D. ABSTRACT--~t|nda~I Incidence ratios for cancers that oc- curred during 1969-73 ~mong A~k~ N~ (Ind~ns ~d Esk(m~A~u~) wart repoRed. ARhough d~ts suggsst~ that ~e overs;] rate of c~c~ In th~s population wM ¢~ose to that of U.S. whit~ d~mnces e~d for c~ln cancer slice Thus inc~a~d dsks ~n N~n N~lves were obse~ for n~ph~ngell, ~livaw g~d. kldnW, and glflblldd~r ~nd I~v~ ~nce~. Con- ~l~lly. d~s~ dsks were f~und for c~ncs~ of the lung, IIwnx, bladder, prostate gland, br~sk and corpus uterus sad for m~llnoml ~ lymphom~.~JNCI ~: 11~-115g, 19~. This paper updates the cancer experience of Alaskan Natives (1, 2), a population traditionally thought to have little cancer. The SIK were tabulated separately for Indians and Eskimo-Aleuts so comparisons with populations elsewhere are facilitated. Alaskan Natives include Indians, Eskimos, and Aleuts. All are thought to be originally of Asian origin. Alaskan Indians comprise several tribes, some of which are linguistically related to Indians of the southwestern United States, whereas Alaskan Eskimos are related to the Eskimos of Canada and Greenland. Aleuts have traditionally occupied the Aleutian Islands and the Alaska Peninsula and possess some cultural and lin- guistic traits in common with the Eskimos. In 1970, the Alaskan Native population was 50,900 (55% Eski- mos, 13% Aleuts, and 32% Indians) and constituted 17% of the population of Alaska (3). Health care of the Natives has been the responsibility of the Alaska Area Native Health Service, Indian Health Service, Public Health Service, U.S. Department of Health and Human Services. Care is administered through a system of eight hospitals, including a referral hospital in Anchorage, the Alaska Native Medical Center. METHODS An attempt was made to idendfy all new cases of cancer occurring in Alaskan Natives during the 5 years 1969-73. The cases were ascertained from coded discharge diagnoses from all Alaska Area Native Health Service hospitals, the surgical pathology files and tumor registry of the Alaska Native Medical Center, a regional tumor registry, and death certificates. The medical record and pathology report on each patient were reviewed, and only cases confirmed by histology, surgical exploration, or roentgenogram were included. Over 95% of the cases had histologic confirmation of the diagnosis. The sites of cancer were coded according to the classification of the ICD (Eighth Revision). Expected numbers of cancers by sex and site were calculated by application of the age-, sex-. and site- specific cancer incidence rates from the U.S. Third National Cancer Survey of 1969-71 to the person-years at risk for the Native population for 1969-7$. Age- and sex-specific population estimates for the 5-year period were determined by extrapolations based on the 1960 and 1970 censuses. Expected numbers of cancers for the 2 subgroups, Indians and Eskimo-Aleuts, were obtained by multiplication of the expected numbers for all Natives by the proportion of the total popuiadon representing each ethnic group (0.$2 and 0.68, respec- tively) because the sex ratio and age structure among Indians and Eskimo-Aleuts were generally similar. Ethnic designation of each patient was based on his or her self-classification and checked for consistency by comparison of the patient's birthplace and residence with knowledge of the traditional geographic distribu- tion of each ethnic group in Alaska. The SIR, the observed numbers of cases identified over ehe 5-year period divided by the expected numbers and then multiplied by 100, were calculated for each cancer site. We tested deviations of these ratios from 100 for statistical significance by referring to published sig- nlficance factors for ratios of a Poisson variable to its mean (4). RESULTS The numbers of cancers among all Alaskan Natives were lower than expected (SIR=84 for males and 83 for females). However, the ratios were no longer significantly low when 26 additional cases identified by death certificate only were included. For certain can- cers, the SIR among Alaskan Natives deviated signifi- cantly from those expected. Elevated rates were seen in males for cancers of the nasopharynx and liver and in females for cancers, of the nasopharynx, salivary gland, gallbladder, and kidney. Low rates prevailed among males for cancers of the larynx, lung, prostate gland, and bladder, for melanoma and lymphoma, and among females for cancers of the breast and corpus uterus and A~IIt~VL~-TION$ US£D: ICD=~ International Classification of Di~ases: SIR:astanclardized incidence ratios. t Pre~nted at the Workshop on Populations at Low Risk of Can- cer conducted at Snowbird, Utah. Augtat 2~-25, 1978. " Alaska Investigations Division. Bureau of Epidemiology, Center for Disease Comrol, )mchorage, t~as'ka 99501. ~ Environmental Epidemiology Branch. Division o{ Cancer Cause and Prevention, National Cancer Institute. National Institutes of Health. Public Health Service. L'.S. Department of Health and Human Services. I~thesda, McL 20205. 1157 j.~cL VOL 6~. .~o. 5. NOVEMBER If,80 T[0937174-4

1158 LanZsr, Blot, Bender, and Fraumen! Alukan Indians Als.~,an Eskim0-Ateuts Male Female Male Female SIR Obt SIR Ot~ SIE Obs SIR Ol~t To~! malignant neoplasms 89 64 92 Lip, ~ ~. ~d p~ ~ 2 N~phs~x 0 0 ~v~ o~ ~d ~n~um 11~ ~p~ 1~ 1 0 S~m~h ~ 1 417~ ~lon 1~ 10 111 ~, ~e~si~oid junction, ~d ~1 167 6 ~r 0 0 G~der 0 0 ~0~ ~r bilia~ ~ 667 2 1.~ P~¢~ ~ 2 77 Rmpi~ ~m 47~ 8 ~ 0 0 0 L~g ~ 8 ~r ~pi~ or~ 0 0 0 ~n~ ~ jo~ 3~ 1 333 Sk~, ~lsnom~ 0 0 O~ B~ 47 F~m~ ~ni~ ~m ~ 47 0 P~ ~nd ~ 9 T~ ~ 3 U~ ~m ~ Bl~d~ ~ 1 Ki~y ~d ~n~l ~l~s 2~ ~r ~ o~ 0 0 0 ~e 0 0 0 Ne~ ~m 0 0 End~ne ~m ~id ~ 0 0 ~er ~ne ~ds 0 0 0 L~phom~ 61 2 40 Ot~r ~ un~o~ p~ma~ 174 4 3 • ICD (Eishth l~wiaion) codes were treed. ~' P<O.O~. ¢ P<0.0L 56 81} 125 79¢ 102 4 138 II 346t 9 1 286 2 800~ 4 2 1,800~ 9 I,~)0 I 21 181 49 147¢ 38 0 2~0 $ ~ 5 167 9 ~2 1 0 ~ 1 0 0 6 81 11 1~ 16 4 107 ~ 136 6 1 615' 8 0 0 3 ~ 2 1,~~ 7 2 0 0 0 0 I I~ 5 143 4 3 71 ~ 68 0 O* 0 0 0 $ 78 ~ 62 4 0 1~ 1 ~ 1 ~ 2 0 0 0 1~ 2 100 1 0 0 0 ~ 1 4 2 ~ 6 1 0~ 0 0 ~ 7 1 77 9 1 5 ~ 8 I~ 1 51 5 37 1 4 71 3 ~ 4 0 0 0 0 0 0 0 0 ~ 1 1 73 3 1~ 3 2 0 0 2~ 0 0 0 0 0 1 37 3 0~ 0 0 0 0 0 0 5 167 8 114 4 for lymphoma. These observations are described else- where in more detail (2). Th~ SIR for the 2 Alaskan Native subgroups, Indians and Eskimo-Aleuts, are given in table I. The SIK for all cancer sites combined were lower among Eskimo- Aleuts than among Indians (81 vs. 89 for males and 79 vs. 92 for females). Although based on small numbers, the site-specific SII~ for Indians and Eskimo-Aleuts showed similarities, especially for the low-rate tu- JNcL VOL. ~. NO. 5. NOYE-MBER 15~ mors. An exception was prosmm cancer, for which dw low rates were limited to Eskimo-Aleuts. Both Eskimo-Aleuts and Indians had increased risks o[ developing gallbladder cancer, but only Indians were prone to cancer of other biliary passages" (i,e., extra- hepadc bile duct). The high rates of kidney cancer were noted mainly in Indians. Also, the rote reported for stomach cancer was significantly high for Indian females, hut not for other groups. T109371745

DISCUSSION These inddence dam show an unusual disu'ibution of cancer among the 50,000 Indians, Eskimos, and Aleuts residing in Alaska. Ascertainment of cases is considered almost complete; n~,ertheless, missed cases and census unde~esthnates may have altered the rotes. l)espi~e these potential limitations and the small hum- "bets o[ cases, the [igures still allow interesting com- parisons to be made with other population groups. Canclr In Alaskan NatNes 1159 REFERENCES (I) ]~LO'~" 1~']. L.~,N~F.R A. FI~I.'ML'~'I JF ~. E~D~ ~. ~r~r mar- mli~y among .~s~n b~dse~ 1~9. J Nazi ~cer Ins~ 1975:55~7-5~. ~} ~zz AP, BZ~ER TR. BLOT WJ, F~L~z~ JF Jx. Ht~zsT W~ ~ncer ir~id~ce in Alas~ Native. Int ~ ~ 197~; 18:4C~-412. {3) U.S. Bu~u of the Ons~. U.K census ~pulatiun I~0. Vol I. ~sdcs o[ ~e ~pu~lion, ~n 3, ,Mas~. W~hingmn, D.C.: U.S. ~ ~im Of L 1977. {¢) ~uz ]C Ill, Eu~a F. Slgnifi~e factors for the m~o of a Poismn ~a~ble to its ex~tion. Biometrics 19~:I~ j.~c~, vol_ ~. NO. ~. NOVEMBER TI09371746

Cancer Mortality in Nonsmokers in the United States Moderator:. Peter Greenwald, M.D. TI09371747

Cancer Mortality Among Nonsmokers In an Insured. Group of U,S. Veterans, F_ Rogot = and J. Murray ABSTRACT--In • l~.yeer f011ow-up of Insured U.S. vetarena, =tandlrcliZed mortality retlol (SMR) for cancer were ealeuleted for r~nlmoker= (55,049); ~ the I~d~d. c~n~ r=t~ of ~1 r~ndln~ (2~,~) wire ul~ For ~1 ~ncl~ non~okl~ had =n SMR 0f 71%; for lung ¢~c~, tbW h~ In SMR of 21~ The ~t= for non~moker~ ware fu~her ~=l~ed with risk of ~ncer ~d ~ccoNIng to r~ldence ~d usu~ occupation ~d lndu=tw. Among the nonsmoker, SMR were esp~l~lty !~ for d~tl=~, ~entem, ~d ~em. ~d for the el~tri¢ light and power Ind printing indu=Ules. ~ dl~=ion of ~e counW, SMR for nonamokm~ wl~ ~p~tllly low for the Moun~ln S~te=.~JNCI ~: 11~11~, lg60. Follow-up of a defined population of U.$. veterans that was planned by Dorn in 1952 with the general objective of studying the relationship of the use of tobacco, residence, and oci:upatlon to mortality has furnished dam for several studies by Dorn (I, 2), Kahn (3), and llogot (4-~). The present study examines cancer mortality in a group of veterans at lower risk of developing cancer, i.e., nonsmokers. SUBJECTS AND METHODS With the cooperation of the Veterans Administration, policyholders of U.$. Government Life Insurance were selected for study. Th/s insurance was available to persons who served in the U.S. armed forces from 1917 to 1940, Most of the policyholders were veterans of World War L All persons with an active policy at the end of 195~ were included except for a few special groups, such as persons with total and permanent disability. Beginning in January 1954, a questionnaire requesting informa- tion concerning the use of tobacco, usual occupation, and indusu'y was mailed to about 300,000 policyholders. Usable replies were received from about 200,000 per- sons or 68% (table 1). This group of respondents was labeled the "195-} cohort." A second questionnaire was mailed to the nonrespondems beginning in January 1957. Usable replies were received from an additional 49,000 policyholders (labeled the "1957 cohort") for a total of 248,000 or about 85% of eligible persons for whom information concerning the use of tobacco, occupation, industry, and residence was obtained. The mortality experience of nonrespondents was recorded. Whenever a claim was filed for the paymem of a policy, a copy of the death notice, usually a copy of the official death certificate, was sent to the office conduct- ing the study. Additional medical information, which included verification of the causes of death entered on the death certificate, the procedures used to establish these diagnoses, a diagnosis of cancer even though it was not considered to be an underlying or conwibumry cause of death, and the histologic type of cancer, was requested from the physician who signed the death certificate or from the hospital where the death oc- curred. Verification of the cause of death was not requested if the death occurred outside the United States, was due to an accident, or was certified by a coroner. Replies m more than 99% of the letters of inquiry were received. These special verification pro- cedures for cause of death were observed for the first part of the study only, i.e., from 195-t through 1982. All smoking classifications used in this report are based on responses to the 195'~ and 1057 questionnaires. The category "nonsmoker" refers to persons who never smoked cigarettes, cigars, or pipes; also included are the "occasional" smokers (persons who smoke once in a while but not every day). Almost all policyholders were white males, drawn mainly from the middle and upper socioeconomic classes. The Veterans Administration provided names of all policyholders known to have died in the study period. For about 75,000 persons whose insurance policies had terminated in 1963-69, special searches were made by the Veterans Administration index staff to ascertain deaths. The overall mortality follow-up, with respect to the fact and year of death, is considered to be almost 100% complete. The full de~ails of the various follow-up procedures appeared in (2-3). With respect to cause of death, special searches were made in state vital statistics offices to obtain copies of certificates for 84,644 deaths needed to complete the latter part of the 16-,year follow-up. All but 2,5~1 death certificates were located. Causes of death were coded according to the ICD (Seventh Revision, 1955). The underlying cause of death, used in this report, was available on 97.6% of all deaths (table 1). The SMI~. for nonsmokers were calculated with the age-specific probabilities oE death in the study period of all respondents used as standard rates. Expected A~II~ZVlACIONS USED:. ICD=:l~nternational Classification of Diseases: SMg=standardized mortality ratioO). = Present~,d at the Workshop on Populations a~ Low Risk of Can- cer conduct~l at Snowbird. L'rah, August ~J-25, 1978. = Epidemiology Branch, Division of Heart and Vascular Diseases. Nadonal Heart. Lung, and Blood Institute, National Institutes Health. Public Health Service. U.S. Depanmem of Health and Human Serv/ces. Bethesda. Md. 20205. ~ Biom~t~. Branch. Division of Cancer Cause and Prev~mi,,n. donal Cancer Institute, National Institutes of Heahh. 1163 JNCL YOL ~3. NO. 5. NOI"E.MBER I.~0 T[09371748

1164 Roger and Murray Death certificates No. in No. cf C~ort each deaths Not Percent cohort Fcund four~l fo~d 1954 196,820 69.858 68,446 1A12 98.0 1957 49~26 16,877 16.521 3~6 97.9 No reply 45.912 20.828 20,0~5 773 96.3 Total 293,958 I07,503 I0~,022 2,641 97.6 numbers of deaths by cause were-calculated separately for the 1954 and 1957 cohorts and then combined. Age groups used were those shown in table 2 except for state data in which the combined age group 30-54 years was used instead of the separate age groups of ~0-34, 35-44, and 45-54 years. The more detailed breakdown by states was not readily available. The difference in expected cancer deaths calculated by means of the coarse and more detailed age groupings was too small to affect the overall SMR for total cancer or for lung cancer. RESULTS In the tables that follow, a small number of cancer deaths allocated to female sites, as well as a few deaths with uncertain causes, have been combined with the "other and unknown" category and are included with the "total cancer" deaths. These number 38 of the 133 other and unknown group and represent about I% of the cancer deaths among nonsmokers. The effect on the SMR for total cancer is considered negligible. Some general characteristics of the population under study are treated in tables 2 and 3. Table 2 shows the age breakdown of the total group of respondents and of the nonsmokers among them at the beginning of follow-up. For each group, the ages of 55-64 and 65-74 years~dominate. About two-thirds of our population were between 55 and 65 years of age at the start of the study. Residence at the beginning of the study for non- smokers and total respondents compared with U.S. TABLE 2.---Number and percent of nonrmekers a~t total rerpende~ by age: U~. t~teruns study° Z95~-69 No. Percent ~on- To~] Non- Total smokers respondents smokers respondents 30-34 1.060 7.464 2 3 35-44 4.205 23,891 8 I0 4,5-~4 2,078 11.5~9 4 5: 55-64 37.8,50 164.399 69 66 65-74 9.319 38.6~5 17 16 75-84 537 2,048 1 1 To~al 55.049 248.046 100 100 Age range is tha~ at l:e~dnning of foll~w-up. Residence~ Percent Non- Total U.S. sn:oke~-s respon- white dents males Urbtn~ 80 83 63 Rural 20 17 37 Div~ion" New England 6 7 7 Middle Atlantic 19 20 21 East North Central 20 20 21 West North Central 12 11 10 South Atlantic 12 12 12 Enst South Central 4 4 7 West South Central 7 7 9 Mountain 4 4 4 Pacific 16 1~ 10 Total I00 I00 IO0 " Rasldence is indicated for beginning of follow-up (1954 or 1957) for veterans and as of April L 1950, for U.$. white males, ~ Urban-pkces of ~2,500 in populffition ns of April 1, 1950: rural-all other. ~ Divisions sre states in designtted census areas. white males in 1950 is given in table ~. More of the nonsmoking veterans lived in urban areas than did their counterparts in the general population: 80.versus 63%. By census divisions of the country, the percent distributions were generally similar, except that a larger proportion of the nonsmokers came from the Pacific States than did the U.S. white males: 15 versus 10%. A comparison of nonsmokers and total respondents with U.S. white males with regard to major occupation 0 ~ SMR ~0 ~O ~1) 100 ! I l f I ! 0 2~ ~O eO 80 100 SMFt TL\-r-Hst'R[ I.~S.MR for cancer si~ for nonsmokers: U.S, veterans study, JNCL rOt- dS. .xO. 5. NOVEMBER I~.~3 TI09371749

Cancer MortaI|ty Among Nonsmoker~ 1165 To~I maUgnant r, ecp|asms Lip. oral c~vity, and pharynx Lip (140.0-140.9) Tongue (141.0-141.9} ~allvary glands (142.0-142,9) Gum and mouth (143.0-145.9} NMopharynx (147) Other and unspecified pharynx (146.0-146.9,148.0-148.9, 149) D~ge~tive organs and peritoneum F~ophagu~ (150) Stom&¢h (151.0-151.9) Small intestine (152.0-152.9) Colon (153.0-153.9) Reaturn, rectosigmold junction, and anal canal (1§4.0--1~4.2) Liver (1~.0, 197.8) ~ Gallbladder (156.0) Other biltary l~ssages (155.L 156.1-156.9) Pancreas (157.0-157.9) l~troperitoneum, peritoneum, and unspt~ifiod digestive orc~ns in- cluding abdomen (153.0-158.9. 159, 195.0.) Respiratory system Larynx (161.0-161.9) Lung (162.0-162.1) Other respiratory organs including pleura, medisstinum, and intra- thoracic site (160.0-160.9. 163) Bones and joints (170.0-170,9) Conn*ctive. su~utaneous, and other soft tissues (171.0-171.9) Skin. melanoma (172.0-172.9) Breast (174) Male genit-q system Prostate gland (18,5) Testis (186) Penis (187,0) Scrotum and other male genital s~mm (17a.5, 167.8, 187.9) Ohs 2,745 24 0 9 5 4 1 1,092 19 194 8 444 155 Exp SMR 3,8,~.7 71 83,7 29 1,6 0 23.3 39 8.2 61 18.2 27 6.8 59 £5.6 4 1,304.5 84 67.0 28 228.0 85 10,8 74 478.1 93 " 164,7 94 59 97,5 61 195 241.1 81 18 17.3 104 190 890,9 21 6 33.7 18 175 844,9 21 9 12.3 73 16 14.4 111 11 15.2 72 27 35.8 75 480 460.4 93 423 453.2 93 5 4.5 111 2 2.7 74 O]:s Exp SMR Urlr,~ry system 217 273.2 79 Bladder a~d other urinary organs 125 1~.~.3 74 (I~8. 189.2-189.9) Kidney and renal pelvis (189.0, 92 104.9 189.1) Eye (190) 6 3.2 188 Nervous system I02 101.9 100 Brain (191) ° Other nervous system (192,0-192.9) Endoer/ne system 17 16.8 101 Thyroid tdand (193) 15 11.1 135 Other endocrine glands 2 5.7 35 (194.0-194.9) Lymphomas 193 18~.5 105 Lymphosareoma and reticulum cell 132 122.2 108 s~reomas (200.0-200.1) Hodgkin's discaso (201) 86 40.8 88 Other lymphome* inelnding my¢o- 25 20.5 122 sis tungoides (202.0-202.9) Multiple myeloma (203) 87 74.1 "117 Leukemia 162 200,'/ 81 I~mphtti¢ (lymphocytic) Acute (204.0) Chronic (204.1) Other (204.9) Other leukemia (e.g.. granulocytic or monoeytic) Acute (205.0. 206.0. 207.0) Chronic (205.1, 206.1. ~7.1) Other (205.9. 20~.9o 207.9, 207,2) Other and unknown prlmaryc 171 197.4 87 (195.1o195.9. 196.0-19~.9. 197.0- I97.7. 197.9o 198.0-198.9. 199.0- 199.1. 173.0-173.9) • These records were coded with the use of the ICD (Seventh Revision). Their equivalent numbers in the ICD (Eighth Revi. siqn) are shown ~ paventhese#. ° The .ICD (Eighth Revision) code numbers for 208 (poly- cythemia vera) and 209 (myelofibrosis) are excluded. I! any total malignant neoplum numbers ~re to be used. these must be included here. and major industry (dam not shown) revealed that nonsmokers were distributed among occupational and industrial categories in the same manner as was the total study population. As expected, the study and U,S. populations were not distributed in the same way on these factors; the veterans occupied more professional and managerial positions in business and public ad- ministration. The main results of our study are summarized according m SMR in tables 4 and 5 and in text- figures 1-7. In text-figure 1, SMR are shown in rank order for nonsmokers by body g/stem. Sites at the top (respiratory system and lip. oral cavity, and pha~'nx) are strongly related to smoking, with SMR of 21 and .09, respectivdy. Those at the bottom (nervous system, endocrine system, lymphomas, bones and joints, and multiple myeloma) have SMR of 100 or greater and are not related to smoking. Comparative cancer mortality levels for nonsmokers are shown in text-figures 2 and 3 by major occupation and industry, respectively. Farmers had the lowest SMR (66), whereas sales workers and private household and service workers had the highest (79). Overall, SMR were not vastly different by major occupation nor by major industry,. The lowest SMK seen was for agricul- ture (65), and the highest was for the wholesale and retail trades (77). SMK for nonsmokers for selected occupations • t00) are ranked in text-figure 4. Dentists had the lowest value (42) followed by carpenters (,t8~ and lawyers (56). At the other end of the scale were auto mechanics (79), mail carriers (80), and doctors (82). At the lower end of the scale for nonsmokers in selected industries (n>~0), ranked in text-figure 5, were the electric light and power services with an SMR of 57. the printing industry, with 58, and legal services jxct. rot.. ~. XO. 5. NOVEMBER t_c$3 T109371750

1166 Rogot and Murray Tx~ 5.~SMR f~ t~al ca~.~r arzI l~-~ ca~ f~ Dive,on of T~tal c~ncer Lung ean~er residence O~ Exp SZ, IR Olin Exp SMR New £01 244.0 82 6 53.5 11 Engl,,nd Middle 563 744.7 76 40 153,8 24 Atlantic East No~ ~ ~1.6 69 31 176.2 18 W~t ~o~ ~ 470.6 ~ 19 103.4 18 Atlan~e EMt ~u~ 110 I~5.~ 71 6 ~.8 18 W~t ~u~ 185 2~.6 65 Ii ~.0 17 ~n~l Mop.in 97 I~.9 ~ 8 ~.9 Pac~c ~9 ~1.3 71 31 116.0 ~ide 6 9.5 ~ 0 2.I 0 Uni~ S~s To~l ~746 8,~,0 71 1~5 ~.8 21 • Obs=~s~rved, exp=expeeted. b Div/si0ns are states in designated census areas. with 59. Higher 5MR were observed for both motor vehicle manufacmr/ng (85) and retailing (I02). Text-figure 6 gives the SMK for total cancer for nonsmokers by population size of place of residence at the beginning of the study. No consistent pattern was observed, and little difference by size of place of residence was seen. This was also u-ue for each of the following major cancers studied: stomach, colon, pan- creas, lung, prostate gland, bladder, and leukemia (data not shown). SMR 20 4O 6O I I I I 1 ~, I , , 0 2O 40 SMR T~.XI"-~Gt'~.~ ~--SMK for to~! c~,nc~r for nonsmok~s by m~jor oc. cupation: U.S. ~rans study. 1954-~g. j.~c~. VOL 6. NO. 5. NOVEMBER SMR 0 20 40 60 I I I I I 0 2o 4o eo SMR TF-'Cl'-HGURE 3.~SMR for tor=l cancer for nonsmokers by major indusu3,: U.$. veterans study, 1954.-69. The SMR for total cancer for nonsmokers by state of residence at the beginning of the study ranged from $2 for Wyoming to 109 for New Hampshire (text-fig. 7). For all cancers, the Mountain States had the lowest mortality redo (58), and the New England States had the highest (82), whereas New England States had the lowest SMR (11) and Pacific States the highest (27) for lung cancer (table 5). DISCUSSION In this population of insured veterans, total cancer SMR 0 ~0 40 ~0 ! I I I l I , f f f I I 0 20 40 60 8O 100 SMR T~-XT.RGVRE -L--S~.IP~ ~or total c~ncer for n~nsm,~kers by TI09371751

I ! 1 1 ,,I I l I SM~ T£.~r.RGV~.t 5.mSMR for total cancer for non~moke~ by ~l~ctecl industry:. U.S. ~mmns study, 1954-69. C~nc~r Morality Among Non~moke~ 1167 SM~ 80 11~ 03S~VED I I 1~0.000+ 13ZS I I ,,, ! I I 0 20 40 60 80 1~ SMR T~XT-~G~ 6,~SM~ for to~l ~ncer for non~oke~ by ~pub- ~on si~ o~ ar~ of ~i~c~ U~. x=~ns study, 1954-69. mortality among the nonsmokers was 71% of that for all respondents. Among the nonsmokers. SMR were especially low for certain workers, e.g,, dentists, car- pentets, and lawyers and for certain industries, e.g., electric light and power, printing, and legal services. By division or section of the counuT, SMR for non- 47 81 61 61 TEXT-~qGL'X£ 7.oSMK for tc"~l cancer for nonsmokers by state of residence: U,S. veterans study, 1954-69..~terisk indicates <10 obsen'ed car~er c'~ s. J.~CI. VOL. ~. .~'O. 5. NO't~MWrl~ I~:) T109371752

1168 RoGot Lqd Muway smokers were especially low for the Mountain States, with several of these States exhibiting values below 50. The key question as to how mortality levels for veterans compare with those of U.S. white males needs to be addressed. For total mortality, estimates by Dorn (l) and Kahn (3) placed the veterans at about 70% that of the U,S. mortality levels. If total cancer mortality among veterans is also assumed to be about 70% that for the United States, we may muldply our SMR by 0.70 to obtain a figure for our nommokers compared with the U.S. white males. Thus we have for total cancer mortality: 0.71 x 0.70=50%. For the major sites we would have: stomach, 85 × 0.70=~)~; colon, 93 X 0.70=65%; pancreas, 81 X 0.70--57%; lung, ~-I X 0,70= 15%; prostate gland, 9~ X 0.70--65%; and for leukemia, 81 x 0.70:~57%. These esdmates should be considered as rough ap- proximations because we have no simple way to verity the accuracy of the 70% figure as applied to any specific cause of death. REFERENCES (I) DO~'~ HF. The mortality o~ ~rnoker~ and nonsmoker. In: Pr~ c~ings of ~e ~ai 5tathti~i ~don of ~e A~d~n S~- ~sti~l ~afion. Wa~ington. D.~: Am S~t ~, 1958: ~4-71. (2) --. To~o cons~p6on ~d mo~lity f~ ~ncer and oth~ ~. Public H~I~ R~ 195~74~81-59& ($) ~u~ ~ ~e Dora stay of s~g and mo~iity ~ong U.S. ve~n~ Rein on eight a~ one-half y~ of obla- tion. Natl ~nc~ Inst M~o~ 1~;19:1-1~. (¢) R~ E. Smoking an6gene~l ~r~li~ among U.S. vete~m. 195401~. W~hington. D.C.: U.S. ~ P~nt Off. 1974 [DHEW ~bli~6on No. ($) ~. S~king a~ mor~llty ~on~ U,S. ~te~ns. J C~onic Dis 1974~:189-20~. (~) ~. Smoking and life ex~n~ ~ong U.S. ~ete~ns. Am JNCI. YOL, 65. NO. .~. NOVEMBER T109371753

Cancer Mortality in Nonsmokers: Prospective Study by the American Cancer Society, Lawrence Gerflnkel, M.A. 2 ABSTRACT-.,.~tandardlz~d mortality retloe ($MR) were computed for nonsmokerl who were c~mpared with the total American Cancer Society etudy population for slte`= of cancer, The SMR were 64 for males and ~ for females for total cancer. Sites with low SMR In males (<30) Included lung, IIWnx, pleura, tongue, mouth, |rid pharynx. SMR less than 60 were observed for cancer= of the esophagus ~nd bladder. For femate nonemoker¢ relatively low SMR were obsewed for lung, I~ynx, and phawnx ¢lnClrl. Death retes for cancers of all sites in females exceeded t~os~ In males In the age groups of 3~.-44. 45-54. and 55-64 years. Death ret~ were about the ¢tme in the 85- to 74-year-old group and about 40% higher in males than in females 75 years old and older. Lung cancer death rates In nonsmokers among both males and females were low and ranked between fourth and sixth In males and fifth to eleventh in females In the 5 age 9roupI.---JNCI 65: 116g.-1173. 19~0. Nonsmokers are a special group of people who enjoy a favorable mortality experience, which has been dem- onstrated in many prospective epidemiologic studies covering nonsmokers in various groups (veterans, phy- sicians, organizations, and o~cupationally exposed work- ers) and in many countries, including the United States, Canada, England, Sweden, and Japan. The mortality advantage of nonsmokers is not re- stricted to one disease. Nonsmokers have lower mortality rates from ischemic heart disease, cerebral vascular lesions, aortic aneurysm, duodenal and gascxic ulcers, and a number of other diseases including cancer. This mortality advantage has been shown to persist when nonsmokers and smokers are subgrouped by urban- rural residence, occupational exposures, family history., educational level, religion, and exercise. The same difference in death rates existed when groups of smokers and nonsmokers were matched on several variables This report presents derailed mortality data on non- smokers in the prospective epidemiologic study of the ACS for total cancer and for cancer of each site by sex and age groups in a 12-year period. The SMR are based on the total mortality experience of all subjects enrolled in this study. MATERIALS AND METHODS Scatting in October 1959, more than 1 million men and women in 1,121 counties in 25 States3 were enrolled in a prospective epidemiologic study. They were asked to complete a derailed four-page confiden- tial questionnaire on family history, history, of diseases, present physical complaints, occupational exposures, education, and a number of habits including eating, drinking, and smoking. The questionnaires were oh- rained by 68,000 volunteers who were asked to enroll people they knew well and would be able to report on over the next several years. Volunteers enrolled as many persons as they could, and ~he mean number enrolled per volunteer was 16. Enrollment was by households, with at least one member of the household over the age of 45 years. Each year the volunteers were asked to report whether the persons on their lists were alive or dead; if a person had died, the volunteer entered the date and place of death. They were also asked to supply information on changes of address or changes of name by marriage. Every other year fdr 6 years, the volunteers asked the subjects to complete a brief confidential questionnaire giving derails on hospitalizations, smoking habits, and several other questions. In 1986, after six annual follow-ups, the tracing was discontinued; at that time, 98.5% of the subjects had been successfully traced. The decision was made to resume the tracing in 1971 and, in 1972, the last follow-up, which included distribution of a questionnaire, was done. Follow-up through 1971 and September 1972 resulted in the successful tracing of 98.4 and 92.8% of the subjects, respectively. Death certificate procedures.~When the subjects were reported to be dead, death certificates were obtained from the health departments in the state (or the foreign country) in which the subject died. For deaths re- ported to be from cancer in the first six follow- ups, a letter that was sent to the certifying phy- sician or hospital in which the subject died requested additional information confirming the primary site of cancer and the basis of diagnosis. However, this procedure was not done for the second 6 years of the study. In the first 6 years, information was obtained confirming the primary, site of cancer for 78~ of subjects who were reported to have died of cancer. There was microscopic confirmation of the primary ASa~V~ATIO.s;S t'S~:D: ACS = American Cancer Societ.v: ICD = Inlerna- tlonal Classification o[ Diseases; SMR =standardized mortality rado~sl. ~ Presented at the Workshop on Populations at Low Risk of Can- cer conducte~ at Snowbird. Utah. August 2~o25. 1978. : Department o~ Epideraiology and Statistics. Araerian Cancer ciety. Inc.. 777 Thi~ Ave.. New York..N.Y. 10017. 3 These states were Arizona. California, Florida. Georgia. Illinois. Indiana. Iowa. Kansas. gemuck~:. Louisiana. Mar.viand. Massa- chusetts. Michi.~Tan. Minnesota. Misstssippi. Missouri. New York. North Carolina. Ohio. Oregon. Penns3'hunLa. South Carolina. Ten- nes.~ee. Texas. and Virginia. 1169 JNCI. VOL. ~. NO. .5. NOVEMBER TI09371754

1170 Garfinkel site by the certifying physidan or hospital in 69%. Over the endre 12 years $2.25 of all cancers in nonsmokers were confirmed. The cases confirmed for each site of cancer ranged from 23% for lung cancer to 67% for lip cancer, the average site-specific confirmation rate being 37%. Coding of death certificates was according to the ICD (Seventh Revision) and was converted to com- parable categories of the ICD (Eighth Revision) code for this report. Ahhough coding was performed by various personnel during the course of the study, all coders were trained by the same persons. The coding was checked by the supervisor of clerks. II correctness was in doubt, the codes were referred to me for a final derision. "Classification o/nonsrnokers.mln this study, we use the term "nonsmokers" for persons who have never smoked regularly. It includes those who never smoked any kind of tobacco and those who wrote on the questionnaire that they smoke or have smoked an. occasional cigarette, cigar, or pipe at present or in the past. The occasional smokers comprised only about a.~sr a~ ~p and p~-~a~ of e~ o~ ~e ~ 1~ ~a~ f~ ~fic~ who ~ ~d ~rlv: ACS ~ai~ Male Female Age group No. of Per~n- No. of Person- in 1959 mzbjeets years of subjeet~ yoars of at start expo~re, at start exp~ure. 1960-72 1980-72 30-34 2,165 24,162 8,532 96,393 35-39 3~89 37,131 17,490 197,742 40-44 6,001 68,180 37,467 422,220 45-49 17,856 199,195 63,170 703,459 50-54 17,801 193,082 65,~,8 721,089 55-59 14,645 154,425 5~,607 622,188 60-~ 11,631 11&8~4 46,688 490,449 65-69 9,280 85,606 35,764 3~5,207 70-74 6,173 49,598 22,418 202,595 75-79 3,419 23,696 11,993 92,061 80-84 1,4~3 7,389 5,775 34,689 85-89 577 2,275 2,231 10,102 ~90 166 442 648 2,022 Total 94,436 961,975 375,381 3,950,186 one-half of I,% of the total who never smoked regularly (2). [The ~:act wording of the questions on smoking habits on the first que~tlonnaire is shown in (2).] Computation el person-years, expected number of deaths, and death rates.--A total of 456,480 males and 590,562 females more than 30 years old was enrolled in the study during a 6-month period beginning October 1959. Of these, 94,456 men and 375,~81 women were classified as having never smoked regularly. The num- bers of subjects and person-years over the 12-year period, classified by the 5-year age group at the start of the study, are shown in table 1. The numbers of person-years of experience at ~ach 5-year age group in the first and second 6*year periods were approximated as shown for one 5-year age group in table 2. Death rates for.cancers of different sites were then computed for all subjects by sex and cohort 5-year age group in each of the two periods. These rates were applied to the estimated number of person-years of exposure for nonsmokers at each attained 5-year age group to obtain expected numbers of deaths for each site. The weight- ings used to compute expected numbers for the first 6 years and second 6 years are shown in table 2. The observed numbers of deaths were divided by the ex- pected numbers and multiplied by 100 to obtain the SMR. The 95 and 99% confidence limits for the SMR, adapted from the method described in appendi~ C of the paper by Haenszel et al. (3), were used for observed numbers of deaths up to 250 in the determinations of upper- and lower-limit estimates o! the Poisson-dis- tributed variable. For observed numbers of deaths of 250 or more, we used a chi-square test to determine the 95 and 99% confidence limits. Death rates (per I million population) were com- puted for each of the S-year age groups and then standardized for 10-year age groups (35-44, 45-54, 55-64, 65-74,. and ~75) weighted according to the proportion of male and female nonsmokers in the 10- year age group. They were then rounded off to death rates per 100,000 population. To make the data more homogeneous and com- parable to the mortality data for some of the religious groups, calculations were made only for the white population in the study. Actually, this was not much TAI~rr 2.--Computation of tt~ights u#ed to alypr~mate attained pem~m.~ars of exposure to risk of nontmoktrs 80-8,~ ~ar~ old" Age. Average attained age in January of each year in the study July First 6 yr of study~ Second 6 yr of study~ 1959 1960 1961 1962 1963 1964 19~5 1966 1966 1967 1968 19~9 1970 1971 1972 30 30.5 31.5 32.5 33.5 ~ 35.5 36.5 "31 31.5 32.5 33.5 ~ 35.5 36.5 37.5 32 32.5 33.5 34.51 35.5 3&5 37.5 38.5 33 33.5 34.5 ~ 35.5 36.5 37.5 38.5 :~9-~ ~ 34.~ ~ s~.5 37.s 38.s 39.5 I 4o.~ 36.5 37.5 38.5 39.5 ~ 40.5 41.5 42.5 37.5 38.5 ~ 40.5 41.5 42.5 43.5 38.5 ~ 41.5 42.5 43.5 44.~ ~9 5 ~ 40.5 41.5 42.5 43.5 44.5 ~ ' 45~5 " 40.5 41.5 42.5 43.5 44.5 ~ 46.5 The death rate for all subjeetn w~ applied to the e~timated persan.year~ to obtain eXl~eted numher~ of deaths. 30-?Affi12.5 3T=0.417 ~ 35-39=17.0 .vr:=0.5,~6 w~ 40-44=0.5 ~T=0.017 wt. 20-34 (3B-40)=,8.0 3T:=0.287 wt: 35--39 (41-45)=20.0 yr=0.6~6 wt: 40.q4 (46-50)=2.0 yr=0.067 wt. jxct. VOL e~. NO. 5. NOV~.MBER ISSO T109371755

-V V different from rotes for th~ entire study ~:ecause non° whites comprised only 3"~ of the total. RESULTS Table ~ shows the observed and expected numbers of deaths due to cancer and the corresponding SMK for males and females who never smoked regularly. Males had an SMP,. of 64 for cancer of any site; females had an SMR of 98. The second figure is not surprising because female nonsmokers comprised 68% of the total female study population. The SMR were low (<$0) for men who had cancers of the tongue, mouth, pharynx (o~her than nasopharynx), larynx, lung, and pleura and moderately low (<60) for cancer~ of the esophagus and bladder. Other sites of cancer with low SMR in males included cancers of the rectum, liver, pancreas, and kidney, as well as other and unknown primary sites. The SMR observed for females was 42 for cancer of the larynx, 48 for the pharynx (other than nasopharynx), and 62 for lung cancer. For certain sites with large numbers of observed deaths, the SMR was statistically significant even when it was close to 100, For example, colon cancer (91) in men and total cancer (96) and cancer of the pancreas (92) in females were statistically significant. Several sites had SMK higher than 100 (retroperitoneum, connective dssue, and melanoma) in males; stomach, colon, gallbladder, and connective tissue had SMR higher, than 100 in females. However, none were statistically significant. Table 4 shows the age-standardized death rates for each site by sex for 10-year age groups in subjects who never smoked regularly. The death rates for females were based on larger numbers of observed deaths than for males; therefore, more age groups with any deaths were reported and fewer 10-year death rates were based on less than five deaths. Total cancer death rates for female nonsmokers exceeded the death rate for male nonsmokers in age groups of ~5-~14, 45°54, and 55-64 years and were the same for those 65-74 years old. The death rate for males age 75 years and older exceeded 'the comparable death rate in females by 40~. A large part of the excess total cancer rate in the younger females was attributable to the breast cancer death rate. Breast cancer was the leading cause of cancer deaths in women in all age groups except the oldest. Death rates for colon cancer were similar in males and females for all up through 74 years of age. In those 75 years and older, the female rate (140/100,000) was slightly higher than the male rate (126/100,000). Compared with the overall lung caner rate, which was roughly four times as great in men than women, rates in nonsmokers were only slightly higher for men than women. Lung cancer, indeed a minor site in male nonsmoke.rs, ranks from fourth m sixth in the 5 age groups shown in table 4. In females, it ranks from fifth to eleventh in the same groups. Caner Mcr~alltt in Nonamokera 117~ DISCUSSION A question arises concerning the degree to which mortality rates are affected when a selected population is used instead of the general population. When comparisons of mortality ratios are internal (e.g., smokers ~. nonsmokers; hypertensives vs. nonhyper- tensives) in a study population, the fact that the study population is not representative of the general popula- tion makes little difference in the interpretation of results. However, when comparisons are made with regisus' or vital statistics data, selection can make a difference. For this reason, SMR for nonsmokers in this analysis are based on total .study population cancer rates, rather than standardized national rates. The ACS study population differs from the general population in several ways: nonwhites, persons who never attended high school, for.eign-born, and institu- tional populations are all underrepresented. Also, per- sons who were sick or hospitalized during the enroll- ment period are underrepresemed, though some of these individuals were enrolled by a family member. Death rates were low in the first year of this study and rose in subsequent years, but they never reached the level of the total death rate in the United States. The death rate in the first 6 years of the study approximated that of the United States for the $5- to 39- year olds, was between 67 and 7~% of U.S. rates for the 5-year age groups from 40 to 60 years, and was 80-87% of the U.S. rates for those 70 years old and over (4). The SMg for the two 6-year periods were analyzed separately. In males, the total cancer SMI~ was 65 in the first period and 6~ in the second. For females the comparable SMR. were 98 and 94. Similar decreases, noted for the major sites of cancer, could be due to diminishing effects of underrepresentation of the sick and hospitalized, with the mortality differences be- tween the nonsmokers and the general population consequently tending to in'tease. However, the most probable reason for the particularly large drop in the SMR for lung cancer in women, from 71 in the first period to 60 in the second, is the sharp increase in cancer rates among smokers in the second period. As noted above, microscopic proof was received for 69% of the cancers in the first 6 years of the study. When a comparison was made of the primary site coded on the death certificate and the one dted on the subsequent physician's or hospital report, they agreed precisely in 92% of the cases. The percent agreement was fairly uniform for sites in all organ systems ranging from 91 to 95%, except [or cancers of the oral cavity and pharynx for which the agreement was only 62%. In many instances when the primaq, site was not exactly the same as that mentioned on the death certificate, it was specified as a site within the same organ system. REFERENCES (1) HA~.I.~O.~D EC. Smoking in relation to mortality and m~rb~dit¥. j.~c~, voL ~. xo. 5. NOVEMBEIg 19~ T[09371756

1172 PrL'~r~ site• Female O~ Exp SMR Obs Exp SMR Total malignant neoplasms 2,999 Lip, oral ~vit~. ~d pht~x ~p 1 Ton~ 2 Gum ~ ~u& N~p~x 7 ~hsr ~ uns~eifi~ p~x 3 Di~ o~ and ~neum 1,170 ~pha~ Swmaeh ~Ion 475 ~ ~i~oid j~c~on, and a~l 130 ~ver 45 Gallbl~der Bili~ ~gns 24 P~e~ ~pi~ sys~m ~x ~ng ~her ~im~ o~ 6 ~n~ ~d join~ Conn~, subeu~n~, and o~er ~f~ S~n, ~ltno~ 53 Fe~le ~ni~l sy~m Ce~x u~ Co~ Ut~ NOS~ V~ ~lv~ ~d o~er female ~ni~l ~m P~, ~ ~ o~er ~le ~ni~l s~m 3 U~ a~m Bl~d,r ~ o~er urinaw o~ 96 Kidney and re~l ~1~ Eye 6 ~e~o~ ~m 114 End~ne sy~m ~er end~ne glands ~mpho~ Lymph~oma ~d ~t~culum ~11 ~r- 146 H~gkin's d~ ~her lymphom~ including mycosis fungnid~ II Multiple ~elom~ 76 ~ukemia 217 ~her ~d unknown p~ma~ 171 4,676.5 64# 12,625 13,159.0 96s 99.3 29~ 91 lISA 80 4.8 21 2 1.5 113 23A 9~ 18 28.8 53c 19.0 92 31 30.4 102 17.8 28# 22 19.0 118 10.7 66 6 8.3 72 30.7 10# 12 ~5.1 48~ 1,481.7 79# 4,062 4,128.8 98 73.0 40# 53 859- 62# 260.8 86c 533 516.7 103 12.1 91 40 38.7 103 521.0 9I~ 1,907 1,877.8 102 175.7 74# 424 431.7 98 72.1 62# 183 193.8 94 19.9 85 150 146.7 102 27.7 87~ 69 78.8 88 300.3 64 649 706.7 92c 191.2 115 64 52.8 102 1,201.3 17~ $37 8~6.1 64# 44.3 11# 6 14.3 42~ 1,183.7 17~ 503 785.0 64~ 23.3 26# 28 36.8 76 16.1 93 41 40.6 I0I 17.4 109 61 58.2 105 44.6 119 114 llSA 99 3.09~ 3,090.7 100 1,839 1,893.3 97 308 3J~.2 87~ 224 233.9 9~ 242 241.6 995 999.8 100 70 72.8 96 576.6 94 566.9 96 6.1 3.6 83 302.2 62t 379 420.1 90c 183.9 52# 196 229.6 8Y 118.3 76# 183 190.4 96 5.8 103 16 14.7 109 131.5 87 292 303.8 96 18.0 61 70 74.0 95 10.9 92 66 64.1 101 7.2 14c 5 I0.0 50 193.8 98 558 5,59.5 100 138.4 105 417 409,8 102 42.7 77 105 112.7 93 12.7 87 36 37.1 97 74.7. 102 193 196.7 99 242.8 89 ~05 508.2 99 271.9 63# 771 806.6 96 • Diagnoee~ were classified by the ICD (Seventh Revision] and converted to # Value was significant at the 99% level. c Value was significant at the 95% level. # NOSffinot otherwise specified. the Eighth Revision. JNCI. rol_ 6~. NO. 5. NOVF-.31BER Tl09371757

Primary site Ferule° 35-44 45-~4 55-64 65-74 -->75 35-44 4~-54 55-64 65-74 -->75 yr yr yr yr yr yr yr yr yr yr Tcml malignant ne~plMms 51 64 1~4 440 1,025 69 146 243 441 774 Lip. ¢~ ~. ~d pha~x -- 1b 1 3 13 -- 1 1 3 9 Gum a~d ~ 1~ ~ <0.5 3 <0"5~ Small in~tine < 4 letli~ne~, lt~neum, ~d In- 2 3o Other ~Pi~l Skin, meh~om~ 5 ~ 2 2 5 13 ~ 3 Bre~t -- 28 56 76 ~ 125 Ferule ~ni~l C~ix u~d ~ 3 5 6 9 17 ~u~ u~ -- 1~ 1 5 10 12 U~ NOS~ -- ~ 2 4 10 14 ~ -- 6 16 24 ~ 41 ~er fe~le ~ni~l s~m -- <0.5~ 1 1 3 6 Male ~ni~l s~m 2~ ..... T.Us ...... U~n~ mm 73 <0.5~ 2 5 12 36 Bt~deP Kidney -- 3 7. 15 ~ <0.5+ I 4 5 14 Eye .... 1~ 1~ <0.5b <0.5 <0.5b 1 Ne~o~ ~m 6b T 12 1~ p) 3 5 10 11 2 L~ho~ 12 ~mph~o~ and ~¢~um ~II ~om~ 8~ 4 15 18 40 3 4 8 16 26 H~gkin'$ d~ 4t <0.5b ~ 6 ~ 2 2 3 4 3 Mul~ple m~lomg l 5 12 31 <0.5~ I 4 9 12 ~em~ 8& 6 II 31 85 6 5 6 18 40 O~¢r ~d ~kno~ 14 70 . I: 7 h?' " Dead, s=no dead,s in ace~ex b Observed deaths were <5 for a 10-yr age group. c NOSfn0t o~erwise specified. Finding~ in the first thirty-four months of follow-up in a prospective study ~tartcd in 1959. J Nail ~nc~ Inst I~: 3~1161-118~. ~o~o EC. G~n~EL L, S~king habits of men and w~ m~. J Natl ~nctr Inst 1~1:~:41~4~. (3) ]-L~ENSZEL ~,V. LOVEL'd~D DB, St~,E.x MG. Lung<ancer mortality ~ reh~ to residence and smoking historic. I. White mal~. J Natl ~ncer lnst 1~2~:~7-1~1. ~) K~M~O~D EC Life ex~n~" of Am~n ~l~ in relation to • ~iv smoking habits. ~ jNcI. VOL. ~. NO. 5. NOVEMBER Ib'30 TI0937175~

Cancer Mortality Among a Representative Sample of Nonsmokers in the United States During 1966-68' James E. Enstrom, Ph.D., = and Frank H. Godtey, Ph.D. ABSTRACT.-OIta ire presentad on cancer lnd total moHlllty ,,mong a representative samite of nonsmokers ¢nd the total population 35-84 years of Ige ~n the United Stltes during 1966-88 that measured the influence of cigarette smoking on mortality rites, independent of other he~lti'Preisted factors. Of U.S. whtta males, those who never smoked cigarettes hive a total Ige-sdjusted cancer dtsth rite which is 37% lisa than that of miles es a wholt and 53% less thin that of those who currently smoke cigarettes. Correspondingly, of all U.$. white fi, mlles, those who never smoked cigarettes have ,, totll ag~-adjusted cancer desth rite which II 15% less than femllel as i whole 33% less then that of those who currently smoke cigarettes. The largelt cancer rst~ reduction In the nonsmoksrs Is concentrated in the re~pirstory system. Nonsmokers hive In age-ld|usted total delth rate which is ~bout 20% less than the population as whole and lbout 43% lisa thMt current cigarette smokers. These Ind other results end methodologic I~suss are discussed.--JNCI 65: 1175-1183. 1980. Previous studies have established a strong, positive correlation between dgarette smoking and mortality, with cancer mortality showing some of the highest correlations. Beginning in 1964, the Smoking and Health Report of the Surgeon General has summarized the evidence (1). Only one previous study in the United States has been based on representative national samples and that was limited to lung cancer deaths during 1958-59 (2, 3). The findings here are based on a 1966-68 survey that measures among a representative U.S. sample the effect of smoking status on all causes of death inde- pendent of other health-related factors. The nonsmokers in this survey are meant to be used as a reference for comparisons with other no.nsmoking populations who may be subject to additional health-related influences and selection biases. Our focus is on cancer mortality among nonsmokers and their risks relative to those L-cperienced by the total population. "Nonsmokers" here are defined as persons who smoked fewer than 5 packs of cigarettes (100 cigarettes) in their lifetimes. However, they include former and current cigar and/or pipe smokers. Unlike most other low-risk populations, nonsmokers comprise a substantial proportion of the adult popula- tion. They are not limited to a particular geographic region, ethnic group, or religion. Moreover, our use of national probability samples to examine the mortality experience o[ nonsmokers minimizes selection biases with respect to other environmental, socioeconomic, and life-style [actors. MATERIALS AND METHODS Investigators who conducted most previous studies of cigarette smoking and mortality used a single-sample, prospective design. Mortality ratios were obtained from the relative proportions of the original sample who died within a specified follow-up period and then were wdghted by survival times. However, serious disadvan- tages associated with representativeness and problems of follow-up are evident in the prospective studies (l). Nonresponse may have been as much as 32~ overall and even greater among smokers. The bias in the mortality ratios of the respondents was estimated to be about a 0.2 or 0.3 overstatement because of lower mortality ratios of the nonrespondents. The low level of nonresponse in the present study (5~) indicated that nonresponse was negligible as a source of bias. Our approach was to use a cross-sectional, two- sample design to esdmate the relative risk of non- smokers' mortality. This study was analogous to the one of representative samples of 1968-59 lung cancer deaths among U.S. whites (2, 3). One sample was representative of U.S. deaths during 1966-68, and the other resembled the general U.S. population during the same period. The smoking characteristics of both samples were determined and then mortality rates were calculated as a funcdon of smoking status. Some limited results of the study, described in detail else- where,4 have been published (4-6). Sources o/data.mData were from two sources. Esti- mates of observed deaths of nonsmokers 35-84 years of age in the United States in 1966--68 were based on the NMS, a follow-back survey linked to a probability sample of 19,526 death registration records that in- Ass~.~vlA-nO.'Cs t'si~ CPS~-Currem Population Surve~(sg ICD-In- ternational C)assificadon of Diseases: NCHS:~National Center for Health Statistics; NMS~, National Mortaihy Survey; SMRa.standard- ized mortality ratio(s). ~ Presented at the Workshop on Populations at Low Risk of Can- cer cond,,,'ted at Snowbird. Utah. August 2.~-25, 1978. ~ School of Public Health and Jonsson Comprehensive Cancer Center. University of California. Los Angeles. Calif. ~ Division of Analysis. National Center for Health Statistics. East-West Highway. Hyattsville. Md. 20782. • Godley. FH. Cigarette smoking, soda| facton, and mortality: New estimates from representative national samples. Unpublished Ph.D. dissertation. University of 3Laryland. College Park. Md.. 1974. 1175 J.~cL VOL ~. NO. 5. XOVlLxffiEI~ lfg3 TI09371759

1176 Er~t~om and Godley eluded deaths of 11,318 white males and 5,638 females.~ The underlying cause of each death was assigned in accordance with the Seventh Revision of the ICD (7, 8). Questionnaires were mailed to surviving family mem- bers and others named on the death certificate who provided smoking histories and social characteristics the deceased. The second source of dam is the CPS conducted by the U.S. Bureau of the Census in August 1967 (9). Smoking and socioeconomic information comparable to that for the decedent was obtained by means oi household interviews for a probability sample of 60,920 adults 55°84 years of age, including 25,266 white males and 29,308 white females: The technical details of each survey are given in the Appendix. Also included there are the age-sex distributions of the deaths and population-at-risk by smoking status and a sample calculation of the weighted number of deaths when the sample number of deaths is used. Inferences that can be drawn from the data in this study are conditioned by certain features of its design, sampling, and dam-collection methods, as discussed in detail in the Appendix. In prospective studies, the decedents are a subset of the total study population; examples are the American Cancer Society and U.S. Veterans cohort studies in which a beginning sample is initially questioned regarding both current and past smoking habits (10, 11). The decedents of a prospec- tive study, who are in the original sample, match exactly with an individual in the study population. The cross-sectional design assumes statistical matching between the two independently drawn samples: de- cedents and population. This problem is common to conventional vital statistics rates and must be balanced with the expense, time, and follow-up problems of prospective studies. The value of a two-sample cross- sectional design was demonstrated in national studies of smoking and lung cancer mortality during 1958-59 (2, Delinition.r.mDetaJled estimates of cigarette smoking history by amount smoked and other relevant smoking variables were not made. Instead, persons were simply classified by lifetime history of having "never smoked" or "ever smoked" cigarettes (at least 5 packs or 100 cigarettes). "Current" cigarette smokers are persons who have smoked at least 100 cigarettes in their life- times and have smoked cigarettes during the past 12 months. We coordinated smoking questions in the supplement to the CPS with those in the death follow-up survey in an effort to gain comparability. Unlike sizable proportions of the CPS sample who provided data on their smoking habits and other personal characteristics, such data were not a~ilable for the NMS death sample. Moreover, a mailed ques- tionnaire was used in the NMS, whereas a household interview was used by the CPS for collection of data on National Center [or Health Statistics. 1966~8 National Mortality SureSt. Unpublished technical notes and computer tapes. 1976. JNcI. VOL ~5. NO. S. NOVE.MB~:K It80 the living population. Evidence from another survey indicated that agreement I:~tween the smoking history reported for a decedent by next of kin and by his own repor~ when alive was about 90%. Moreover, Haenszel et al. (2) noted that this type of bias is largest with _ respect to medium and heavy smokers but virtually nonexistent for nonsmokers. RESULTS Table 1 contains the average annual age-specific total death rates for 1966-68 for U.S. white males and females by smoking status: never smoked cigarettes, currently smoke cigarettes, and the total sample. The rates are given in 10-year intervals for adults 35-84 years old, with an overall rate age adjusted to the 1940 U.S. population. The 1940 .U.S. population is the standard reference used by the NCHS (7). Included in table 1 are age-specific total death, rates for all U.S. whites based on the NMS and CPS and the average total death rates as published in the 1966-68 and 1970 Vital Statistics of the United States (8). The NM$-CPS rates generally agree with the 1966-68 U.S. rates to within 3%, except in the 75- to 84-year-old category, in which they are up to 10~ higher than the U.S. rates. (The CPS includes only the noninsdtutlonalized popu- lation as explained in the Appendix.) Agreement be- tween the survey rates and the U~. vital statistics is good. Finally; it is evident that the nonsmokers in this survey have significantly lower rates than those of the total population, which conforms with previous epi- demiologic studies of nonsmokers (l, 10, 11). Non- smokers have an age-adjusted total death rate about 20% less than that of the population as a whole and about 45% less than the rate for current cigarette smokers. In addition to the 1966-68 average annual age- specific cancer death rates for U.S. white males and females by smoking status, table 2 includes the rates for total cancer and for cancers o[ the lung, breast, and prostate, all of which are based on more than 100 sample deaths. The U.S. white males who never smoked cigarettes have a total age-adjusted cancer death rate ~7% less than that of U.S. white males as a whole and 5~% less than that of white males who are currently dgarette smokers. With the rates for total cancer based on the NMS and CPS are those given in the 1966-68 and 1970 U.S. vital statistics. The agree- ment between the survey and the vital statistics rates is within 11~ for those less than 65 years old. Over 65 years, the survey rates are 7-17,~ higher than those for vital statistics. These differences are greater than the corresponding differences for total mortality primarily because of statistical fluctuation in the sampling of cancer deaths in the NMS. The 1966-68 weighted numbers of observed cancer deaths for U.S. whites who never smoked cigarettes for the sites listed based on at least 10 sample deaths in males and/or females are given in table 3. Also included arc the expected numbers of cancer deaths T109371760

C~nnr Amang U,S. Nansrnokem 1177 U.S. whims U.S. whites Al! U.S. whites who never who ,'ur- 1~]--~ I~ U.S. 1970 U.S. c~" ci~e~~ CPS• s~tfs~ smtis~ Males 35-44 216 442 352 342 344 ~ 570 1.183 937 904 ~ b~-64 1,686 ~.S~8 ~Z ~ ~ ~74 B.~ 6,572 4,9~ 4,~8 4.810 7~ 9.~ 13.9~ 10.749 9.918 I0.~ To~l: I.~ 9.489 1.918 1,862 I,~ Femal~ 3~4 1~. 245 ~)1 195 193 45-~4 333 597 464 462 468 M 771 1,444 1.016 1.022 1.015 ~-74 2,398 ~ 2,70~ ~6~ ~ 7,~ 18,~ 7,~ 6,8~ 6,6~ To~l: 8~ L~ 1,0~ 1,017 989 • Death rates are based on the 1966-~ NMS and August 1967 CPS. ~ Rates are age adjusted by the direct method to the 1940 U.$. population 35-84 yr old. based on the death rotes among all the whites in the 19~.-68 NMS sample. With th/s group as the reference population, the total cancer SMR for whites were 67 and 87,% for male and female nonsmokers, re- specfively. Using 1970 U.S. whites as the reference population, as was done in a previous publication summarizing some of these NMS results (.~), we de- termined, that the total cancer SMK were 70% for white TABLE 2~T~e I96&~8 az~raffe am~ual a~.~,ciflc, can~er d~ath ~,ates (d~ths/]O0,O00) among U.S. ~tbitts by ace. se~, and cigarttt~ Lun~ cancer Total cancer U.$. U.S. whit~ whites whites who U.$. whites U,S. whites All U.$. whites who who All U.~. never who never who cur- 1966-68 1966-68 U.$. 1970 never currently white" smoked smoked rently ~mok. e NMS + v/tal v/tai smok~i b smoke cigarettess eigarett~ cigarettes° CPS~ statistics stat/~ti~ cigarettes cigarettes~ 35-44 45-64 65-74 75-84 Total: 3~84 2.3 (2) 17.5 (38) 12.9 (48) 0.7 (1)" 32.0 (28) 56.8 (125) 49.2 (185) ~0 ,50 3.5 (3) 81.1 (163) 59.3 (213) 2.1 (3) 80.7 (42) 211.2 (340) 172.4 (476) 170 172 33.4 (24) 270.4 (365) 177.6 (493) 15.2 (18) 308.0 (102) 685.7 (646) 527.5 (946) 479 498 33.8 (43) 533.9 (298) 288.5 (499) 79.9 (89) 655.5 (204) 1,531.3 (543) 1,079.7 (1.041) 985 997 87.9 (36) 746.1 (122) 267,2 (211) 244.4 (165) 1,360.8 (294) 2,224.3 (262) 1,548.0 (760) 1.472 1,593 19.9 (108) 179.8 (986) 103.3 (I,464) 24.0 (276) 232.8 (670) 497.4 (1,916) 368.6 (3.408) 343.9 3,54.9 Females 35-44 0.4 (1) 8.6 (16) 4,2 (19) 23.8 (46) 56.7 (110) 80.1 (134) 72.9 ~(288) 65 62 45-54 2.8 (6) 25.4 (43) 15.3 (65) 45.6 (46) 133.2 (141) 197.2 (179) 172.1 (377) 175 177 M 10.9 (24) 51.9 (46) 25.4 (8~) 85.9 (45) 268.5 (166) 412.6 (141) 335.7 (360) 330 339 65-74 18.7 (37) 103.7 (31) 34.3 (82} 121.2 (28) 576.9 (182) 877.1 (66) 627.1 (274) ,565 5~5 75-84 48.4 (55) 143.9 (9) 5,5.5 (68) 145.0 (19) 921.8 (190) 1.303.9 (21) 999.2 (229} " 868 904 T©tal: 7.5 (123) 39.4 (145) 17.5 (319) ~gA (184) 220.9 (789) 328.8 (541) 260.6 (1.528) 244.5 246.2 ~ Unwei~hted No. of sample deaths are given f~ Death ram~ ~-e based cn 1M6-~8 NMS and August 1967 CPS. In column 5, values are for prostate cancer in males and breast cancer in females. ~ l~,s ar~ age ~djusted by the direct method to the 1940 U.$. population 35-84 yr j.~c~, vo- ~. No. 5. NOVEMBER I.~0 TI09371761

1178 ICD 9.525.416 male nc:~smokers 23.945.73,3 female Sample Sample Total malignant r~oplasms 149-20~ 106,478 6-/0 157.'/98 67 237~61 789 271,~ ~7 ~c~l ~ ~d pha~ 14~1~ ~7~ 56 4,403 62 1,702 28 3~48 52 ~ o~ns ~d ~ri~neum 15~159 45,~9 135 ~,457 93 74,~ 180 ~7,5~ 86 ~ ~pha~ 1~0 1,782 37 3~16 55 1,601 26 1,~2 ~ S~m~ 151 7.~0 17 10,~ 74 10,~3 27 13,~ 79 ~lon l~ 17.~T 44 13.~3 133 ~.~ 61 ~.446 94 ~m 1~ ~.~ 18 6.~3 117 8.117 18 10.573 - 77 ~ ~r, ~lbiadder. ~d biU~ 15~$6 3~82 6 5~2 62 11,717 ~ 13,311 ~ Pa~ 157 4,759 9 7,~0 60 11.6~ ~ 13.983 83 ~i~ ~m 1~!65 !3,038 115 46.~1 28 12.~8 131 21.7~1 59 L~g, b~nch~ ~d ~a~ea ~62-~ 8,7~ 108 39,~1 ~ 9,~9 1~ 17,1~ ~ ~ B~t ~70 ~,05~ ~84 ~,659 94 Ferule ~ni~l ~m 171-176 ~,~5 124 41.576 92 ~x 171 7,790 30 11.~2 67 ~ u~d 172 1.126 6 1,937 ~ Ut~, NOS~ ~7~174 5,~8 16 6,316 ~ ~ O~, fallopi~ ~ ~d 175 21,~3 68 19.393 111 b~ li~ment ~le genial s~m 177-179 13,784 278 16,~3 83 ~rina~ s~m 18~18~ 5.~ 13 9,~78 6~ 6,816 ~ 7.~6 86 Ne~ ~m 193 2.478 10 2.~9 95 6.~0 24 5.~ 123 ~hom, ~ 4.187 14 6.417 65 7.~1 21 9.131 88 ~emia ~ 3.~1 12 5~ ~ 8.784 24 10.107 87 To~ ~lt~ant n~ 14~205 106.478 670 1~.~7' 70 237~61 789 ~0.521' 95 ~ ! ;" ! • Sample No. of duths from which weighted No. of.observed deaths is ¢zl~lat~d with the use of ~s~t~fi~ ntio es~mztion p~u~ (~ ~g . " gx~c~ No. of d~ms f~m ~ncer is b~ on g~, ~iflc dea~ ra~ de~in~ for U.S. whi~ (t~8) from the ~1 ~ 1~ NMS a~ Au~st I~7 CPS (s~e ~xt). " SMR h the ~n~ defin~ ~ O~+Exp X 1~. ~ NOS=not o~e~ s~ifl~ ' Ex~ No. o~ deat~ f~m ~neer is b~ on ~, a~ific dea~ ~ for U.S. whi~ in 1970 (8). ~ male and 95% for female nonsmokers. For male non- smokers, the lowest SMg values occurred for the tradi- tional smoking-related cancer sites of the respiratory system, buccal cavity and pharynx, esophagus, pan- c~eas, and urinary system, with a combined SMR of 39%, whereas the remaining cancer sites had a com- bined SMK of 91%. However, the lowest SMK values for female nonsmokers' occurred for the cancer sites of the lung and buccal cavity and pharynx, with a combined SMK of 55,%; the SMK values for most of the other sites were between 80 and 95%. Because of the complex sampling and weighting procedures used in these data, confidence limits on an individual SMP, cannot be precisely calculated, although approximate limits can he determined with the use of the sample number of deaths, if one assumes they are subject to a Poisson distribution (2). DISCUSSION The 1966-68 NMS represents the first and only at- tempt to date to determine death rates [or all causes of death among a representative sample o[ the U.S. popula- tion classified according to smoking status. T~e basic findings in this paper show that during 1966-68, U,S. whites 35-84 years old who never smoked cigarettes have age-adjusted death rates from cancer and all'. causes substantially below those of the population as a whole and only about one-half the corresponding rates of those who currently smoke cigarettes. Phrased in other terms, whites who never smoked dgarettes have overall SMR which are substantially less than 100%. Specific comparisons vary. by a few percentage points with different reference populations or with direct versus indirect standardization, but the basic patterns remain the same. Fmthermore, these results indicate that most of the reduction in cancer deaths among male nonsmokers occurs among the traditional smok- ing-related cancer sites of the respiratory system, buccal cavity and pharynx, esophagus, pancreas, and urinary. ~stem.. The single greatest reduction in cancer deaths among female nonsmokers occurs in the respiratory system; however, small reductions also occur at many other sites. These results provide a methodologically sound dam set on which to base comparisons with other popula- JNCL VOL ~5. NO. 5. NOVE.MBEK TI09371762

- Cancer Among U.S. Nonsmoker~ 1179 tions at low risk to cancer. In other words, any nonsmoking l:opulation who has cteath rates similar to the 18f~=65 NMS nonsmokers is probably at low risk solely because they do not smoke. These results help delineate the specific influence of cigarette smoking on mortality rates from other health-related factors. REFERENCES (I) Advhow Committee to the S..r~-on Ger~ral of the PuMic Health Service. Smoking and h~l~. W~hin~on. D.C: U~. ~vt Print Off, 1~ [DHEW public,on No. (PHS)ll0~]. {2) ~N~L W, LO~ND DB, SIgN MG. Lung~n~ ~mlity - ~ ~h~ to ~sld~ and ~oking his~o~es. I. White J Nad ~c~ Inst O) ~NS~L W. TAEt'~R ~. LunB~nc~ morulity ~ relat~ r,i~ a~ ~oking hhtoH~. IL ~ite fe~l~. ~ Nad (#) ~ F. ~G~ D~ ~w :mo~ng and ~f{erem~l In: H~I~: Unit~ S~. I~5. R~k~lle. Md.: ~r f~ H~I~ S~. 1976:4~ [DHEW publi~tion No. (~)~I~2]. O) Es~oM ]E. ~r and ~I morality ~ong a~ive Mo~on~ (#) ~. ~sins lun~ ~ ~ity ~ong non~o~n. (7) G~ ~, H~L ~L Vi~l s~io ~t, in the Uni~ S~t,. I~I~. Washington. D.~: U.S. ~vt P~nt I~ [DHEW publi~d~ No. (8) National ~nt~ {or H~ Suti~d~ Yi~l s~tics of the Unit~ S~te~ ann~l r~m for I~ I%7. I~8. ~ 19~0. (~) ~. ~,~ ~ng s~tu~une I~. August I~. and Aunt l~8. Mon~ly ~I s~fis~cs re~rt. Vol 18. No. (I0) ~ONV E~ S~klng in reh~on m the d~th ~, of one mHli~ men and womb. Nail ~ Inn Mono~ I~;19: IT-~. (II) ~HN ~. The Dora st~ of smo~ng ~d mollify U.S. ~tmns: Re~rt ~ ei~h~ ~d on,half y~, of ob~r- ~ti~. Nail ~ncer Inst Mono~ I~;19:1-125. (12} U.S. Bu~u of ~e ~u~ ~ of ~e ~pu~don of Uni~ S~t~ ~ age. ~I~. ~d ~x. July I. I~7. C~em ~puhfi~ ~m. ~, P-~. No. 385. Washin~n. D.~: U.S. ~vt Prim Off. ([~) ~GXW~ EM. ~S~R PM. Dilf~n~l morality in ~e Uni~ S~: A study ~ ~~ ~demiol~. ~mbfid~: H~ U~v ~,. I~. (I~} F~N~L MR. Infe~nce ~ s~y ~pl,: An empifi~l in- v,fi~ti~. ~n Ar~. Mi~.: Institme for S~l ~rch. Univ ~chi~n. 19~I. (~) ~R~o~ H. ~ RH. Smoking ~bi~ o{ 21.612 i~ivld~Is in Te~ J Nail ~ [~t APPENDIX National Mortality Survey The NMS is based on a systematic random sample of all deaths o~curring and registered in the United States,5 Death registration is essentially 100% complete and is performed on a state-of-occurrence basis with the use of the official State death certificate. The States generally follow the federally recommended model death certifi- cate. Date, place, and cause of death, age, sex, and other demographic characteristics are important legal facts noted on the document. One can extend the research potentials of death certificates hy using them as a sampling frame for sur~-eys (such as the NMS) in which additional information is collected through informants, funeral directors, physicians, and others nam~t on the certificate. The scope of the NMS is all deaths of persons ~5-84 years old that occurred in the United States during 1966, 1967, or 1968. The sample was stratified for month of death, geographic area, age, and selected causes of death related to smoking. The total sample of 19,526 decedents between 35 and 8'~ years of age was drawn in two stages. 1) A I0.~ sample is represented in the monthly "Current Mortality Survey" of death certificates submitted to the NCHS on microfilm by all States, the District of Columbia, and other death registration areas. 2) The strata of the 10,% sample are subsampled; each 26th case was taken on the average. The overall sampling rate is therefore 1/260., The demographic and medical information coded from the sampled death certificate is supplemented by informa- tion from a mailed questionnaire on smoking habits, marital statu~, income, and other characteristics of the deceased. One-half of the respondents are spouses or ex-spouses of the deceased, one-fourth are either parents or offspring, one-eighth are sihlings, and the remain- ing 10,% are comprised of more distant relatives, friends, neighbors, and other associates. The overall response rate of 95% in the NM$ is comparable to that of the CPS. After follow-up queries of questionnaires with in- complete, inconsistent, or otherwise inadequate -~ponses, smoking information was still missing for about 8% of the sample deaths. For these cases, smok- ing history was assigned by computer processing of the records o~ the decedents. This procedure, in the absence of better information, assumes that the characteristics for which reported information is missing are like those of demographically similar individuals for whom the information in question was reported. This processing was performed by the NCHS, which also supplied the magnetic tape of 1966~8 records. Each record combines the smoking and other NMS questionnaire information with that from the death certificate. In addition, each record contains an inflation factor (sample weight) that reflects the samp- ling fraction applicable to the decendent's demographic cause-of-death stratum. This sample weight is de- termined by a poststratified ratio estimation procedure which forces the total number of deaths estimated from sampling to agree with the total number of deaths from which the sampling was done? The weight is a function of age, year of death, and cause of death and varies from about 10 to 6-t0. Appendix table 1 shows an example of how the total weighted number of 55,051 deaths is calculated from the 184 sample deaths due to breast cancer among U.S. white females who never smoked cigarettes. The age-sex distributions of the total weighted deaths from all causes according to smoking status are given in Appendix table 2. The underlying cause of death (item ~18 of the death JNCL YOL. ~. NO. 5. NOVEMBER I~0 T109371763

1180 Ermtrom and Godley U'~i~ females Total Average Unwefzhted No. cf s~nple Pc~mtracffied ratio e~- ~i~h~ snn~ ~nal by ~ ~mple at~sk (du~ cf 19~ U.S. :~r-44 18 15 13 46 81 82 84 3,7~0 5,301,563 23.8 0.346440 45-~ 17 12 17 48 163 162 1~6 7,~37 5,514,745 45.6 0.2922S3 ~ 17 I0 18 46 318 319 814 14,163 5,498,482 86.9 0.199205 68-74 II 7 I0 28 639 640 611 17,619 4,847,416 121.2 0.120142 75-.84 5 7 7 19 689 640 611 11,952 2,785,616 143.0 0.042~80 Total: 68 51 6~ 184 55,051 23,945.722 59.4s 1.000000 • Welzht No. a~ s~-nple deaths in each lO-yr age group is obtained by multiplication: T.Tnweighted No. of sample deaths in each death year times the pest~tretified ratio estimation sample weight for the same death year. add results for the 3-yr totals • "~ Age-adjuSted death rate is ¢~Iculated by the direct method with the 1940 U.S. population 3,5-84 yr of ~ge aa the standard. certificate) is determined from information provided by physicians and coded at the NCHS according to the rules of the Seventh Revision of the ICD (7, 8). Current Population Survey The CPS (9) is the source of estimates of the non- smoking and smoking populations. Smoking informa- tion was collected by the U.S. Bureau of the Census in iu CPS of August 1967. Designed to be a representa- tive sample of the Nation's noninstitutionalized popu- lation, the total consisted of about 50,000 households (residents of military bases, homes for the sick and aged, and other institutions were excluded). The over- all response rate was 95%; of the 95,532 aduh~ in the sample, those between the ages of ~5 and 8d years numbered 60,920 and included 25,266 white males and 29,308 white females..Based on this sample, the age-sex distributions of the noninstitutionalized population ac- cording to smoking status are shown in Appendix table 8. Also shown for comparison is the CPS for July 1, 196-7 (12), which represents the total resident popu- lation. The U.S. Bureau of the Census interviewers recorded information about homehold and personal characteris- tics of those in the sample. Demographic and sodo- economic information for all persons in the household was reported by a single adult respondent. Smoking information was obtained from each adult responding for himself, if this was possible during the one interview. However, the smoking information [or those not at home was also obtained from the person inter- viewed for the household. Inasmuch as men are more likely than women to be away from home when the interviews take place, only about one-third of the men, as compared with three-fourths of the women in the sample, reported on their smoking habits. Smoking in[ormation was missing for only about 3% of the sample. Estimates of population frequencies o| nonsmokers by demographic characteristics were obtained from A~PgNOtX T~L~ 2.--.Estimated I96~..~8 U.S. total d~aths o.f whites ba~¢d on repre~ntatice sample of deat~ from tl~ NM8 by a~, s~, aml ciouret~ smoking ~tatu~ U.S. whites who never smoked U.S. whites who currently smoke Age, yr cigarette" cigarettes" All U.S. whi~" Male Female Male Ferule M~e Fem~e 35-39 6,700 (81) 7,840 (98) 28.181 (346) 12,067 (148) 88.858 (476) £2.600 (2/9) 40-44 8,521 (105) 16.2~ (200) 49.3~9 (603) 21,116 (262) 66.804 (824} ~ (521) 45-49 13.734 (93) 20,697 (132) 74.468 (528) 33.045 (221) 103,119 (732) 60,191 (398) 50-54 25,868 (170) 34,468 (222) 118.466 (8,50) 39,771 (£/3) 169.694 (1,207) 84,566 (566) 55-59 39,723 (172) 47,388 (179) 146.149 (774) 48,336 (206) £35,209 (1.193) 110,514 (448) 60-64 58,025 (265) 79,733 (309) 168,468 (909) 46.68~ (209) £93,031 (1.527) 141.862 (589) 65-69 90,047 (2/3) 131,184 (304) 133,307 (691) 48.520 (139) 32/,646 (1,351) 195,314 (488) 70-74 109,&~6 (397) 217,599 (515) 144,226 (656) 41,403 (113) 363.764 (1,547) 280,660 (698) 75-'~3 158,£26 (498) 290,082 (708) 113.174 (482) 39.244 (96) 3'~,164 (1,426) 349,985 (862) 80"-~4 157.309 (478) 299.569 (691) 70.960 (2/0) 20,659 (53) 309.449 (1,035) 3~.428 (787) Torah 667,~89 (2,532) 1.143.840 (3,338) 1,0~6.748 (6.114) 330.849 (1.720} 2,23¢928 (11,318) 1.625.460 (5.636) 35-84 • Tl'.efir~t No. i~ eac£ cslur~ is the weighted No. of deaths with the u~e of weight based on the~pcststratified ratio estimation l~rocedure? ~ur~rs ~ ~r~tb.~,s ace ~e unweighted No. of sample d~ths used in the JNCL VOL f,5. NO. 5. NOVEMI~EK 198~ TI09371764

Cancsr Among U.S. Nonsmokers 1181 U.S. whites who never U.S. whi~es wh~ currently All U.S. whi~ ~ ~ ~le M~e Fe~le Fe~e ~e" Fete" ~le" Fe~e~ (Aunt (~F ~. (Au~ (J~y ~, l~f I~" 19~7)" I~7)" 35-39 1,168,42.1 2,527,M2 2,779.025 2,20,5.703 4fl79.120 5.~)2.000 5.1373,$6 40-44 1,185,1§1 2,774.021 ~,054,389 2.307.518 5~.35.199 ~,~78.000 5.~88,969 $,627.(~0 45-49 1,19~°62 2.788.674 2.894.887 2~48.005 5,054.911 5.161,000 5,434,324 §,4~4,000 ~0-~4 1,119,008 2,72~,071 2,541,111 1,819.272 4,547,741 4,723,0~ 4,968.615 55-59 1,009,872 2,778.288 2,084,528 I,~0,491 4,095,280 4,153.000 4,454.~]02 4,489,000 60-64 923,158 2,718,194 1,559.549 8~3,~03 3,392,068 3,452,0~) 3,935,809 3,875,000 65-69 937,471 2,5&1~26 9~,542 484,001 2,563,209 2,714,0~0 3~21.374 70-74 883,178 2~283,190 552,483 241.868 2o009,687 2,067.000 2,550.990 2.726.000 75-?9 695.297 1,834,020 326~79 107~78 1,439.539 1,462,000 2,002,333 2,038,000 80-84 408,598 951,496 114,336 44,8~0 689,562 800.000 1.012.735 1,195.000 Total: 9,525,416 2~,9~5,'~22 16,873,1~9 11,652,521 34:006,326 34.912,000 ~8,304~-~)7 38,8~6,000 35-84 • Numbers are from the August 1967 CPS of nouin~titutionalized U.S. population by cigarette smoking status (9)." Numbers are from the July 1, 1967 CP$ of total r~ident U.$. population (I2). inflation factors supplied by the U~S. Bureau of the Census in a magnetic tape of the sample person records. The inflation factors take into account the sampling fraction and demographic characteristics of those in noninter~iewed sample households. Missing demographic characteristics of those in interviewed households were computer assigned except for smoking information. "'Not stated" or otherwise unknown smok- ing status is distributed statistically in proportion to the relative frequencies in the known smoking cate- gories. This procedure is followed, for lack of evidence of a smoking-related pattern of nonresponse, so the best estimate of population can be provided. Because it incorporates U.S. census control totals, the CPS is representative of geographic units, age, sex, and other demographic characteristics of the national population. Standard Death Rates Expected numbers of deaths in table 3 are the product of a set of standard death rates for 1966-68 (Appendix table 2) and the August 1967 CPS estimates of whites who never smoked cigarettes (Appendix table 5). The standard rates reflect the mortality experience of the total white population (irrespective of smoking history). These rates are specific for 5-year age groups. sex, and the underlying causes of cancer death shown in table $. We computed the standard rates by dividing the total of the NMS death estimates for 1966, 1967, and 1968 by the CPS population estimate for August 1967. Thus both expected and observed deaths were derived from the same data base of deaths. In other words, for each cause of death, the total number of expected deaths equals the total number of observed deaths for all white males and females. This procedure minimizes some biases that otherwise may be present in the SMK when the expected deaths are based on an arbitrary standard such as the 1970 U.S. white popu- lation. Sources of Error Various types of systen'mtic erro~ likely to affect the estimates of mortality ~rom smoking are discussed below. Also, the reliability of the estimated statistics in relation to sampling error and other sources of random variability is evaluated. 1) Matching between deaths and populations: The major source of bias affecting the results of this study arises from the cross-sectional two-sample design. The- oretically, the decedents are a subset of the total study population; ideally, the prospective design best ap- proximates this concept. The American Cancer Society and U.S, Veterans cohorts (10, II) are prospective studies in which a beginning sample is initially questioned regarding current and past smoking habits. The cohort is followed longitudinally, and its decedents are recorded with respect m length of time in the study. Death rates are computed as follows: the total number of deaths divided by the total number of person-years lived during the study period by persons in various subgroups. These persons are classified by smoking habits and other characteristics. The decedents of a prospective study, included in the original samplc,~ match exactly with an individual in the study popu- lation. The present cross-sectional study, however, must rely on statistical matching between the two independently drawn samples: decedents and population. This prob- lem is common to conventional vital statistics rates (7, 8) and must be balanced with the expense, time, and follow-up problems of prospective studies. Because mortality risk is measured by ratio of deaths to population, the matching problem is a basic con-" sideration in the evaluation of errors from sampling, the dam-collection process, and measurement techniques. 2) The samples: A comparability problem exists in the extent to which the target populations differen- tially represent their universe. The universe of all JNCL VOL. ~S. NO. 5. NOV$,MSER I~ TI09371765

1182 Enstrom aftd God]~y deaths occurring to persons 35-84 years old in the United States during 1966-68 is represented by official registrations of deaths occurring in the United States in any of the 3 years, irrespective of month of death or usual residence of the deceawd. However, the popula- tion interviewed in the CPS consists of individuals who were not residents of institutions in the United States as of August 1967; this count was at about the midpoint and is believed to repre~nt the 1966-68 population adequately. The CPS pracdce of replacing part of the sample every 4 months (rather than monthly) extends part., of the target population to cover a span of 7 months, May to November 1967. Conversely. exclusion of resident, of institutions introduces a systematic bias that understates the popu- lation and thereby inflates death rate,. A first con- sideration is also to exclude deaths of resident, of insti- tutions. The rationale for not doing ~o is as follows: a) In terms of death rates, the bias is really a function ¢ff tx'r,on-year~ of excluded institutional population. Fragmentary evidence suggest, that many per~om in the institutional population are there for only a fracdon o! a year.4 b) There is no sound basis for the estimation o[ the person-years of excluded institutionalized popula- tion for the various ~ocioeconomic groups classified by smoking status, c) The net effects of the bias are likely to be canceled in measure~ of differential mortality like the SMR. d) The bin, is most likely to affect only the extremely old, a small fraction of the study population for whom differential mortality is greatly reduced and not of major interest. Only for persons 80-8'i year, of age is the excluded population more than 5% of the total (Appendix table 3). $) The respondent,: Unlike sizable proportions of the population ~ample who could provide information on their smoking habit~ and other personal characteris- tic2, no such data were available from the death ~ample. Moreover, a mailed questionnaire was used in the death survey, whereas a household interview wa~ conducted for collection of data on the living popula- tion. Other sources report substantial agreement be- tween the smoking habit, reported for a decedent retrospectively and his reports as obtained in a house- hold interview prior to death (2).* With regard to social and demographic characteristic*, evidence indi- cates agreement between the 1960 NMS responses and the matched 1960 censu~ report, (13). The overall response rates to the NMS and the CPS were nearly equal at 95%; these high levels eliminate any seriou~ bia~ from differential nonresponse. There is no evidence on nonrespome differentials by smok- ing status from this investigation. In retrospective studies, in which nonresponse may be as much as 3£% o~rall and even more among smokers, the bias in the mortality ratio of the xxspondent has been estimated to be about a 0.2 or 0.~ overstatement because of lower mortaliw ratios 0[ the nonrespondent* (1). Again, the low level of nonre~ponse in the present study (5%) in- dicates that nonresponse is negligible as a source of bias. JNCL VOL ~.5. NO. 5. NOVEMBEP. l_C$:~ 4) Measurement of cigarette smoMng:. The opera- tional definitions of lifedme history concepts, of never ~mo'~ng cigarettes are almost exacdy the same in the death survey (<5 packs of cigarettes) and popula- tion sur~.ey (<i09 cigarettes). Copies of the actual NMS and CPS questionnaires are shown elsewheref Although the slight variadom in these definitions do not appear to be a problem, a comparability problem doe, ari, e from the fact that those reporting for the decedent tend to understate cigarette consumption com- pared with what the decedent would have reported as his habit~. Haenszel et al. (2) noted that this bias is largest with respect to medium and heavy smokers but virtually nonexistent for nonsmokers. 5) Sampling and data processing: Complex sampling design, coding of reported information, construction of data files, and production of the final data tables are all possible sources of systematic error. The assump- tion is that these errors are negligible. Table 1 and Appendix table 3 show that the study estimates of death rates and population-at-risk.are representative of the United State~ for the study period. Table 1 illm- trates the good agreement between the total age-speciflc death rates as calculated in the survey and the 1966-68 U.S. vital statistics (8). The good agreement between the total age-specific U.S. white popuIation as deter- mined by the August 1967 and the July 1, 1967, CPS (I2), which was used to calculate the 1967 U.S. death rates (8), is depicted in Appendix table 3, 6) Reliability: In addition to systematic error, the estimates reported in this study contain some random error. This kind of variability is i'nherent in the sampling of deaths and population at a particular time. In more detailed analyses, random variation may be present in the indicators of mortality that are used for empirical tests of hypotheses. The 1964 Surgeon General's Report evaluated the stability of the SMR by a main assumption that deaths occurring within a given period are generated by a Poisson process (1). The random variability of SMR was small and de- pended on the number o! deaths. The analysis was of 26,000 deaths in 7 p.rospecdve studies. Inasmuch as this study is based on about 90,000 sample deaths, one can assume that variability in its SMK is a negligible problem for the major causes of death. 7) Sampling error:. In addition to variability over time, the SMR may contain some random variation due to sampling. Both death and population estimates are based on complex, multistage sample designs. Another analysis of the 1967 CPS shows that simple random sampling standard errors understate the true standard errors of estimated populadon statistic* Standard tests of statistical significance are therefore inapplicable to this study. Hammond (10) used a frequency of 10 deaths as a minimum for standards of reliability, Although arbi- trary, the same rule of thumb (adopted in this study) primarily affects deaths when classified by detailed underlying causes. The sample numbers of deaths from the 1966--68 NMS are pre~enred in tables I-3. TI09371766

8) Response error. At the level of the indi~qdual, reported information may yap." from one observation to another. This s-ariation ma.v introduce error in esti- mates of fixed characteristics like a history of ever smoking. Kirchoff and Rigdon (15) compared the smoking habits o! a group of Texans and reported Csncer Among U,S. Nonsmokem 1183 different smoking habits for the same reference per/ad when reslSondents were reinte~'iewed. This difference pardy reflects some memory lapse and some response error in smoking information. We assume that the extent to which this sourc~ of variation affects esti- mates of the present study is negligible but unknown. jNcl. VOL c~. xO. ~. NOVEMBEg I.~0 T109371767

Summary and Comments on t_he Workshop Moderator:. Joseph F. Fraumeni, Jr. TI0937176~

- High-Rate Cancers Among Low-Risk Populations,., Joseph F. Fraumen], Jr., M.D. 3 AEISTRACT~A r~Iaw of the tsbu~lr material at the Workshop on reduced for all forms of cancer. Thl= report I~rlefiy discusses =one high-rate tumors occurring among American ethnic and rallglou= groups who are traditionaJly thought to have = low dsk of cancer overall. In some groups expedendng a cttsnglng life- style, the rites for ¢~rtain cancers =re dsing or faltlng toward the levels prevailing In the general U.S. population. Ely eluold=tlng the high-risk end the low-risk states for vadous ¢tnoera In these population groups, one mlxlmlzes the opportunltle~ for research In cancer etiology and preventlon.--JNCI 65: 1187o1189, lg80. This Workshop focused on population groups within the United States who, taken as a whole, are thought to show a low risk of cancer compared with the risk prevailing in the U.S. white population. Blacks were not included because their overall incidence rates exceed those for whites, even though for certain forms of cancer (e.g., skin, testis, and bone) they show interestingly low rate~. Inasmuch as population groups with low rates of one type o[ cancer may have high rates of another, it would seem unwise to restrict attention to the low-rate tumors. Consideration of cancers that appear only a| sites w/th low rates without mention of sites with high rates in the same popula- tions could give the reader the simplistic impression that the genetics, environments, or life-styles of these populations are conducive to a uniform reduction in cancer incidence. Discussion of sites with high rates serves to remind us that, when we speak of "cancer at any site," we are referring ~o a conglomeration of different diseases with great variations in form and etiology. This is not to say that popuiation groups at low risk for certain tumors experience a compensatory high risk for other tumors. Litde evidence exists to support Cramer's hypothesis (I) that a relatively constant pro- portion of the population around the world is at risk of comracting some type of cancer. Indeed, it often happens that tumors developing excessively in an otherwise low-risk population still represent only a fraction of all cancers in the group. A certain urgency exists for investigators to study high-rate tumors in many of these populations among whom the life-style is changing, because over time the risks may approach those in the general population. In this brief review, our subgroup has not attempted an all-embracing survey of high-rate tumors and etio- logic hypotheses, but we have tried to single out some tumors and population groups that seemed especially interesting on the basis of the tabular material available to us. Cancers oj the mouth ond ph=rynx.--Incidence rates for these tumors are high in Puerto Rico and have l~en related m the consumption o£ alcoholic beverages including homemade rum. Cancer o~ the lip is reported to be equally high in Mormons and non-Mormons in Utah. presumably due to sunlight exposure, which suggests that the low rates for many tumors in Mor- mons compared with non-Mormons cannot be ex- plained by underreporting. Elevated rates for salivary gland tumors are recorded in Alaskan and Canadian Eskimos and apparently also in certain ethnic groups of Hawaii. Exceptionally high rams for nasopharyngeal cancer have been documented among Chinese through- out the world, although the incidence is lower in those born in the United States. Higher risks of this cancer have also been described in Filipinos and other Asian groups with Chinese admixture and in Eskimos from Aiaska, Canada, and Greenland. The susceptibility of Chinese to nasopharyngeal cancer may result from an interaction between genetic factors linked to certain histocompatibility antigens and environmental expo- sures including the consumption of salted fish or infection with Epstein-Barr v/rus. C=~ce~ oJ =he ~soph=g~.~In Puerto Rico. high incidences of esophageal and oropharyngeal cancers have been related to alcohol consumption and tobacco smoking. High rates of esophageal cancer also prevail among Chinese males, but they appear to be declining; that these rates are especially high in Chinese in northern China and also in chickens (which share certain dietary items) in the same region suggests a novel source of clues to its etiology. Stomach cancer.~The incidence is high not only in Japanese m/grating to the United States but also in Chinese Americans, Hawa/ians, Mexican Americans, Puerto Ricans, and American Indians. These excesses apply to the intestinal rather than to the diffuse type of gastric carcinoma. In case-control studies of Japanese in Hawaii, the tumor has been related to both a deficiency of various Western vegetables and an excess of pickled vegetables and dried or salted fish in the diet. Although it is unlikely that specific food items are responsible in all high-risk groups, further dietary studies are requ/red. High-risk groups also provide ' Pre~nted at the Workshop on Populations at Low Risk of Can- car conducted at Snowbird, Utah. Augusl 9.~o25. 1978. : This paper i~ ~ on d/scu~ions ol a subgroup comiuing of Dr. Fraumeni (Chairman). Dr. Leo Kinlen (National C.ar~'er Insd- sure), Dr. Anne Lanier (Center for Disea~ Contm|. Anchorage, Al=cka). and Dr. John McCullough (Urdversity o( Utah, Salt Lake City. Ulah). ~ Environmental Epidemio~ogy Branch. Divhlon o| Cancer Cause and Prevention. National Cancer Ir~titute0 Na6onal Insthute~ o| H~ahh, Public Health Service. U.S. Depanr.'tent ef Health and Human Service=. Betheu~. Md. ~205. 1187 j.~c~. VOL ~3..~o. 5. NOVEMBER I~0 T[09371769

1188 Fraumen! opportunities for the investigation of precursor lesions, such as chronic atrophic gastritis and intestinal meta- plasia, and mechanisms by which dietary factors may promote or inhibit carcinogenesis. The suggestion of a high risk in certain Hutterire families indicates that generic factors should not be neglected. Cancer of the large boweL--Among th'e populations considered at this meeting, there is a notable absence of any with a higher risk of this tumor than that of the general U.S. population, though the rates tend to rise in certain migrant groups, notably those from China and Japan. These trends, which also affect precursor lesions such as polyps, may resuh from the adoption of traditional American diets high in fat and meat and low in fiber content. Of some interest is the observa- tion from the New Mexico registry of an excess risk of rectal carcinoid tumors in American Indian females, although the numbers involved are small. Liver cancer.mElevated rates are recorded in several ethnic groups including Chinese and Japanese (espe- dally the foreign born), Mexican Americans, Puerto Ricans, Indians, Eskimos, Hawaiians, and Filipinos. Many of these groups, who alga show high rates of hepadds B infection and cirrhosis, provide oppor- tunities for research on these diseases as precursor states to liver cancer. Cancer o! the biliary tract.mHigh rates are well documented among American Indians and to a lesser extent among Mexican Americans and Eskimos. In some but not all reports, high rates are recorded for Japanese and Chinese Americans. The incidence of bilia~ tumors in population groups everywhere ap- pears to parallel that of gallstones, which, in turn, may be influenced by dietary, endocrine, and familial fac- tors. Generic mechanisms are probably partially re- sponsible for the exceptional risks of gallstones and biliary cancer in Indian tribes across the country and in population groups with Indian admixture. Cancer of the #anereas.mNadve Hawaiians and the New Zealand Maoris, another Polynesian group, share the highest rates of cancer of the pancreas reported around the world. The rates for females, but apparently not for males, are elevated among American Indians. The population groups prone to this tumor also seem predisposed to diabetes, which appears in certain case-control and cohort studies to be a risk factor for cancer of the pancreas, especially in women. Respirator~ tract cancer.~For cancer of the nasal cavity and sinuses, a high incidence rate is suggested in American Indians, ahhough this estimate is based on small numbers. For laryngeal cancer, high rates are reported for Spanish-American females in several pans of the country (Los Angeles, Calif.; Texas; and Puerto Rico). Ahhough in the past the rates for lung cancer were higher in Mexican-American women (mainly immigrants), they appear to have declined over the years and are not reflected in the recent statistics reported at this meeting. However, the risk of lung cancer continues to be high in Chinese Americans, especially females, and is consistent with the high rates reported in 5ingapore'zv indicate that particular t noma, are implicated cancer in Chinese. N also seem to have hi~7. Cancers of the bone a high rate of hone Americans from Los A; cation o[ the cell type'-~ tumors would be help: sarcoma is also indic Americans in Los An Breast cancer.~There risk of breast cancer am~ at this Workshop, but some groups (particular.._ Japanese Americans), th approaching that of t~ Dietary factors are sus[ this increase. Cancers ol the reproduc patterns of rates repor'" uteri, ovary, and prost pattern of rates reported.~- cervical cancer are recor American Indians, and - opportunities for etiolo~ role of herpesvixus type rr" high risks for cancers of • may share etiologic fact risk of penile cancer is especially in Puerto RicbT. tion with cervical cancer high rates among Amet testicular cancer is not r certain other Asian womb" noma is well documented, rates [or these groups a whites. Cancers of the urinal. kidney cancer are seen in mos and Indians. Howe groups covered at this to bladder cancer. Gancer of the nervous considered here,, none ap! of brain tumors. Pineal ently occur more often amt and the United States, population. The reported some Latin American data on eye cancer among Cancer o/ the thyroid reported in Mexican Filipinos, and Chinese incidence of thyroid cancer ethnic groups of Hawaii. Cancers of the lyrnpha terns.mAn excess of chiIo~.. Puerto Rico and elsewhere j.~ct. VOL eZ, NO. 5. NOVE.MBEK IS~3 TI09371770

reflected in the d~m reported in this ~;'orkshop. I~ut it would have been most interesting if breakdowns by a~e and cell types v-ere provided for lymphomas among the Spanish-American groups. Although the rstc for mul- t/plc myclorna in Indians is high, the numbers arc small. An increased risk of leukemia is reported in Huttcrites and has been related to inbreeding, but, again, the numbers involved are small and further work is required to evaluate th/s question. Conclusions.--For some time, it has been known that the risk of var/ous cancers may differ strikingly from one continent to another. Indeed, these variations have served as the basis for a rough estimation of the burden of environmentally induced cancer in our society. Compared with international patterns, dif- ferences in cancer risk within any country seem much less pronounced. Nevertheless, geographic variations in the United States, such as those revealed by the National Cancer Institute's county-by-county survey, High-Rate Cancers 11S9 and ethnic and religious differences, su~x as those highlighted in this Workshop, are pro~,~ding useful clues to the c~nses "of cancer. Population groups, however defined, s;enerally show a mixture of high and low risks for various forms of cancer. Although this meet/rig was primarily directed to the study of populations at low risk of cancer, the dam have also served to clarify our knowledge of people at high risk of developing cancer at certain sites. The nonmndom distribution of risk factors, reflected in the patterns presented here. suggests com- plementary strategies for etiologic research and preven- tion, with applicadorm beyond the population groups under study. REFERENCE (I) CRAM£1t W. The p~evention of cancer. Lancet 1934:1:!-5. JNCL VOL. ~5. NO. 5. NOV'EMBER 19~3 T109371771

Lip. oral cavity, and F,~ar~rxx Lip 140.0-140.9 Tcn~ 141.~141.9 ~liz~ g~s 1~14~9 Gum ~ ~u~ 14&~145.9 N~ha~x I~7 O~er uns~Y~ p~x 1~.~!~.9, 148.~ Di~ve c~ns a~ ~Hwne~ S~ IM.~151.9 Small immune I~l~Z9 ~lon 15].~15~.9 ~. r~t~oid j~ction, and 1fi4.~154.2 a~l ~1 Li~r 155.~ 19Z8 Gallb~ 1~.0 O~ bil~ ~mg~ 155.1. P~z 15Z~ISZ9 Re~H~e~. ~6tone~, ~ 158.~15&9. 159, ~s~fi~ di~sti~ o~ns 195.0 R~pimto~ ~x 161.~1~.9 ~ng 1~!6~1 ~og~c ~i2 ~ a~ joinn 170.~I70.9 ~w, su~u~n~uz, a~ ~ ~ft 171.~171.9 Skin, ~ma 1~17~9 B~lt 174 ~ in dm ~ u~ I~.0 Ute~ NOS" 1~9 Va~ 184.0 Vul~ and cllm~s 184.1 O~er f~le geni~l sDtem 181. IM.~IM.9 • Male S'rnital system Pr~ta~ gland T~s Pmis 187 ~ot~ ~ eth~ ~e genital sys~ 17&5, 18Z8, 187.9 B~d~ and o~ ~na~ ~n~ I~ I~-189.9 ~dn~ and re~i ~lvis 189.0, I~.i O~ ~n~ or~ I~-!8~.8 Eye N~ sys~ . O~er ~ s~em 192.~1~9 End~rine ~id gland 19~ O~er ~ne g~n~ 1~.~i~.9 L~phcm~ Lymphom~a and ~ cell ~.~.1 O~ l~ph~ i~l~ing myco~ ~.~2~.9 [~goides Mul~ple m~loma Lym~adc ~.1 Oth~ O~er leukemia ¢e.S.. ~n~y~ or ~5.9. ~.9, ~.9. ~.2 19&1=1~.9. 1~.~ !~.~, I~.~ !~.7, 1~.9, 1~.~19&9, 1~.~i~.1, Other and unknown primary • NOSz~not ou~erwise specified, 1191 JNCL VOI, ~, NO. 5. NOVEM~EP. I.~0 T109371772

PARTICIPANTS Mr. W. Lawrence Beeu3n Department of Biosmtistics and Epidemiology I.oma Linda University Loma Linda. Calif. 92550 Thomas R. Bender, M.D. Alaska Investigations Division Bureau o! Epidcmiology Center for Disease Control Anchorage, Alaska 99501 William J. Blot, Ph.D. Envimnmentai l~pidemiology Branch Division ~ Cancer Cause and Prevent/on National Cancer Institute National Institutes of Health Public Health Service U.S. Department of Health and Human Services Bethe~da, Md. 20205 Mr. Burton Brin Department o[ Biostatistics and Epidemiology School of Health Loma Linda University Loma Linda, CaliL 92350 Genrose Copley, M.D. Special Programs Branch Division o/ Cancer Cause and Prevention National Cancer lmtitute National Institutes of Health Public Health Service U.S. Department o[ Health and Human Servic~ Bethesda, Md. 20205 Ms. Judith g. Dunn Cancer Canter Northwestern Uni~r~it¥ Medical School Chicago, IlL 60611 Sherman FAias, M.D. Department of Obstetrics and Gynecology Northwestern Univenity Medical School Chicago, Ill. 60611 James E. Enstrom, Ph.D. School of Public Health University of California and Jonsson Comprehensive Cancer Cemer Los Angeles. Calif. 90024 Joseph F. Fraumeni, Jr., M.D. Environmental EpidemloIogy Branch Division af Cancer Cause m'xd Prevention National Cancer Institute National Institutes of Health Public Health Service U.S. Department of Health and Human Se~rices Bethesda. Md. 2O'2O5 Mr. Lorln Gaa~lner Mountain States Health Corporation Boise, Idaho 85705 Eldon Gardner, Ph.D. Department of Biology Utah State University Logan, Utah 84322 J. w. Galdner, M.D. Division of Epidemiology E~artment of Family and Community Medicine University of Utah College of Medicine Salt Lake City, Utah 84152 Mr. Lawrence Garfinkel Department of Epidemiology and Statistics American Cancer Society, Inc. New York, N.Y." 10017 Frank H. Godley, Ph.D. Division o[ Analysis National Canter for Health Statistics Hyamville, Md. 20782 Mr. Michael Grace Department of Medicine University o/ Alberta Edmonton, Alberta T6G IT], Canada Peter Gtcenw-ald, M.D. Cancer Control Bureau New Yo~k State Department of Health Albany, N.Y. 12237 Vincent F. Guinee, M.D. Department of Epidemiology The University of Texas System Cancer Center M. D. Andewon Hospital and Tumor Institute Houston, Tex. 770~0 Mr. William Haenszel Illinois Cancer Council Chicago, III. 60f~)7 1193 JNCI, VOL. ~5. NO. 3. NOVEMBER I~0 TI09371773

bits. Fr~,ces 'B. Lsdk~ ~: Biometr~' Branch." .-.,~.~ -- Dividon d ~m~ ~u~ and Preyer Nadcnal ~ncer Imfitute National Imtitutcs o[ Hmlth Public ~!~-S~ " ~ U.S. D~t~eut ~ H~ and Hun and Epid- Anne P. Lanier, M.D. Alaska Investigations Division Bureau of EpidemioIogy Center for Disease Conuol Anchorage. Alaska 99501 J. Murray, Ph.D. Biometry Branch . Division of Cancer Cause and Pre~( National Cancer Institute Nadonal Institutes of Health Public Health Service U.S. Department of Health and ~ Bethesda, McL ~0'205 .INCl. VOL f~5, NO. 5, NOVEMBER Tl09371774

.. ; / ,(" TI09371775